On Machine Translation

To illustrate this point, ALSC had a webpage on bone tumors intended for a general audience translated by Babel Fish of Alta Vista. Feel free to either reproduce this yourself by inserting the bone tumor URL into the appropriate Babel Fish dialog box, or call up the results directly by clicking on Bone Tumors - Babel Fish translation into German. If you do not understand German, you can get an appreciation for the usefulness - or rather the lack thereof - of the Babel Fish translation program, by having the German translation backtranslated into English. Again, you may either go through the motions yourself or call up the results by clicking on Bone Tumors - Babel Fish back translation.

Except for very short sentences, the MT results are useless, and quite frankly it would take longer to fix (post-edit) the MT translation than to translate from scratch. Please note that in the bone cyst paragraph the sentence "It has a characteristic appearance on X-ray, and does not require any treatment if small" has been converted to express the opposite meaning in the back translation, namely: "It does not have a characteristic appearance on X-ray, and needs handling, if small." So, do you really want to trust a machine with your web site translation? Except for very repetitive originals, restricted subject fields (such as weather reports) and controlled source languages, MT translation may not even be helpful to a human translator. Babel Fish itself admits that it cannot deliver finished products: "The Babel Fish can help you learn a foreign language. But do not rely on it to do all your homework! ;-)."

Similar poor results as with the Babel Fish program are obtained with the On-line MT program SYSTRAN (for a German translation of the Bone Tumor page with this program click on Bone Tumors - Systran translation into German), which is not surprising since the Babel Fish program is derived from the SYSTRAN program. Feel free to check out more machine translation programs at TranslateNow or the Personal Translator Online. For comparison with a human translation, click on Bone Tumors - ALSC translation into German.

Will MT programs get much better in the near future? Not very likely. "Since nobody really knows how the brain works, nobody knows how to model a computer after the human brain either" (New York Times 4/20/00: Untangling the Web's Languages). But don't take ALSC's word for it. Here is what experts and visionaries have to say (you can access any of the corresponding full articles by clicking on the respective author's name):

Timothy R. Hunt (CEO of TermSeek Inc., a software development corporation) in The ATA Chronicle 31, 49-52 (2002):
"... Just think about it. Accountants deal mostly with numbers and reports, and computers handle numbers very well. I think it would be a lot easier to replace accountants than translators.
I am not trying to promote the idea that we should replace accountants with computer software. I am trying to point out that the idea of software programs replacing translators is illogical. Programmers don't try to replace accountants with computers because accountants are treated like professionals who add value. Programmers understand that there is an element of human decision needed in accounting, so I don't know why they miss this fact when they think about translators...
Someday computers may be programmed to provide a higher level of artificial intelligence. When they develop computers smart enough to write prize-winning novels that everyone would like to read, then maybe the computer will have enough skill to do the work of a human translator. After all, the translator is nothing more than a skilled writer who takes ideas from one language and creates them anew in another."

Alexandra Russell-Bitting (Fall 2001): "In a PBS special called 'Planet Work: Finding Solutions in the World Wide Work World" aired in September, host Will Durst explored how globalization of the world economy is transforming the way we work, including how we translate...
...when Kevin put Babelfish to the acid test of Durst's own writing, Durst saw the light. Translated from English into Italian and back to English, the phrase 'Although our American culture may be the red, white, and blue bowling ball on the ping pong table of commerce...' became 'Even if our cultivation American can be the sphere white woman and red blue bowling on the table of pong of metallic noise of the commerce...'
Clearly, Durst got the message that current machine translation programs are okay for simple sentences but not okay for complex phrasing."

Katie Hafner (July 2001) about Dr. Vinge, a visionary computer scientist: "And computers are at the center of Dr. Vinge's vision of the challenges that the coming decades will bring. A linchpin of his thinking is what he calls the technological singularity, a point at which the intelligence of machines takes a huge leap, and they come to possess capabilities that exceed those of humans. As a result, ultra-intelligent machines become capable of upgrading themselves, humans cease to be the primary players, and the future becomes unknowable...
Dr. Vinge readily concedes that his worldview has been shaped by science fiction, which he has been reading and writing since childhood. His dream, he said, was to be a scientist, and 'the science fiction was just part of the dreaming'...
Computers and artificial intelligence are, of course, at the center of much science fiction, including the current Steven Spielberg film, "A.I." In the Spielberg vision, a robotic boy achieves a different sort of singularity: parity with humans not just in intelligence but in emotion, too. 'To me, the big leap of faith is to make that little boy,' Dr. Vinge said. 'We don't have evidence of progress toward that. If it ever happens, there will be a runaway effect, and getting to something a whole lot better than human will happen really fast.'"

Steve Silverman (May 2000): "This dream of accurate, automatic, real-time translation by computers ... has been a consuming obsession for some of the brightest minds in computing, linguistics, and AI research for more than five decades. It has marshaled heroic R&D efforts on academic and commercial fronts from IBM to MIT, burning through billions of dollars in pursuit of what is either the supreme embodiment of a borderless global society or the ultimate vaporware... And it's still largely a dream...Why does the future of MT never seem to arrive?"

Susan Larsson (Jan 2000): "Do human translators worry about their jobs with all of these free online translation services? Not really; if anything, MT services highlight just how much they are still needed. When the message is important, no machine will replace the human ability to pick up the nuance, understand the source culture, and find the perfect turn of phrase to communicate the essence to the target language reader. None of these online services claim to provide perfect results; "post-editing" is needed and recommended, and many also offer the services of human translators from the same Web site."

John Hutchins (June 1999): "From this survey it should be apparent that the application of computers to the task of translating natural languages has not been and is unlikely to be a threat to the livelihood of professional translators. Those skills which the human translator can contribute will continue always to be in demand... As far as the dissemination function (production of publishable translations) is concerned, human translation is more satisfactory and less costly overall whenever it is a question of translating one particular text in a unique subject domain (whether scientific, technical, medical, legal or literary). Machine translation demands the costly investment of dictionary maintenance and updating and the costly involvement of post-editing. This can be justifiable (i.e. cost-effective) only when large volumes of documentation within a particular domain are being translated."

Fred Klein (December 1997): "MT requires an online bilingual glossary, a transfer mechanism, and human post-editing as a minimum. If you happen to need 30-year-old terms from Russian to English (updated in part), you are lucky because a large glossary already exists (Systran). In other languages, vendors offer general glossaries of from 5,000 to 100,000 words, but it is unlikely that your particular technical vocabulary exists online in the language you need.
So you run a count of words not found. A document with a corpus of 30,000 (the total amount of words) may include 14,000 words that are not found. Every word not found, including proper names, misspelled words, unknown acronyms, even verbs and pronouns, has to be researched (in dictionaries), input, coded and tested. The amount of time and money needed is prohibitive. Remember, the machine has rules, but does not know any words other than those in the original online glossary!
... In testing MT systems, beware of "prototypes" and edited texts. Sit down at the keyboard, try it yourself, and beware of salespeople!"