Evolution of MT with the web

Since the Cranfield-94 conference, we have come to a better understanding of the nature of MT systems by separately analyzing their linguistic, computational, and operational architectures. Also, thanks to the CxAxQ metatheorem, the systems ’ inherent limits have been clarified, and design choices can now be made in an informed manner according to the translation situations. MT evaluation has also matured: tools based on reference translations are useful for measuring progress; those based on subjective judgments for estimating future usage quality; and task-related objective measures (such as post-editing distance) for measuring operational quality. Moreover, the same technological advances that have led to “Web 2.01 ” have brought several futuristic predictions to fruition. Free Web MT services have democratized assimilation MT beyond belief. Speech translation research has given rise to usable systems for restricted tasks running on PDAs or on mobile phones connected to servers. New man-machine interface techniques have made interactive disambiguation usable in large-coverage multimodal MT. Increases in computing power have made statistical methods workable, and have led to the possibility of building low-linguistic-quality but still useful MT systems by machine learning from aligned bilingual corpora (SMT, EBMT). In parallel, progress has been made in developing interlingua-based MT systems, using hybrid methods. Unfortunately, many misconceptions about MT have spread among the public, and even among MT researchers, because of ignorance of the past and present of MT R&D.