Printer Friendly

Universal translater still years away.

During the 1950s, computer pioneers thought machine translation (MT) would be easy, a mere engineering effort. Early euphoria has turned to frustration, however.

The demand for faster translation is exploding increasingly in international business relations. The European Community employs more than 2,000 full-time translators for nine languages. Yet. many companies quietly have withdrawn their MT projects. That leaves a firm such as Apple Computer to hire local translation companies to produce manuals in 21 languages for its European market alone.

"Machine translation is in a pitiful state," reports Martin Kay, professor of linguistics, Stanford University. Some computational linguists have shown that translations require knowledge about the real world, and the meaning conveyed by a word often is encoded implicitly in the context. Human hearers understand it by including information from previous words into their interpretation of whatever follows.

Consider the sentence, "John got himself into this mess, and he must get himself out of it." There seems little room for confusion, but suppose this were translated from Finnish, which has no gender distinctions in its pronouns. An entirely possible translation is: "John got himself into this mess, and she must get herself out of it."

The feeling that "he" is overwhelmingly more likely comes not from any linguistic properties, but from common-sense knowledge that the person most in need of extraction from a mess is the one who got into it. Kay points out that there is no concept yet to make such knowledge accessible to computers.

Many sentences require additional contextual information to be translated accurately. "He ate the deer" can not be translated into German without providing explicit information that only is implicit in English because German divides the semantic territory occupied by the English word "eat" into two words, One represents the savage mastication of animals, the other specifies human dining. Translation difficulties are aggravated even more for languages that are not related, such as Chinese and French.

Kay maintains that MT designers have spent the last 30 years failing to solve the problem and that a different approach is overdue. "Trying to automate everything is wrong; we should rather develop symbiotic connections of people and machine." For example, computers could help humans who translate technical manuals by keeping track of the vast technical vocabulary. The machines could do most of the translation, but ask a human for clarification of contextual ambiguities.
COPYRIGHT 1993 Society for the Advancement of Education
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 1993 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Publication:USA Today (Magazine)
Date:Dec 1, 1993
Previous Article:Clarifying the meaning of fatherhood.
Next Article:Drilling into the San Andreas Fault.

Related Articles
The Mag Movies: Something too familiar about Polly; ALONG CAME POLLY (12A).
The Razz: Keeping the love alive.
Lovers split by the horrors of warfare.

Terms of use | Copyright © 2017 Farlex, Inc. | Feedback | For webmasters