CHAPTER FOUR RESULTS & DISCUSSION
In this chapter, four research questions of the present study will be addressed respectively. Before addressing the four research questions, an overview depicting general findings of the present study will be given in section 4.0. In section 4.1, effects of listening comprehension anxiety on listening comprehension will be explored. The relationship between listening comprehension anxiety with personality and linguistic intelligence will be investigated in section 4.2. The interviewed data regarding some other factors affecting listening comprehension anxiety and solutions to it will be discussed in section 4.3, followed by section 4.4, investigating the interaction of all the constructs examined, i.e., listening comprehension anxiety and performance, personality and linguistic intelligence. Finally, a summary of the results and a discussion of this chapter will be given in section 4.5.
4.0 Overview
This overview comprises two parts. First, the descriptive results of the questionnaire assessing the participants’ listening comprehension anxiety (ANX), personality (PER) and linguistic intelligence (LI) levels are presented. Second, a correlation matrix incorporating all the constructs, i.e., listening comprehension performance, listening comprehension anxiety, personality, and linguistic intelligence is reported.
Table 4-1 shows the descriptive results of the questionnaire that investigated the
participants’ listening comprehension anxiety, personality traits and linguistic
intelligence.
Table 4-1: Means and Standard Deviation for the Questionnaire Items of All the Participants
1Item Mean SD Item Mean SD
anx1 3.86 .941 per3 3.53 .912
anx2 3.14 1.053 per4 2.95 .978
anx3 3.26 1.103 per5 3.25 .892
anx4 2.95 1.125 per6 2.74 .908
anx5 3.44 1.018 per7 2.90 1.008
anx6 3.33 1.042 per8 3.15 .940
anx7 2.95 1.112 per9 3.27 .895
anx8 2.97 1.059 per10 3.46 .884
anx9 2.83 1.129 li1 3.12 .849
anx10 2.77 1.078 li2 2.83 1.000
anx11 2.86 1.108 li3 2.76 .975
anx12 3.49 .988 li4 2.94 1.011
anx13 2.53 1.131 li5 2.78 .937
anx14 2.66 1.033 li6 2.78 .887
anx15 4.21 .825 li7 2.88 .872
per1 3.73 .872 li8 2.69 .832
per2 3.49 .881 N=1,400
In Table 4-1, it was found that the participants’ listening comprehension anxiety levels were generally high as shown in all the items assessing the anxiety level (anx 1 to anx 15), ranging from 2.53 to 4.21 (Disagree to Strongly Agree). In addition, most of the items were close or even above the score of 3.00, indicating that the participant had a high degree of listening comprehension anxiety. In particular, Items anx15 and anx1 were found to be the indicators that provoked the participants’ anxiety most easily since they came to the first and second place among the 15 items used to examine the participants’ anxiety levels. This implied that the participants tended to feel anxious about English listening simply because they were not satisfied with their English listening ability (Item anx15) and because they were not sure whether they
1 The mean scores of the anxiety, personality and linguistic intelligence scale all ranged from 1 to 5.
understood what they heard in English (Item anx1).
Similarly, the personality scale showed that the scores of all the items (Items per1 to per10) examining the participants’ personality level were slightly higher, ranging from 2.74 to 3.73 (Disagree to Agree). The results that most items for the personality category were above 3.00 showed that the participants of this study were slightly extroverted. This result was confirmed by the highest scores found in response to Item per1, which asked the participants whether they considered themselves an extroverted person.
The participants’ linguistic intelligence levels assessed via the Items li1 to li8 showed that their linguistic intelligence was slightly low, ranging from 2.69 to 3.12 (Disagree to Agree). Moreover, except for Item li1, all the other items were lower than the score of 3.00, indicating that the participants of this study did not possess a high degree of linguistic intelligence.
