A computer-based training system, called 'Pronto', will be developed for use in improving the intelligibility of the English of non-native speakers. Among 20,000,000+ non-native speakers of English in the U.S., many have intelligibility problems that seriously reduce their educational and employment opportunities. In the Pronto system automatic speech recognition, employing both speaker-dependent and speaker- independent modes, will provide evaluative feedback in pronunciation training lessons. Pronunciation models from native English speakers will be used in perception training lessons as will video segments showing the production of native speech. Phase I research has shown that the intelligibility of accented speech can be improved using these procedures. Two native-language-specific modules, one for Spanish speakers and one for Chinese speakers, will be developed. Each module will include a Phonological Evaluation Test, a formal method of estimating the effects on intelligibility of various segmental errors, and a training curriculum. Training procedures for teaching English speech rhythm will be developed, following a Phase l finding that the intelligibility of accented English improves when temporal parameters are changed to be more like native English speech. The effectiveness of training with the system will be evaluated with independent intelligibility testing.
This project will develop a computer- based system to improve the intelligibility of non-native speakers of English, (there are 26 million U.S. citizens for whom English is the second language). The total number of ESL (English as a Second Language) programs now operating, and the recent targeting by the American Speech- Language-Hearing Association of accent correction as a significant clinical problem suggests that individuals, employers, and educational institutions represent a large potential market for such a device, possibly 100,000 units. Since the principles used in the intelligibility training system are language-independent, the potential worldwide market includes any person who wants to learn accentless production of any language.
| Rogers, Catherine L; Dalby, Jonathan (2005) Forced-choice analysis of segmental production by Chinese-accented English speakers. J Speech Lang Hear Res 48:306-22 |
