Material 1 (announcements on board the plane) once
4.6 Results
The hypotheses tested were that subjects would attain
different comprehension scores by different pre-listening formats.
The comprehension measures were of four types: (1) the
combined scores of post-listening comprehension tests 1 and 2;
(2) the separate scores of two post-listening tests respectively; (3)
the combined scores of the post-listening tests classified by their listening proficiency level; (4) the scores of follow-up tests. A
one-way analysis of variance (ANOVA) was conducted, with the alpha for the F set at .05, to analyze the statistical difference of
the mean scores among the groups. When appropriate, a
subsequent post hoc comparison test (LSD) was applied.
4.6.1 Combined Scores on Post-listening Comprehension Tests Table 2 indicates the mean scores and standard deviations of combined scores in two sorts of post-listening comprehension tests. All of the four experimental groups outscored the control greup in listening comprehension performance.
Table 2. Data of Combined Scores on Post-listening Tests
Control QPrev Visual Voc
SM
Mean
NSD
14.38 39 5.87
18.84 38 5.57
17.78 41 4.79
22.46 39 5e14
16.03 39 5.21
ANOVA indicated the significant difference statistically,
F(4,191)=12.929, p<.OOI (See Appendix 8, Table 1), on the average scores among the groups. An LSD test revealed
significant differences between the control condition and three
pre-listening forms: question preview (Q Prev); visual presentation Nisual) and vocabulary pre-teaching (Voc), although there was no significance between the control and soundmodification(SM). Amongthethreeexperimentalgroups
in which the significaBt effects of pre-listening treatments were
Appendix 8, Table 2 for details).
4.6.2 Respective Scores on Post-listening Comprehension Tests
Let us look at the subjects' listening comprehension
performance separately in two serts ofpost-listening tests. The listening text of comprehension test 1 was a monologue, and that of comprehension test 2 was a dialogue. The data of each test are presented in Table 3, and Figure 5 graphically illustrates the mean scores. Each pre-Iistening activity seems te enhance listening comprehension to the same degree, regardless of the text type.
Table 3. Data from Two Kinds of Post-listening Tests
Control QPrev Visual Voc
SM
Test1 (Monologue)
Mean
NSD
7.56 39 3.67
9.95 38 3.30
9.41 41 3.35
12.13 39 3.58
8.26 39 3.18
Test2
(Dialogue)
Mean
NSD
6.82 39 3.21
8.89 38 3.24
8.37 41 2.53
10.33 39 3.01
7.77 39 2.77
14 12 10 8 6 4 2 o
Cont QPrev Visual Voc
Figure 5. Mean Scores on Post-listening Tests
+Test 1
-ae- Test 2
SM
In order to determine whether or not the efficacy of pre-listening activities vary according to the text type (i.e.,
monologue and dialogue), the same procedures were applied:
ANOVA and post hoc LSD test. In comprehension test 1
(monologue), a significant difference was found in subjects'
comprehension performance, F (4,191) = 10.346, p<.OOI (see
Appendix 8, Table 1). Further investigation, using LSD testing,revealed significant pre-listening treatment effects between the control condition and three pre-listening activities: Q Prev,
Visual and Voc. The difference between Cont and SM, however, was not significant enough to be verified statistically. Among
the three pre-listening activities whose efficacy was confirmed in listening comprehension, Voc significantly outscores Q Prev and Visual (see Appendix 8, Table 3 for details).
Identical results of analyses were obtained in
comprehension test 2 (dialogue): significant differences exist among the five groups, F(4,191) = 7.653, p<.OOI (see Appendix 8, Table 1); Q Prev, Visual and Voc significantly outperform the control, although SM does not indicate statistical significance;
Voc outscores Q Prev and Visual in statistical significance (for details see Appendix 8, Table 4).
4.6.3 Combined Scores by Differences in Listening Proficiency Table 4 shows the mean scores and standard deviations on the combined scores of post-listening comprehension tests 1 and 2, according to the subjects' listening proficiency level. Figure 6
proficiency levels, the vocabulary group obtained the highest score followed by the question preview or visual group. The sound raodification group had the lowest score among the four experimental groups.
Table 4. Data of Combined Scores by Listening Proficiency on Post-listening Tests
Proficiency
Level Control QPrev Visual Voc
SM
Mean 17.70 23.18 22.ll 26.23 19.00
Upper
N
10 11 9 13 12SD 5.31 4.07 5.30 3.22 5.06
Mean 14.69 2o.eo l7.53 21.73 16.08
Middie
N
16 13 19 11 13SD 5.67 3.42 2.48 5.95 3.77
Mean 11.46 14.36 15.15 19.73 13.43
Lower
N
13 14 13 15 14SD 5.41 5.09 5.18 3.99 5.39
30 25
20
+ Cont
-)e Voc o
Upper Middle Lower
Figure 6. Mean Scores by Listening Proficiency on Post-listening Tests
ANOVA showed significant differences among the five
groups at all three levels: Upper level, F(4,50) = 6.414, p<.OOI;
Middle level, F(4,67) = 5.691, p<.Ol; Lower level, F(4,64) = 5.358,
p<.Ol (see Appendix 8, Table 5). Subsequent LSD testing
revealed the following results.
