ITR: Translation Technology for Language Modeling

Principal Investigator
Mari Ostendorf, Jeff Bilmes, Katrin Kirchhoff

Sponsor(s)
National Science Foundation (NSF)

Award Period
09/15/2003 - 08/15/2008

Abstract
This project seeks to advance multilingual language processing technology by investigating new methods for both statistical machine translation and language modeling for speech recognition in an integrated research program. A key component of the program is the use of n-way parallel translations from several languages for robust learning of general structures and for better assessing generality of algorithms. The automatically learned structure and cross-lingual dependencies will then be used to help alleviate the problems of limited training data by facilitating data sharing across word classes, task domains and languages. The resulting models will impact many aspects of human language technology, but in particular will be evaluated in the context of machine translation and speech recognition. In addition, the project will look at implications of computer processing for bilingual education.

Updates or corrections to this page should be sent to gheaton@u.washington.edu.