The growth of the Web has facilitated a revolution in translation technologies that are primarily based not on language rules but on vast volumes of text translated by humans into different languages. For example, Google's Translate project can instantaneously translate text among 41 languages. Google research director Peter Norvig's vision for Google Translate is to make billions of routine pages more accessible to ordinary people.
Norvig says that running a newspaper article through Translate allows the readers to get the gist of the article, although he adds that "it will be very rare that you think a native speaker did the translation." He foresees the inevitable advent of cell phones that translate conversation, and in November Google unveiled an application to search on any topic by talking into an iPhone.
The U.S. Defense Advanced Research Projects Agency is testing a universal translator that converts spoken English words into native tongues, and vice versa. It is the agency's goal to equip each U.S. soldier with an affordable iPod-size interpreter via its Spoken Language Communication and Translation System for Tactical Use (TRANSTAC) project. TRANSTAC program manager Mari Maeda wants the devices to be networked so that if one soldier spots an error in translation, all the devices will learn.
Carnegie Mellon University's Alex Waibel says that translation by machines — even highly accurate translation — is complicated by the fact that most people speak broken English, while speaker/listener bias can add to the difficulty of understanding the machine's translation.
From The Washington Post
View Full Article
Abstracts Copyright © 2009 Information Inc., Bethesda, Maryland, USA