Sakhr Software, a company that builds automatic language translators, recently unveiled a prototype smart phone application that transforms spoken English phrases into spoken Arabic, and vice versa, in near real time. The technology isn’t quite ready for your next trip to Cairo, but thanks to recent advances in machine-translation techniques, plus the advent of higher-fidelity microphones and increasing processing power in smart phones, this mobile technology could soon allow two people speaking different languages to have basic conversations.
Before the 1990s automatic translation meant programming in an endless list of linguistic rules, a technique that proved too labor-intensive and insufficiently accurate. Today’s leading programs—developed by BBN Technologies, IBM, Sakhr and others as part of a Defense Advanced Research Projects Agency effort to eliminate the military’s need for human translators—rely on machine-learning techniques instead. The software works from a database of parallel texts—for example, War and Peace in two different languages, translated United Nations speeches, and documents pulled off the Web. Algorithms identify short matching phrases across sources, and the software uses them to build statistical models that link English phrases to Arabic ones.
John Makhoul, BBN’s chief scientist, says the current technology is at its best when confined to subject areas with specific phrases and terminology—translating a weather report from English into French, for example, or helping soldiers gather basic biographical information from people in the field. Makhoul envisions the first consumer applications, five years from now, being similarly constrained. A tourism-related translation app on a smart phone could help an American in Florence get directions from a non-English-speaking local, but they won’t chat about Renaissance art. “It is not going to work perfectly,” he says, “but it will do a pretty good job.”
