Identification vs. discrimination training: Learning effects for trained and untrained sounds

This paper presents the results of two high variability phonetic training regimes aimed at improving the perception of five English vowels, namely /æ, ʌ, ɪ, i, ɜː/ by Spanish/Catalan speakers learning English as a second language (L2). 60 learners of English as an L2 took part in a pretest/training/post-test study and were assigned to one of two training methods (forced-choice identification training, AX discrimination training) and a control group. Vowel training was administered with CVC nonsense words. Preand post-tests involved consonant and vowel identification of nonsense words and generalization to real words. Results indicate that both trained groups significantly outperformed the controls on the trained sounds, showing a positive effect of both training regimes. Identification trainees improved to a greater extent than discrimination trainees on the perception of trained segments. However, a tendency towards improvement with the untrained segments was observed for the learners who received discrimination training only.