One of the most important technologies for globalization is Automated Translation.

In order to appreciate its potential, let's examine the facts and implications underlying this important technological trend.

First, we need to recognize that real-time, voice-to-voice, conversational translation is by far the most complex and demanding task ever contemplated for computer science. It consists of more than translation from one language to another. It also requires highly-reliable natural-language processing that depends on the proper parsing of sounds into words. Plus, it must interpret the meaning of those words within the correct context, while also factoring in the human emotions behind what is spoken.

To understand what this means, we can think of Automated Translation as a three-step process that would be used to convert one language to another, such as English to Japanese.

• First, the computer must listen to an English speaker and properly convert that speech into written English, including punctuation.


• Second, it must translate the written English into Japanese with enough accuracy to properly communicate what the English speaker meant to say to a Japanese reader familiar with the subject.


• Third, it must transform the written Japanese into spoken Japanese with the proper inflection and tone to communicate the emotions of the English speaker in addition to the factual meaning.

To do all this, and in real time, represents one of the toughest challenges of our century.

Of these three steps, translating from one written language to another is the most important for globalization. Over the past decade, the demand for language-to-language automated translation has soared as globalization has forced companies to do business across borders.

In fact, by 2007, the already huge translation market is expected to reach $13 billion. But while it is mostly done by hand today, there's already much more text than anybody could hope to translate manually, and it's growing exponentially. Trillions of words and catalogues, advertising, Web pages, and even internal company documents have to be tailored to different countries.

Creating effective translation software requires solving many of the same problems as conventional natural-language processing. As Carnegie Mellon University adjunct professor Chip...