Table 4-2 shows the correlation matrix of all the investigated constructs:
Table 4-2: Correlation Matrix of All the Constructs Concerned
LC ANX PER LI
Pearson Correlation 1 -.54(**) -.03 .14(**)
Sig. (2-tailed) .000 .290 .000
LC
N 1,400 1,400 1,400 1,400
Pearson Correlation 1 -.13(**) -.28(**)
Sig. (2-tailed) .000 .000
ANX
N 1,400 1,400 1,400
Pearson Correlation 1 .31(**)
Sig. (2-tailed) .000
PER
N 1,400 1,400
Pearson Correlation 1
Sig. (2-tailed) LI
N 1,400
As shown in Table 4-2, the participants’ LC performance significantly correlated with their ANX and LI scores (ANX: p<.01; LI: p<.01). To be more precise, the participants’ ANX negatively correlated with their LC performance, while their LI positively correlated with the LC. This indicated that the higher the participants’ ANX scores were, the worse their LC performance was; on the other hand, the higher the LI scores, the better the performance. However, the participants’ LC performance was found not to correlate with their PER scores at a significant level. Concerning the participants’ listening comprehension anxiety with the two examined personal factors (i.e., PER and LI), it was found that the participants’ PER and LI both correlated negatively and significantly with their ANX scores. That is to say, the participants with higher personality traits (being more extroverted) or with higher levels of linguistic intelligence had lower levels of listening comprehension anxiety. When investigating the relationship between these two personal factors, a positive correlation was also obtained at a statistical significant level (p<.01), which suggested that the participants who possessed a higher level of personality were those who had a higher level of linguistic intelligence and vice versa.
To sum up, the general findings illustrated here suggested that the participants of this study were the learners with higher levels of listening comprehension anxiety and possessed more extroverted personality traits. They were also the learners who did not have a higher level of linguistic intelligence. Furthermore, it was found that the participants’ listening comprehension performance might be influenced by their listening comprehension anxiety and linguistic intelligence. And the participants’
listening comprehension anxiety, personality traits and linguistic intelligence were in
a certain relationship with one another.
4.1 Effects of Listening Comprehension Anxiety on Listening Comprehension Performance
In response to the first research question probing into the relationship between listening comprehension anxiety and listening comprehension performance, this section will first present the descriptive results of all the participants’ ANX scores and their LC performance. The correlation between these two investigated constructs (i.e., ANX and LC) will be examined. Second, the group effect will be discussed by comparing the two anxiety groups’ listening comprehension and their anxiety levels.
4.1.1 Overall Performance of the Participants
Table 4-3 shows the descriptive results of the LC performance and the ANX scores of the 1,400 participants:
Table 4-3: LC Performance and Anxiety Scores of All the Participants
2N Mean SD
LC 1,400 51.31 18.37
ANX 1,400 3.15 .673
As shown in Table 4-3, regarding the participants’ LC performance, the overall mean score was 51.31 out of 100 indicating that, overall speaking, the participants did modestly on their LC test. As for their ANX scores, consistent with the earlier analysis, the mean score of 3.15 out of 5 showed that the participants were quite anxious about listening to English (the value 3.15 was between the response of “Neither Agree Nor Disagree” and “Agree” in responding to their listening comprehension anxiety levels assessed by the FLLAS used in the present study).
The result of correlation between the LC performance and ANX scores of all the
2 The mean scores of the LC test ranged from 0 to 100, and that of the anxiety scale was from 1 to 5.
participants is reported in Table 4-4 and the correlation tendency is shown in Figure 4-1:
Table 4-4: Correlation between LC Performance and Anxiety Scores of All the Participants
ANX
Pearson Correlation -.54(**)
Sig. (2-tailed) .000
LC
N 1,400
** Correlation is significant at the 0.01 level (2-tailed).
Table 4-4 shows that the correlation coefficient (Pearson’s r) of the LC performance and ANX scores of all the participants was -.54 and they correlated with each other significantly (p< .01). In addition, the negative correlation obtained between the two constructs suggested that when the participants’ ANX scores increased, their LC performance decreased and vice versa. Moreover, a correlation obtained here was slightly high (r= -.54 > .50)
3indicating a close relationship between the participants’ ANX and LC.
