The Future of MT is Now and Bar.Hillel Was (Almost Entirely) Right

Proceedings of the Fourth Bar-Ilan Symposium on Foundations of Artificial Intelligence

Volume

Proceedings of the Fourth Bar-Ilan Symposium on Foundations of Artificial Intelligence

Contents

A seer, according to one dictionary definition, is a person who can predict events or developments; someone, in other words, who can foretell the future. As far as I know, Yehoshua Bar-Hillel never actually claimed to be a seer. He was first and foremost a scientist (a logician and philosopher of science), not a prophet. But in his many writings on machine translation in the 1950’s and early 60’s, Bar-Hillel did make a number of clear and important predictions about the future of MT; and based upon his assessment of what was possible in the field, he suggested certain priorities for future development. In this paper, I shall summarize some of the more striking of those early predictions and attempt to evaluate them in light of the current state of the art of MT, some thirty-five years later. As the title of my paper suggests, I believe that Bar-Hillel’s predictions have been largely borne out; and yet his suggestions for more reasonable ways of putting the power of computers at the service of translators have gone by and large unheeded. In the final section of this paper; I will describe a system currently under development at the CITI called TransCheck, which is a novel kind of support tool for human translators. The approach to translation automation that underpins TransCheck is wholly consistent with Bar-Hillel’s own call for a modest and judicious use of mechanical aids as an alternative to classical MT, and as such I am quite confident he would have endorsed it. At the CITI, we are convinced it is the way of the future.