Experiment 2 .1 Participants

The participants were divided into two groups, A and B. Group A was a treatment group consisting of 21 students, aged 18 to 20, while Group B was assigned as a contrast group with 18 students whose age ranged from 18 to 19. They were all Japanese university students who were non-English majors and their anonymity was secured. The teaching materials were provided with the two groups of learners using two different presentation modes on the Internet. Learners in Group A were assigned the presentation mode of “animated letters with audio,” while the students in Group B learned pronunciation using the “still letters with audio” mode. The learners in both groups logged onto the designated Internet site where each presentation mode was available.

Table 4-1.

Descriptive Data of the Two Groups

Group A (n= 21) Group B (n=18) Presentation Mode Animated Letters with audio Still Letters with audio Listening Test M =12.57, SD = 3.60 M = 11.78, SD = 5.06 Independent t-test

(two-tailed)

df = 37, p = 0.5820, t-value = 0.5554 n.s.

Note. The full mark of the listening test is 20. N=39

Prior to the experiment, a listening test, comprising 10 questions selected from the second, the pre-second, and the third grade of the STEP test, was conducted on both groups to examine their English proficiency levels. A t-test was administered to confirm the homogeneity of the learners’

English ability across the two groups. The results of the test revealed that there was no significant difference in English proficiency between the two groups. A listening test was used here since it is known that learners’ listening ability is highly correlated with the other three linguistic skills, i.e.,

reading, writing, and speaking (DeMauro, 1992; Thompson, 1996). Please refer to Table 4-1 for the details of the descriptive data of the learners in the two groups.

4.4.2 Procedure

The experiment was carried out once a week for five weeks in a CALL classroom from April to May as a part of the regular class activities for both Groups A and B. The pronunciation practice was conducted on the Web for 15 minutes at the beginning of each lesson. The learners in both groups accessed the designated website and practiced using the listen-and-repeat method. They clicked a speaker button on the PC screen to listen to the model speech which was recorded by a Canadian male English instructor at a moderate speech rate and repeated in a CALL room. No explanation on pronunciation was given to the learner while they practiced.

The student’s learning history was recorded via a CGI (Common Gateway Interface) program which was designed to accept and return the data on the Internet that conformed to its specifications.

This system enabled instructors to trace back the record of the students’ learning process. Instructors, for instance, were able to keep a record of the number of times learners repeated, or more precisely, clicked each sentence.

The eight English sentences used for the practice are listed in Appendix C. Words in these practice sentences were chosen from the corpus of the JACET list of 8000 (JACET 8000) basic words, one of the most widely used vocabulary frequency lists in Japan. The words used in the practice sentences were picked up from Level 1 (82.9%). Words belonging to Level 1 represent the most frequently used vocabulary in daily lives and Level 2 (10.5%) represents the second most frequently used words. The vocabulary in Levels 1 and 2 account for 93.4% of the corpus of daily use vocabulary.

The speech rate of the model speech by the NS in this experiment was controlled to be from 4.2 to 5.5 words per second, taking into consideration the number of words and syllables to meet the standard of the “Perceptual Sense Unit (PSU)” suggested by Kohno (2001). Kohno maintains that

holistic information processing will be made possible when units of this length are presented to learners.

Experiment 2 proceeded as follows: Among the eight sentences which were used for pronunciation practice, one test sentence, i.e., “I should have sent her a message.” was chosen for detailed analysis. Before starting the experiment in April, all participants were given five minutes so that they could practice reading it aloud. The readings by the participants were recorded individually by the author. The readings were saved in WAV digital form and were stored as “pre-test” files. An application program called SoundEngine, which was designed to edit audio digital data, was used for the recording. No demonstration readings by an NS, or reading instructions as to how to pronounce the sentence were given to the learners. Their speech samples, with a sampling frequency of 44.1kHz, underwent acoustic analysis using a computer speech analyzer, WaveSurfer. Upon completion of a 5-week session of pronunciation practice with the Website, learners in both groups recorded the same test sentence again and their oral readings were saved as “post-test” files to be used for both acoustic and assessment analysis.

The speech files were randomized and the raters (one Canadian, one American and one Japanese English teacher) were then asked to listen to each speech sample and give pronunciation evaluation scores. Both of the two native speakers of English had MA degrees in sociology and had been teaching English in Japan for more than ten years. Both of them were quite familiar with the phonetic features of Japanese. The Japanese rater, who majored in English in university, had worked in England and the U.S.A. for eight years. He had experience of teaching English at a national university in Japan for twelve years. Thus, he was considered to be eligible to rate Japanese learners’

English pronunciation.

The evaluation was made on a 10-level Likert scale based on the raters’ total impression as to the naturalness of English. They were asked to make their ratings on a scale ranging from 10: “It sounds perfectly natural.” to 1: “It sounds least natural.” The inter-rater reliability score among the three raters was Cronback α = 0.851, which is considered to be very high.

4.5 Results and Discussion