4.00 2.00
ANX 100.00
80.00
60.00
40.00
20.00
0.00
LC
Figure 4-1: Correlation between LC Performance and Anxiety Scores of All the Participants
3 According to Cohen (1988), in describing the strength of the association between two variables, a small correlation was defined as having a value of approximately 0.1, a medium correlation as 0.3, and a large correlation as 0.5 or greater.
As shown in Figure 4-1, a negative correlation between the participants’ ANX scores and LC performance was found since the line formed by quite intensive plots went from the top-left to the bottom-right side in the scatter-graph. This further demonstrated the tendency of the participants’ LC performance going down with their ANX scores going up.
The results presented above can be interpreted as follows. First, the descriptive statistics showed that the average ANX score of the FLLAS was 3.15, indicating that the participants in the present study had a higher level of listening comprehension anxiety compared to Cheng’s (2007) study obtaining a mean score of 2.95 and also to Elkhafaifi’s (2005) finding of 2.77 (also out of a five-point Likert scale). It is not surprising that the participants’ anxiety intensity obtained in the present study was higher than the participants in Elkhafaifi (2005) since it is believed that different cultural contexts may come into play in affecting the results. Generally speaking, the oriental learning context is more anxiety-provoking due to the traditional Chinese culture, in which “Chinese teachers and students [were predisposed to] regard education as a serious undertaking that is least likely to be associated with light-heartedness but requires deep commitment and painstaking effort” (Hu, 2002, p.97). Thus, the result might lead to a higher level of the participants’ listening comprehension anxiety in the present study than the anxiety level conducted in a western context with western students (cf. Elkhafaifi, 2005). Cheng’s (2007) participants were accustomed to taking a weekly listening test given by the teacher;
therefore, they might not feel that anxious even though the FLLAS was given to
assess the anxiety level right after the weekly listening test. In contrast, the
participants in the present study faced a totally different situation. The FLLAS was
required to be completed right after they took a formal English placement test
administered by the school’s authorities concerned. This placement test would place
them into different levels of classes based on their performance on the test. Therefore, they were under great pressure in this situation.
Second, a correlation analysis showed that the listening comprehension anxiety
and listening comprehension performance of all the participants correlated with each
other significantly and negatively (r= -.54, p<.01). This indicated a negative
relationship between the participants’ listening comprehension anxiety and their
listening comprehension performance. Such findings corresponded to Cheng (2007),
Cheng (2005), Elkhafaifi (2005), Song (2005), and Vogely (1998), who all suggested
a negative effect of listening comprehension anxiety on learners’ listening
achievements. In addition, the results lent supports to the claims of “debilitating
anxiety” as a more typical language learning anxiety found in the previous studies
investigating general language learning anxiety or other sub-skill language learning
anxiety (Aida, 1994; Cheng, 1998; Horwitz et al., 1986; MacIntyre & Gardener,
1994a, 1994b; Price, 1991; Young, 1986). One possible explanation for this is given
by Tobias’ (1986) model. Tobias (1986) contended that anxiety could interfere with
learning at three stages, i.e., input, processing, and output. At the input stage, anxiety
acts like a filter impeding input from getting into the cognitive processing system. At
the processing stage, anxiety disturbs the coming information especially when
learners tackle with a difficult task since it requires more processing space in
processing a more difficult task. Finally, at the output stage, anxiety serves as a
disruption and it hinders the retrieval of the existent information. Consequently, it is
not surprising to find that listening comprehension anxiety had a negative impact on
listening comprehension performance.
