2
listening test scores
4.2 Effects of Two Types of Enhanced Pre-Listening Supports on Participants’
Listening Comprehension: Background Knowledge Pre-Teaching versus Vocabulary Pre-Instruction
This section attempts to answer the first three research questions by examining the effects of two types of enhanced pre-listening supports on students’ listening comprehension. Research question one addresses whether enhanced background knowledge pre-teaching facilitates participants’ listening comprehension. Similarly,
research question two aims to see whether enhanced vocabulary pre-instruction helps students’ listening comprehension. Research question three further asks which type of support is more helpful as a whole. Specifically, these three questions are:
RQ1: Does enhanced background knowledge pre-instruction facilitate participants’ listening comprehension?
RQ2: Does enhanced vocabulary pre-teaching facilitate participants’ listening comprehension?
RQ3: On the whole, which type of pre-listening support is more helpful?
Enhanced background knowledge pre-instruction or enhanced vocabulary pre-teaching?
To answer these three questions, one-way between-groups ANCOVA was used to analyze the collected data. One-way ANCOVA is the statistical tool for detecting the differences between groups while statistically controlling for a variable suspected of potentially influencing the dependent variable (Pallant, 2001). It is acknowledged that learners’ proficiency level has an influence on comprehension performance;
therefore, in the present study, this confounding covariate was statistically controlled and students’ listening comprehension scores were adjusted as well. In other words, by using one-way ANCOVA, the influence of students’ achievement scores on the post-listening comprehension tests can be statistically removed.
Before ANCOVA can be used with reason, five underlying assumptions should be checked first (Pallant, 2001).
(A) Assumption 1: Measurement of the covariate—the covariate should be measured before the treatment (Pallant, 2001, p. 239). The present study met this assumption because participants’ achievement scores were collected from their three mid-term English scores prior to the experiment.
(B) Assumption 2: Reliability of the covariate—the covariate should be reliable
by choosing the most reliable measuring tools (Pallant, 2001, p. 239).
Although strictly speaking participants’ proficiency level cannot be determined by their achievement scores, the fact that three of students’ past mid-term scores were included may more or less compensate for this deficit.
So we could say that the present study passed for this assumption.
(C) Assumption 3: Correlations among the covariates—the covariates should not be too correlated to one another (Pallant, 2001, p. 239). The present study did not have this concern because there was only one covariate in the study.
(D) Assumption 4: Linearity—the covariate should have a linear relationship with the dependent variable (Pallant, 2001, p. 239). The current study met this assumption because there were straight lines formed in each group (see Graph 4-6).
Graph 4-6: Linearity among groups
100
(E) Assumption 5: Homogeneity of regression slopes—the relationship between the covariate and dependent variable for each group is the same, namely, there are similar slopes on the regression line for each group (Pallant, 2001, p.
236). This assumption holds in the study because the Sig. value greater
than .05 (F=.11, df=2, p> .05) was derived, which indicated no interaction between treatment and covariate (see Table 4-3).
Table 4-3: Homogeneity of Regression Slopes
Source SS’ df MS’ F Sig.
Group*achievement scores .56 2 .28 .11 .90
After it was confirmed that the current study had not violated the five assumptions, one-way between-groups ANCOVA was conducted to explore the differences among three treatment groups. ANCOVA statistically adjusted means on the dependent variable for each group. Therefore, as Table 4-4 revealed, vocabulary class was adjusted to have higher corrected means because it had the lowest achievement scores whereas background class had lower corrected means because it had the highest starting scores.
Table 4-4: Means and Adjusted Means for Treatment Groups
Group N Achievement scores Mean Adjusted Mean
Background 32 74.16 8.41 8.21
Vocabulary 33 64.95 7.42 7.55
Control 34 66.88 6.71 6.77
Results of ANCOVA showed that there was significant difference among groups (F=6.70, df=2, p<.01, eta squared=.12), after controlling for participants’ achievement scores (see Table 4-5). As indicated by the eta squared value, the treatments that three groups received respectively accounted for 12 percent of the variance in participants’
post-listening comprehension performance. This, according to Cohen (1998), had
nearly reached a large effect size5, suggesting that the treatment itself played an important role in determining students’ performance in listening comprehension tests.
Table 4-5: ANCOVA of Between-Groups Effects
Source SS’ df MS’ F Eta Square
Achievement scores 62.22 1 62.22 24.36** .20
group 34.23 2 17.11 6.70** .12
**P<.01
Also shown in Table 4-5, the effect of the covariate, in this case the participants’
achievement scores, was significantly related to their comprehension performance (F=24.36, df=1, p<.01, eta squared=.20) while controlling for “group” variable. This information showed that students’ proficiency level was highly correlated to their listening comprehension, accounting for 20 percent of the variance. Research question 4 and 5 further addressed the interaction between participants’ proficiency level and the types of pre-listening treatments on listening comprehension. Results would be presented in section 4.3.
Now that a significant difference was detected among groups, exploring where such difference lay by using post hoc analysis was the next step. As shown in Table 4-6, enhanced background knowledge class had significantly better performance than control class on comprehension tests (p<.01). Likewise, when compared to control class, enhanced vocabulary class reached a .05 significant level in their listening comprehension (p<.05). In other words, both of these two treatments were proved to be effective pre-listening supports to junior high school students in this study which helped them listen better in listening tasks.
5 According to Cohen (1988), .01 is a small effect size, .06, medium, and .14, large.
Table 4-6: Background v.s. Control / Vocabulary v.s. Control
Group Simple Contrast Contrast Estimate Sig.
Background v.s. Control 1.45 .000**
Vocabulary v.s. Control .79 .047*
**p<.01, *p<.05
Given that both enhanced background knowledge group and enhanced vocabulary group performed significantly better than control group, this study proceeded to explore which one of the two had even greater facilitative effects.
Post-hoc analysis indicated that there was no significant difference between these two pre-listening supports (p>.05) (see Table 4-7).
Table 4-7: Background v.s. Vocabulary
Group Repeated Contrast Contrast Estimate Sig.
Background v.s. Voc .66 .10
To further strenthen the result, t-test analysis was also used to compare the mean difference between [background-control] (mean=1.72) and [vocabulary-control]
(mean=.76). The result showed that the mean difference between the two categories was not significant (p>.05) (shown in Table 4-8), signifying that background treatment did not outperform vocabulary treatment, and vice versa. The above two findings therefore suggested that enhanced background knowledge pre-teaching and enhanced vocabulary pre-instruction were equally facilitative in enhancing Taiwanese junior high school students’ listening comprehension in the present study.
Table 4-8: T-Test of Mean Difference b/w [background-control] and [voc-control]
t-test for Equality of Means
t df Sig.(2-tailed) Mean dif.
Contrast Equal variance assumed 1.544 63 .128 .96 Contrast Equal variance not assumed 1.545 62.99 .127 .96
In brief, the above results answered the first three research questions. That is, the pre-listening support of enhanced background knowledge pre-teaching indeed facilitated Taiwanese junior high school students’ listening comprehension in the present study and so did the treatment of enhanced vocabulary pre-teaching.
Furthermore, when these two supports were compared, no significance was found, indicating that both of them were equally helpful to students’ listening comprehension.