Accent training by computer
March 19th, 2007Our company has been working on pronunciation training systems for more than 20 years (see comdistec.com). The general approach has been to use computer recognition technology to evaluate the way single words are pronounced, and then tell the student/user/trainee whether his/her pronunciation was closer to (a) the way a midwestern US speaker would say that word, or (b) closer to some common mispronunciation that often is made by non-native speakers of English. This approach (used in a system we call HearSay) has been shown to work in research studies, but of course there is more to speaking intelligible English than pronouncing single words correctly. People learning English as a Second Language (ESL students) also need to learn the rhythms and stress patterns used in English, which often differ a great deal from those in their native language. For this reason, computer-based systems for teaching word pronunciation (like HearSay) should be used together with instruction by some speech professional, such as a linguist, ESL instructor, or speech pathologist. We at CDT are very interested in hearing about people’s experiences in improving their accents, so that the training systems we develop can be of the greatest possible help.