The Effectiveness of a Computer Assisted Pronunciation Training System for Young Foreign Language Learners

In line with the European Union`s policy to foster multilingualism in the member countries (Barcelona, 2002), Italy has recently made it compulsory to learn a foreign language from the first year of primary education onwards. Currently, 75% of Italian pupils are learning English as a foreign language (Eurydice, 2005). Within language learning, pronunciation obviously plays an important role because communication cannot take place below a certain level of pronunciation quality, even if grammar and vocabulary have been mastered. For children it is particularly important to start off with correct pronunciation examples to avoid that incorrect forms are acquired and possibly fossilized later on. This need is acute for Italian learners of English, because of the differences in the phonetic/phonological systems of these languages, and because in English the relationship between graphemes and phonemes is not as straightforward as it is in Italian. To help children achieve a correct pronunciation, teachers should offer individualized feedback, but the amount of feedback to the individual pupil is generally small due to time constraints. Some experts believe that a CALL (Computer Assisted Language Learning) system based on ASR (Automatic Speech Recognition) technology and on sound pedagogical guidelines could integrate classroom teaching by providing individualized practice and feedback (Eskenazi, 1999). The question that is often left unanswered is whether this type of systems can indeed offer valuable help towards the improvement of pronunciation skills, especially for children. To test this hypothesis, a group of researchers at ITC-irst developed PARLING, a CALL system for Italian children learning English, which also includes ASR technology. The design of Parling was based on work with teachers and children and on a study of literature (Mich et al., 2005). The system offers English children stories and games in which the pronunciation of new words can be trained by means of ASR technology. The study presented here aims to investigate whether this system can supplement traditional language teaching by helping young learners acquire the pronunciation of English words at least equally well as if the training were provided by a teacher. To this aim, we compared a group of pupils receiving teacher-led instruction with a group receiving ASR-based CALL (PARLING). A pretest-posttest design was used to measure the effects of 4 weeks of training. Pre- and post-test consisted of 28 words, selected from those present in the training material so as to cover most English phonemes. Each word was read aloud and recorded by each subject. The recordings were subsequently evaluated by 3 experts. We performed two different analyses: an analysis on global pronunciation skills for all words, and an analysis of the quality of (previously) unknown and difficult words, for which the support offered by the system may be more apparent. Results show that 1) overall pronunciation quality improved significantly for both groups of pupils, 2) both groups also significantly improved in pronunciation quality with respect to difficult and unknown words, which indicates that the system was equally effective as the teacher in improving the pupils` pronunciation.