CHAPTER THREE METHODOLOGY
This chapter covers descriptions of the subjects, materials and data collection procedures for production and perception experiments, data analysis procedures for the production experiment, and scoring of the perception experiment.
A production experiment and a perception task (a listening test) were executed to test the null hypotheses mentioned in the first chapter: 1) Mandarin subjects do not produce the vowels under study differently from the English native speakers, 2) male subjects and female subjects do not perform differently in the experiments, and 3) Mandarin speakers do not contrast the similar/new vowels differently from the English native speakers.
The production experiment was conducted before the perception experiment because if the perception experiment were carried out first, the subjects would know the sounds to be studied; then when producing the vowels, the subjects might produce the vowels unnaturally. On the other hand, while the production experiment is conducted first, the subjects may be exposed to the vowels under study; yet actually when they see the test items on the test sheet of the listening test, they know
immediately the material to be tested are vowels. Nevertheless, the subjects would
not know which vowel is being tested. Thus, it was deemed better to conduct the
production experiment before the perception experiment.
3.1 The production experiment
3.1.1 Subjects
Two groups of subjects participated in this experiment. The first group consisted of fourteen Mandarin speakers (7 males, 7 females) who were sophomores in the English Department of Providence University. They were about 18-20 years old. Mandarin Chinese is their native language and for those who claimed to speak Taiwanese at home (five subjects) Taiwanese was actually used in a limited topic domain and was often code-switched with Mandarin. Most of them started learning English in junior high school with most of their learning having been in school.
Most of them have not been to an English-speaking country; for those few who did go abroad, their stay was no longer than one month and their purposes were for
sight-seeing. For the seven and a half years of English learning, they have had oral training courses in the last one and a half years. The other group was made up of the same number of English native speakers (7 males, 7 females). The latter group was used as a comparison to the former group. These native speakers varied in their ages and background. Five of them were from Texas and were 20-year-old college students. The others’ ages range from 23 to 66 years old and they were from everywhere of the USA.
3.1.2 Materials and data collection procedures
Words containing the English high vowels (heed, hid, who’d, hood) were
presented in the carrier sentence “I will say h_d” (adapted from Hillenbrand et al.,
1995) for the subjects to produce individually.
1A sheet of paper containing the four sentences printed on was shown to the subjects. The phonetic symbols of the vowels under study were provided on the sheet for Mandarin speakers but not on that for English speakers (see Appendix A). The Mandarin subjects are supposed to have learned the phonetic symbols in high schools. It was assumed that the symbols could help the subjects produce the vowel under investigation and avoid the possibility that the Mandarin subjects did not know how to pronounce the word.
Thus, the production by the Mandarin speakers would actually reveal their
pronunciation of the vowels under study. Their productions were recorded on an AIWA HS-JS245 stereo radio cassette recorder.
3.1.3 Data analysis procedures
For each vowel, the length (L) and the first two formats (F1 and F2) were measured from each subject. Vowel length (in seconds) was measured on wide-band spectrograms by a KAY DSP Sonograph, Model 5500. F1 and F2 (in Hertz) of the vowels were captured from the formant history by a KAY CSL system. The data obtained underwent ANOVA test on length, F1, and F2 of each vowel token for Mandarin speakers and English speakers, both male and female (2 language speakers x 4 vowels x 2 sexes). That is, the factor native language contained two levels (Mandarin and English), the factor vowel included four levels (/ i , I , u , U /), and the
1