(1) Upper Level
Three types of pre-listening activities (Voc, Q Prev and Visual)
indicate a significant pre-listening treatment effect in
comparison with the control condition, while SM does not reach
significance. Among the three effective activities, Voc
significantly outscores Visual (see Appendix 8, Table 6 for details).(2) Middle Level
In comparison with the control condition, a significant
difference was found in two types (Voc and Q Prev), although the other two (Visual and SM) did not indicate significance. Itwas also revealed that Voc and Q Prev were in the same rank order, indicating no significant difference between them (for details see Appendix 8, Table 7).
(3) Lower Level
It was Voc only foT which a significant treatment effect was verified compared with the control condition. The other three pre-Iistening activities (Q Prev, Visual and SM) did not reach statistical significance (for details see Appendix 8, Table 8).
4.6.4 ScoresonFollow-upComprehensionTests
In order to examine whether or not pre-listening activities contribute to learners' progress in listening proficiency, follow-up tests, identical to the post-listening comprehension tests, were
and standard deviations of the follow-up tests, with graphical
illustration of mean scores in Figure 7. Although the
experimental groups barely outscored the control condition, the difference in the mean scores is extremely trivial at this time, which is u' nlike the case with the post-listening tests.
Table 5. Data from Follow-up Tests
Control Prev Visual Voe
SM
Testl (Monologue)
Mean
NSD
8.23 39 3.81
8.50 38 3.41
9.00 41 3.04
9.67 39 2.90
8.74 39 3.67
Test2
(Dialogue)
Mean
NSD
7.31 39 3.02
7.87 38 2.80
8.39 41 3.38
8.46 39 2.59
7.82 39 2.66
Combined Scoreof
Tests1&2
Mean
NSD
15.54 39 6.05
l6.37 38 5.50
17.39 41 5.15
18.13 39 4.53
l6.56 39 5.42
14 12 IO 8 6 4 2 o
+Test 1
ma Test 2
Cont QPrev Visual Voc SM
Figure 7. Mean Scores on Follow-up Tests
ANOVA was performed to determine whether the
differences among the five groups were significant or not. No significance was revealed in any of the three cases: Test 1
(monolegue), F (4,191)=1.026, n.s.; Test 2 (dialogue), F
(4,191)=1.033, n.s.; Combined Scere of Tests 1 and 2, F
(4,191)=1.349, n.s. (see Appendix 9, Table 1). Additional analysis classified by listening proficiency also exposes no significant differences among the five groups at all threeproficiency levels: Upper level, F (4,50) ex e.693, n.s.; Middle level, F (4,67) = O.705, n.s.; Lewer level, F (4,64) = 1.304, n.s. (for details see Appendix 9, Tables 2 and 3).
' Table 6 compares the cembined scores in the follow-up tests with those of the post-listening tests. In Figure 8, the mean scores of both tests are displayed graphically, and they reveal notable decreases in the cases of Voc and Q Prev.
Table 6. Comparison of Post-listening and Follow-up Tests
Control Prev Visual Voc
SM
Post-listening
aests1&2
combined
Mean
NSD
14.38 39 5.87
l8.84 38 5.57
17.78 41 4.79
22.46 39 5.14
16.03 39 5.21 Follow-up
(']]estsl&2 combined)
Mean
NSD
15.54 39 6.05
16.37 38 5.50
17.39 41 5.15
l8.13 39 4.53
16.56 39 5.42
25
20 15 10
5 o
+Post--listening - FollowMup
Cont QPrev Vlsual Voc SM
Figure 8. Comparison of Mean Scores on Post-listening and Follow-up Tests
Discussion
Based on the findings from the experimentation as
described in the previous chapter, the following discussion attempts to explain the results obtained. Implications for the classroom are also presented in order to arrive at appropriate pre-listening activities for Japanese senior high school students.5.1 Facilitative Effects ofPre-listeningActivities on Listening
Comprehension
With respect to the first hypothesis, the data partially supported that listening comprehension is facilitated by the pre-listening activities regardless of the type of activity. The
treatment effects of three types of pre-listening activities (Voc, Q
Prev and Visual) were verified, while SM, contrary to our
expectation, did not contribute to the extent that significantdifferences could be observed. The reason why SM did not reach significance is considered to be that the experimental materials, which were extracted from textbooks for Japanese senior high school students, do not include phonological modifications as numerously as are found in spontaneous English speech. These results suggest that familiarizing students with phonological modifications, the importance of which is not ignored, should not be so efficient at the pre-listening stage in the case of textbook materials, and that such an activity should be desirable as
feedback after listening. The facilitative effect of vocabulary pre-teaching, which was not significant in Berne (1995), was ascertained. The difference is probably due to the defects in her experimental design as described in 3.1.
