Linguistic support of CAPT-systems: Prediction of pronunciation errors and creation of exercises

Abstract

This article discusses the possibilities of using linguistic information in CAPTsystems. Information about the user's native language allows to partially predict the pronunciation errors that he or she will make in the target language. This data can be used in phonetic simulators to improve error localization at the level of phonemes. In the first part of the article we summarize the results of two experiments in which we conducted an acoustic contrastive analysis of the Russian and German vowel subsystems and predicted three pronunciation error types that Russian speakers are most likely to make in their German speech. The second part of the article discusses the possibilities of correcting typical pronunciation errors by means of special exercises.