4.1.2 General Performance of the Two Anxiety Groups
To investigate the group effect, among all the participants, the High Anxiety (HI-ANX) Group and the Low anxiety (LOW-ANX) Group were selected from the top 25% and bottom 25% participants based on their listening comprehension anxiety level in the FLLAS. Thus, altogether 700 participants were selected from the overall subject pool with 350 participants assigned to each ANX group. In other words, the HI-ANX group consisted of 350 participants and so did the LOW-ANX group. Table 4-5 reports the descriptive results of the ANX score of the HI-ANX and LOW-ANX groups together with the LC performance of these two groups:
Table 4-5: LC Performance and Anxiety Scores of the Two Anxiety Groups Group Mean SD
ANX 4.00 .33
HI-ANX
(n=350) LC 39.41 14.29
ANX 2.29 .34
LOW-ANX
(n=350) LC 65.33 15.99
As reported in Table 4-5, the mean ANX score of the HI-ANX group was 4.00,
while that of the LOW-ANX group was 2.29 only. The t-test comparison also yielded
a significant difference between the ANX mean scores of the two anxiety groups
(t=-66.939, p< .001). In a word, these results indicated that the HI-ANX participants
did experience a higher level of the anxiety while engaging in English listening
related activities or events than their counterpart, LOW-ANX participants who were
found to be not that anxious for English listening. These results can be further
supported by the interviewed data. Among those 20 focused group interviewees (10
from the two anxiety groups respectively), nine out of ten participants from the
HI-ANX group reported that they did experience listening comprehension anxiety
when engaging in English listening activities. Contrary to the HI-ANX group, all the
ten interviewees of the LOW-ANX group reported that they did not feel anxious at all or even they felt so when listening to English, it was just to a slight degree. In terms of the LC score, the participants’ mean score of LC performance of the HI-ANX group was 39.41 out of 100, and that of the LOW-ANX group was 65.33 higher than the former group. Moreover, the LC mean scores of the two groups were found to be significantly different according to the t-test comparison (t=22.606, p< .001), indicating that the LOW-ANX group did significantly better than the HI-ANX group on their LC performance.
Correlation analyses were also done to compare the LC performance and ANX score of the two anxiety groups. Table 4-6 presents the results:
Table 4-6: Correlation between LC Performance and Anxiety Scores of the Two Anxiety Groups
Group ANX
Pearson Correlation -.28(**) LOW-ANX
(n=350)
LC
Sig. (2-tailed) .000
Pearson Correlation -.06 HI-ANX
(n=350)
LC
Sig. (2-tailed) .279
** Correlation is significant at the 0.01 level (2-tailed).
Table 4-6 reveals that, for the LOW-ANX group, although the r value was quite
low with only -.28, a significant level (p< .01) was reached in this case, implying that
a small but significantly negative relationship was found between the LC performance
and ANX score of the LOW-ANX group. That is to say, these two constructs were in a
certain relationship. As for the HI-ANX group, while a negative correlation was found
between their LC performance and ANX score, it did not reach a significant level
(p> .01). Furthermore, the r value of the correlation, -.06 was very low, showing that
the LC and ANX score of the HI-ANX group did not seem to be in a strong
relationship between each other in the present study.
The above results of the two anxiety groups can be interpreted as follows. Since the same criteria was employed by Cheng (2007) and the present study in terms of the subject selection criterion for identifying the HI-ANX and LOW-ANX groups, the results of the two studies could then be compared. The same tendency (i.e., lower levels of the anxiety found in Cheng’s learners comparing all the participants in the two studies) was found with the results of the HI-ANX groups across the two studies.
The mean score of the HI-ANX group in Cheng (2007) was 3.50 which was lower than 4.00 obtained from the HI-ANX group in the present study. Nevertheless, the mean score of our LOW-ANX group was 2.29 lower than 2.43 obtained from the LOW-ANX group in Cheng (2007). Therefore, the more anxious listeners of the present study, i.e., the HI-ANX group, might be the main source to promote the anxiety intensity of the overall result. The reasons accounting for the higher level of listening comprehension anxiety of the overall participants than Cheng’s (2007) overall results provided earlier (due to different contexts and natures of the test given) thus can be applicable to this specific group of listeners. Concerning the lower level of anxiety found in the LOW-ANX group of the present study than that of Cheng’s (2007) study, a possible reason for this may lie in the different proficiency levels of the two subject groups. The more proficient listeners demonstrated a lower degree of listening comprehension anxiety; thus, our college participants of the LOW-ANX group might be more proficient than the participants of Cheng’s (2007) LOW-ANX group because Cheng’s (2007) subjects were senior high students. Therefore, the lower anxiety level was obtained in the present study than in Cheng (2007). However, this account may need to be testified by actually comparing the proficiency levels of the two groups of learners across the two studies in the future.