It is meaningful that the treatment effects of three types of
pre-listening activities (vocabulary pre-teaching, visual
presentation, question preview) were substantiated. Taglieber et al. (1988) recommends three kinds of pre-reading activities (pictorial context, vocabulary preteaching and prequestioning) as effective and practical techniques for reading comprehension, which is considered applicable to listening comprehension. In lesson planning, we can select one of the three pre-listening techniques, depending on student needs and the characteristics of the text, or we can combine all three in the same class.Familiarizing the students with sound modification prior to
listening, which received no confirmation in the present
experimentation, might be an effective method if it is presented with the meaning of the chunk constituent that is about to be heard.
5.2 VocabularyPre-teachingvs.SchemaActivation
In terms of the second hypothesis, the data supported that vocabulary pre-teaching is more facilitative for Japanese senior
high school students than schema activation by visual presentation. Vocabulary pre-teaching Noc) produced
significantly greater degrees of comprehension enhancement
both in monologue and dialogue materials and in all the levels of subjects' listening proficiency. This finding is consistent with Shizuka (1994) concerning the effects of pre-reading activities, contrary to the commonly held generalization that vocabulary pre-teaching is not as effective as inducing relevant schemata.
It is true that the comprehension advantage derives from
the experimental design, in which sufficient numbers of unknown words and expressions were listed together, and in which the sounds and meanings of the vocabulary were adequately connected. If the treatment were reduced, the
outcome would be different, as the precedent research indicated and which was discussed in Chapter 3. The type of vocabulary instruction in this experiment, however, is frequently observed and not special in the least to our situation.
Two other reasons of great importance should not be
overlooked, though. First, the vocabulary level ofthe textbooks for Japanese senior high school students is considerably high.
Hatori (1979) and Takanashi (1995) argue that the ability to infer unknown words, in the case of reading comprehension, is almost impossible when they exceed 50/o of total word count. In listening comprehension, which requires on-line information processing, inferring unknown words is still more burdensome
than in reading comprehension. The percentage of unknown
words, which the dictionary prescribes as those not yet learned in junior high school, was 10.80/o of all the words in Material 1
and 7.60/o of all the words in Material 2. That is to say, the
materials used in the experimentation include so many new
words for the students that they cannot infer the meanings of the words from context, and consequently they fail to understand a passage or a conversation.
A second reason is concerned with the contents of the
materials. In the precedent studies stressing the importance ofbackground knowledge, arduous materials were intentionally
employed that were ambiguous or hardly interpretable withoutcultural or religious background (Anderson et al. 1977,
Steffensen 1981, Markham and Latham 1987, Long 1990). In
the textbooks for high school students, on the other hand, such peculiar materials are rarely found.
Many reading researchers argue that L2 reading
comprehension is influenced by language problems rather than reading problems, especially when students are not yet reading at the prescribed proficiency (Clarke 1980, Alderson 1984,
Yamashita 1993). This assertion is considered valid for listening comprehension as well. It should be deduced that
familiarizing the students with unknown vocabulary prior to listening can be a highly efficient technique for increasingIistening comprehension among Japanese high school students, especially when textbook materials are used.
5.3 TextIlypeandListeningProficiency
The third hypothesis, which proposed that effective types of
pre-listening activities are common between monologue and
obtained in both cases. In both cases, the efficacy of three kinds of pre-listening activities (Voc, Q Prev and Visual) was verified, and the results show that vocabulary pre-teaching Noc) is the
most facilitative for comprehension. Each type of activity
results in the same degree of comprehension enhancement,
whether the material used is a monologue or a dialogue.
The reason for this is thought to be that the identical non-reciprocal task, in which students listen to oral texts in order to obtain information, was imposed on the subjects in both cases,
regardless of the number of speakers (i.e., monologue or
dialogue). In classroom application, teachers are encouraged to use the three types of activities as previously stated, whether the material is a monologue or a dialogue.
With regard to the fourth hypothesis, the results of our experiment supported that the effectiveness of each type of
pre-listening activity differed according to the L2 pre-listening
proficiency of the subjects. At the upper proficiency level, theefficacy of three types (Voc, Q Prev and Visual) was
demonstrated, at the middle level two types (Voc and Q Prev) and at the lower level only one (Voc). These findings indicate that vocabulary pre-teaching is an efficient technique for facilitating comprehension in spite of learners' listening proficiency, and that visual presentation and question preview become less effective aids as the listening proficiency lowers. The reason why Visual was effective only for the upper-level students is thought to be that the learners with middle and lower proficiency failed to
induce relevant schemata, or the schemata they activated were inappropriate. The lack of treatment effect of Q Prev among lower-level learners is thought to result from those students' insufficient knowledge of the language, especially in vocabulary.
Limitations to their vocabulary impeded their comprehension.
The learners with high listening comprehension proficiency seem to make the best use of all types of pre-Iistening activities for their comprehension, while low-proficiency learners seem to avail themselves only of direct and specific information. The present study agrees with Takefuta (1997) on this point, arguing that the information for top-down processing is ambiguous in many cases, so that learners, especially beginners, frequently fail