The Effects of Two Methods on Training EFL University Students in Taiwan to Identify Three Non-Native Phonemic Contrasts

Abstract

The present study investigated and compared the effects of two methods on training 71 EFL university students in Taiwan to identify three non-native phonemic contrasts. The study also investigated whether the perceptual training effect could be generalized to new tokens, new talkers, and untrained final stop contrasts as well as whether perceptual training facilitated production. Moreover, to understand the participants' language background and language attitudes toward accents, a survey was conducted.

Eighty-eight EFL university students in Taiwan were randomly assigned to one of two experimental groups, identification (ID) (9 males, 20 females) or same/different (SD) (9 males, 20 females), or to a control group (10 males, 20 females). After attrition, the total number of participants who finished the pretest and posttest was 71, including the identification (ID) group ( n = 24; 7 males, 17 females), the same/different (SD) group ( n = 23; 9 males, 14 females), and the control group (n = 24; 8 males, 16 females). The experimental groups received different training in identifying the English /i/-/&igr;/, /ϵ/-/æ/, and the word-final /t/-/d/ contrasts fifty minutes per week for eight weeks in a computer lab.

The results indicated a significant positive training effect for the two training methods. In contrast, the control group achieved no gains at all. Although the ID group outperformed the SD group in identifying all the three contrasts, the mean difference between the two trained groups was not statistically significant. The effect of perceptual training was generalized to certain contrasts in new tokens, new talkers, and other final stop contrasts. The training effect appeared retained for certain contrasts in new tokens, new talkers, and other final stop contrasts three months after the completion of the training. The perceptual training did not facilitate production. The native speakers of Mandarin outperformed native speakers of Mandarin-Taiwanese on all three contrasts at posttest. An overwhelming 97% of the participants would like to have a native-like accent and 75% of the participants agreed that EFL teachers should have a native-like accent. The results of the survey also show that English is regarded as symbolic capital in Taiwan.

With regard to difficulty, the duration manipulated /i/-/&igr;/ contrast posed the most difficulty for the participants to identify. The /ϵ/-/æ/ contrast proved to be the most resistant to differentiate due to a high degree of spectral overlap of the two vowels. The participants' production performance appeared to be related to specific phonetic environments.

On the whole, the results show a consistent trend that the ID training method is superior to the SD training method although the mean difference between the two trained groups was not statistically significant. However, the large effect sizes of both trained groups indicate that either of the training methods is effective for perceptual training for ESL/EFL students.

Taken together, the findings of the present study provide support for the efficacy of high talker and duration variability in perceptual training. The finding that the trained groups did not improve their oral production also indicates the necessity of production training.