Moreover, in terms of the group effect found in the present study, the listening
comprehension performance of the LOW-ANX group outperformed the HI-ANX
group significantly. This further confirms the possible debilitating effects of listening comprehension anxiety on the performance as suggested by the earlier studies (Cheng, 2005; Cheng, 2007; Elkhafaifi, 2005; Vogely, 1998). However, a negative correlation (r=-.284) was found between the listening comprehension anxiety and performance of the LOW-ANX group but there was no significant correlation found in the HI-ANX group. Thus, the group effect regarding the performance of more anxious and less anxious listeners was not perfectly ensured in the present study, which was similar to the findings of Cheng (2007). This may be accounted for by the fact that some anxious listeners still had a good command of English listening ability. Through a closer examination of the raw data, it was found that some of the HI-ANX participants still performed well on the given LC test. The ways how they coped with their anxiety and achieved good performances are worthy of further investigation.
To sum up, with a slightly higher listening comprehension anxiety intensity obtained than the previous studies, the present study confirmed the claim that listening comprehension anxiety was harmful to learners’ listening comprehension performance in the previous studies. However, while the group effect showed that the LOW-ANX participants could perform significantly better than the HI-ANX participants on their listening comprehension test, it was found that some anxious listeners still could achieve good listening performance in the present study.
4.2 Effects of Personality and Linguistic Intelligence on Listening Comprehension Anxiety
This section addressed the second research question investigating effects of
personality (extraversion vs. introversion) and linguistic intelligence (higher
intelligence vs. lower intelligence) on the participant’s listening comprehension
anxiety. In the following, first, the overall performance of all the participants
concerning their personality traits and linguistic intelligence will be presented and discussed. Statistical comparisons will also be made by relating the participants’
personality traits and linguistic intelligence to their listening comprehension anxiety.
Also, in addition to the results provided above, the personality and linguistic intelligence level of the two anxiety groups will be further compared. This serves as a back-channeling to check the relationship among the participants’ listening comprehension anxiety, personality and linguistic intelligence.
4.2.1 Overall Performance of the Participants
Table 4-7 shows the descriptive results of the PER and LI scores of all the participants via the scale used in the questionnaire:
Table 4-7: Personality and Linguistic Intelligence
4Scores of All the Participants
N Mean SD
PER 1,400 3.25 .56
LI 1,400 2.85 .57
Overall speaking, all the participants in this study demonstrated a slightly extroverted personality since the mean score obtained was 3.25 out of 5.00. In other words, their PER level lay between “Neither Disagree Nor Agree” and “Agree”
considering whether they are extroverted. On the contrary, their LI scores were slightly lower (mean= 2.85 out of 5.00), indicating that on the average, the participants’ LI levels were not very high since 2.85 was between “Neither Disagree Nor Agree” and “Disagree” considering themselves as persons who possess higher degrees of linguistic intelligence. The results above were basically consistent with the previous analyses.
