V 4.6.2 Inferential Statistics Results
6.3 RQ1a: What are the effects of two different forms of language teaching (TBLT and TSLT) on the development of students’ L2 speaking performance over time?
6.3.4 Effects on Fluency
Two measures were employed in this study to operationalise fluency: number of pauses (FNP) and number of repairs (FNR). The quantitative results revealed that both groups, the EIST group and ST group, managed to improve breakdown fluency by reducing the number of pauses from pre- to post-test with small and medium effect sizes, respectively. In contrast, for repair fluency, only the ST group significantly reduced the overall number of repairs in the post-test with a small effect size (see Section 4.6.2 for fluency results). It should be noted that the characteristic of the tasks in the speaking tests (pre- and post-tests) were different from those used in the intervention. The questions in the speaking tests were designed as a monologic task to assess students’ overall speaking performance individually, whereas the speaking tasks throughout the intervention period were dialogic in nature. The conditions in which both tasks were performed also differed and need to be taken into consideration. The speaking tests had a specific time limit for students to perform and planning time was given prior to the tests. In the intervention, however, time was not restricted and ample time was given for group discussion and pair work. Therefore, this allowed for an existing speech gap between the interlocutors to comprehend information, plan their speech systematically and negotiate meaning with their peers using both linguistic and non-linguistic resources effectively.
A thorough analysis was carefully conducted in order to find pausing patterns. The results were that both groups produced more filled pauses than unfilled pauses in both pre- and post-tests.
188 The filled pauses (e.g., uh, um) in both groups were highly used in mid-clauses of the speech and frequently occurred after a repair was made. Meanwhile, unfilled pauses or silent pauses were found to be more prominent in both mid-clauses and at the beginning or end of the clauses. For repair, both groups seem to have approximately the same amount of repetition (repeated words and/or phrases) and replacement (word and/or phrase replacing) from pre- to post-test. However, the major difference was in reformulation (correcting words and/ or phrases) between both groups. The ST group drastically reduced the amount of reformulation from pre- to post-test compared with the EIST group which had a small decrease. Therefore, these findings indicated that speaking tasks without the influence of explicit instruction had more positive effects on students’ fluency. There are possible explanations for these effects.
First, the occurring pauses might be caused by linguistic or cognitive processes. A study by Tavakoli (2011) investigated the differences in pausing patterns between L2 learners and native speakers in four oral narrative tasks (picture-story narration). The results of that previous study suggested that L2 speakers produced more unfilled pauses in the mid-clauses and paused less at the end of the clauses across all four tasks. Similar results were found in the present study where the unfilled pauses occurred in mid-clause. As for repairs, Tavakoli (2011) suggested that L2 speakers paused before they repeated a vocabulary item (repetition), replaced a word or a phrase/expression with another (replacement) and/or prior to correcting (reformulation). In the present study, these pauses were inconsistent, depending on the students in which pauses could occur both before and after repairs. This indicates that students in both studies used pauses when they were thinking to change or reshape the structure of an utterance (repetition and replacement) and when monitoring their own performance or checking whether they were using accurate and appropriate language (reformulation) (Tavakoli, 2011). Furthermore, in the present study, the speaking data indicated that some pauses occurred during online planning. From this it could be assumed that pauses helped students when they were formulating their
189 thoughts, organising ideas and figuring out how to convey the message into words. Another interesting fact about the current findings which is similar to the previous study is that students paused outside the formulaic sequences and rarely in the middle of a formulaic sequence (e.g., last+but+not+least, in+the+meantime, and+so+on). It could be assumed that these chunks of words require less attention and no additional processing is needed (Tavakoli, 2011) and thus they were ‘ready to go’ and were uttered automatically by the students with ease.
Second, the level of proficiency might have affected their fluency outcome. The students in the present study were considered as intermediate second language learners of English and currently in their first year of undergraduate study. The positive effects of the intervention on fluency especially breakdown fluency in this study differ from those in Ellis et al. (2019). The previous study showed that students were having difficulties in producing fluent speech, with more frequent and longer pauses, especially in the group which received explicit instruction. This might be arguable because, in the previous study, English was considered as a foreign language where students do not use the language frequently in their daily life. Moreover, the students in Ellis et al. (2019) were secondary school students and the average age was 14 years. However, in the present study, students were in their tertiary level and English proficiency level was considered intermediate, and thus their experience perhaps diminished the negative effect of increasing demands on fluency.
Third, explicit instruction affects fluency performance. Although pre-task planning was given, the EIST group did not make any significant decrease in reformulation. This suggests that the EIST group’s attention was more on accuracy in which they were keen to correct themselves in order to produce accurate language and meet the target structure. This is in harmony with Ellis et al. (2019) who argued that the explicit instruction group concentrated more in producing the grammatical target structure, which explains why students paused more regularly and for a longer period. However, the previous study did not look into repair fluency. Although it is