The first research question tackles students’ improvement in fluency after
receiving the training from ETA for one semester, and in this section, the scores of the perceived fluency offered by the raters were listed below. For passage reading section, the results of the perceived fluency were listed in Table 3 as follows:
Table 3 Perceived Fluency in The passage-reading Task
Perceived Fluency Pretest Posttest
M SD M SD
Overall 2.537 1.307 2.759 1.059
Based on Table 3, it could be observed that the mean of perceived fluency in the passage-reading task has increased from 2.537 to 2.759, suggesting that their
perceived fluency was enhanced in the posttest. The standard deviation, on the other hand, decreased from 1.307 to 1.059, which implied that the difference between the participants was lessened; that is, the group of participants were considered more homogeneous after receiving training.
For the second task, storytelling, the difference of perceived scores in pretest and those in posttest test was shown below:
36
Table 4 Perceived Fluency in The storytelling Task
Perceived Fluency Pretest Posttest
M SD M SD
Overall 2.388 1.295 2.648 1.223
According to the table 4, it was clear that the participants increased their perceived fluency on the average. The mean score of perceived fluency has increased from 2.388 to 2.648. The standard deviation, nevertheless, did not show an obvious difference in this task. In sum, the results only suggested that their oral fluency was perceived to be better at the end of the semester.
While referring to the overall improvement of fluency as perceived by the raters, it was interesting to reveal that the temporal indicators employed in this study
contributed to the judgement of perceived fluency perceived fluency, as proposed in Lennon’s (1990) and Riggenbach’s (1991) studies. The increase in speech rate has been deemed a significant indicator of improvement of fluency, and the findings of this study showed that most students increased their speech rate in passage reading section. Similarly, reducing the frequency of pauses was an indicator of improving perceived fluency (Lennon, 1999), which was also confirmed in the results of the participants’ frequency of pauses. In sum, it could thus be assumed that rate of speech and the frequency of pauses distinctively influenced the perceived fluency in this study.
To affirm whether the correlation between the indicators and perceived fluency could be established, a correlation test was conducted on all the variables in this study, and the results are presented in the Table 4 below:
37
Table 5 The Correlation of Variables in The passage-reading Task
Rate of Speech Frequency of Pauses
The results of the correlation test showed a clear and strong correlation between the rate of speech and perceived fluency. With the correlation coefficient being 0.779 and 0.774, rate of speech is highly correlated with perceived fluency. As for frequency of pauses, it was slightly negatively related to the perceived fluency.
As for the storytelling task, to understand the variables that contributed to the enhancement of oral fluency, a correlation test was also conducted to confirm the relationships between perceived fluency and rate of speech and frequency of pauses.
The results were presented below:
Table 6 The Correlation of Variables in The storytelling Task
Rate of Speech Frequency of Pauses
The results pointed out that rate of speech is highly correlated with fluency, and frequency of pauses was slightly associated with fluency.
However, aside from the indicators which have already been known to contribute to the perception of fluency, another salient improvement noticed in the tests was accuracy. Since this study focused on ETA and their impacts on speaking fluency, the
38
rating of accuracy was beyond the scope of the study. However, when analyzing the answers of the participants, many representative samples could specify that students increase their accuracy in the pretest. The progress of accuracy was made manifest in an example from the storytelling test below:
One day a girl and a boy (2.0) and (.6) to house (2.1) / they are (.6) going a one house (2.1) /they looking for a thing (.6) / and (.9) their (.4) on they need water (.) / and they (.) get help (1.2) /they get out and find a new swimming pool (1.0) / they are play very happy (4.8) / and everyday they are going to new (1.0) swimming pool.
(Sample 01 pretest) In the example above, the speaker had a hard time completing an error-free sentence, but numerous grammatical mistakes and awkward meaning could be seen.
The speaker, contrarily, uttered much more accurate speech in the posttest as follows:
One day the boy and the girl (.7) are going to play/ (.4) and they (.4) going to their house (.3) / but their no one in house (.5) / they go to the kitchen (1.1) but (.7) the thing is broken / and they’re (.3) all in the yard and meet her meet their mother (.4) /Mother said (.6) / we have a swimming pool (2.6) / you can go into (1.0) swimming pool / and (.4) play with (1.1) swim (.7) water (.1) / I will (1.1) make juice water and some snakes (.) / Have fun.
(Sample 01 posttest) As observed above, the speaker better conveyed the message and produced more accurate speech. Without quantitative empirical evident, it would be too wild a claim that accuracy affects the judgement of accuracy, while previous research claimed that the accuracy of speech may affect the judgement of fluency (Kormos & Denes, 2004).
Kormos and Denes (2004) claimed the following in their study:
39
Our research suggests that accuracy also plays an important role in fluency judgements and sometimes overrides the effect of temporal factors on listeners. The correlations between the temporal and linguistic
variables also reveal that accuracy is positively related to temporal variables that are influential in fluency judgements. (Kormos & Denes, 2004, p. 160)
Based on the above, it could be suggested that speaking accuracy could be significantly positively correlated to perceived fluency. Namely, when giving score of fluency, raters still took accuracy into account unconsciously. For example, in passage reading task, the rating might be controlled by the exact wordings on the test,
suggesting that they might consider the correct wordings as the norm. When students could not complete all the words on the test or make multiple mistakes in their
utterance, the raters might judge the speakers to be nonfluent since their accuracy was inferior.
To summarize, this section focused on the difference of the means of perceived fluency in pretest and posttest. The results indicated that most students increased their perceived fluency in the posttest, with an exception of class 703 in storytelling task.
To further analyze the reasons why students were perceived more fluent, a correlation test pointed out that accuracy strongly influenced raters’ judgement of fluency in passage reading task, while it was not significantly associated with fluency in
storytelling task. Furthermore, the increase in speech rate was highly correlated with perceived fluency. More details of speech rate would be provided in the following section.
40