| This current study investigates and compares the effectiveness of High Variability Phonetic Training (HVPT), Low Variability Phonetic Training (LVPT), production training, and a combination of either HVPT or LVPT plus production training on improving both the perception and production of three English vowel pairs, /I/-/i:/, /e/-/ae/ and / ℧ /-/u:/, by Cantonese-speaking ESL learners. HVPT is a perception training paradigm which provides the subjects with training stimuli under various phonetic environments produced by various speakers, whereas LVPT offers stimuli produced by just a single speaker. With a 3-phased "pretest-treatment-posttest" structure, the current study recruited 85 participants and employed an experimental design in which 9 groups of the target participants took part in different training sessions in over a period of time, with also a control group. Perceptual training intensity (intensive -- 10 training sessions per day; or standard -- 2 training sessions per day) was also one factor under investigation.;Analyses showed that the subjects generally had difficulties in both the perception and production of the three vowel pairs before training. While the production training was found to have only mild benefits to the perception learning, both HVPT and LVPT were effective in improving the subjects' perception of the three vowel pairs. Groups receiving both HVPT/LVPT and the production training improved as well, but their improvement was not as great as the groups receiving only the perceptual training. Perceptual learning could also be generalized to new words and new speakers, but only the HVPT group demonstrated significantly better performance than other groups. Perceptual training intensity did not have significant influence on the subjects' performance.;As for the learning in the production domain, all subjects with training improved significantly after the intervention. The HVPT group outperformed the LVPT and the control groups with significance whereas the LVPT group did not display robust difference with the control group. The production training was also beneficial to the subjects' improvement of the production accuracy. Even different training groups improved significantly in the posttest; however, only the HVPT with production training group outscored the other groups with significance. Again, perceptual training intensity was not an influential factor. The formant frequencies and durations of the vowel pairs produced by the trained groups after the training became not as conflated as before and were closer to native productions. However, although the number of target productions in the contextualization test of production at the sentence-level was high, only the HVPT group with production training outperformed the other groups with significance.;The above findings show that training in perception alone can be sufficient for significant improvement in both the perception and production domains, with the HVPT being a more effective paradigm. Production training can facilitate more learning in production than in perception. Future studies can investigate the acoustic cues that these learners rely on, as well as how different training methods can benefit L2 learners.;Keywords: speech perception and production, High Variability Phonetic Training (HVPT), L2 vowel perception, non-native phonemic contrast, production training, perceptual training intensity, Cantonese ESL learners. |