4 The mean scores of the two scales ranged from 1 to 5.
Probing into the relationship, the correlations of the participants’ PER and LI scores with their ANX scores are reported in Table 4-8 along with the scatter-graphs in Figure 4-2 and Figure 4-3:
Table 4-8: Correlation between Personality and Linguistic Intelligence with Anxiety Scores of All the Participants
PER LI
Pearson Correlation -.13(**) -.28(**)
Sig. (2-tailed) .000 .000
ANX
N 1,400 1,400
** Correlation is significant at the 0.01 level (2-tailed).
Table 4-8 shows that both the PER and LI scores of the participants correlated with their ANX scores negatively at a statistically significant level of .01 (PER:
p< .01; LI: p< .01). This indicated that when the participants’ ANX scores increased,
both their PER and LI scores decreased. A further demonstration can be seen from the
negative correlation lines which seemingly went from the top-left to the bottom-right
side in Figures 4-2 and 4-3. In addition, the correlation coefficients of the two cases,
(both ANX with PER and with LI) turned out to be low with only -.13 for the PER
level and -.28 for the LI level showing that there was a negative relationship between
the PER/LI and ANX scores of the participants, but the degrees were not strong.
4.00 2.00
ANX mean
4.00
2.00
PER mean
Figure 4-2: Correlation between Personality and Anxiety Scores of All the Participants
4.00 2.00
ANX mean
4.00
2.00
LI mean
Figure 4-3: Correlation between Linguistic Intelligence and Anxiety Scores of All the Participants
Seen from the two scatter-graphs, the plots in them were a bit loose since the
value of the correlation coefficient was not high. Also, as shown from the lines in the
two figures above, the direction of the line in Figure 4-3 seemed to be clearer going
from the top-left to the bottom-right than the direction of the line shown in Figure 4-2.
This further demonstrates the smaller value of correlation of the participants’ ANX scores with their PER traits than with their LI scores.
However, the correlation results can only tell the relationship between the two variables/constructs, not the direction and influential intensity of the variables concerned. When there are more than one independent variable with a single dependent variable and when the predicting power of the dependent variable from the independent variables is of interest, the multiple regression analysis is suggested to obtain the results (Brown, 1992). Table 4-9 below provides the regression results of the participants’ PER and LI scores with their ANX scores and Table 4-10 shows the collinearity results used to detect the overlapping problem of the two independent variables, PER and LI, and to assess their predicting power toward the ANX.
Table 4-9: Regression of Personality and Linguistic Intelligence with Anxiety Scores Change Statistics
R R Square R Square
Change
F
Change df1 df2 Sig. F Change
.28(a) .08 .08 60.15 2 1397 .000
a. Predictors: (Constant), LI, PER
As reported in Table 4-9 above, the multiple correlation coefficient between the
participants’ PER and LI and their ANX scores was .28. With the R square of .08, this
indicated that the combination of PER and LI scores can explain about only eight
percent of the variance in the ANX scores. Still, the significant result of the F-change
was .000 indicating that the model of the participants’ PER and LI successfully
predicted their ANX scores but only with a small amount of variance.
Table 4-10: Collinearity of Personality and Linguistic Intelligence with Anxiety Scores
Unstandardized Coefficients
Standardized Coefficients
Collinearity Statistics
B SD Beta
t Sig.
Tolerance VIF
(Constant) 4.213 .116 36.256 .000
PER mean -.06 .032 -.05 -1.774 .076 .903 1.107
LI mean -.31 .032 -.26 -9.733 .000 .903 1.107
With the VIF of both the PER and LI scores lower than the value of four (Miles
& Shevlin, 2001), there was no collinearity problem with the two independent variables, PER and LI. Although both the participants’ PER and LI scores were found to influence their ANX scores in the regression model, only the LI influence on the ANX scores reached to a significant level (β=-.26, p< .05). Moreover, with the beta absolute value of the PER score (-.05) smaller than that of the LI score (-.26) on the ANX scores, the predicting power of the PER on the ANX was confirmed smaller than that of the LI on the ANX.
More comparisons were made by grouping the participants into four different
levels according to their PER and LI scores. By so doing, the effects of the
participants’ PER and LI on their anxiety level can be explored. In the following, the
descriptive results regarding the relationship of the participants’ PER and ANX scores
are presented in Table 4-11 by dividing the participants into four PER groups (PER-1,
PER-2, PER-3, and PER-4) based on their personality levels from low to high
extroversion. Similarly, the participants were divided into four LI groups (LI-1, LI-2,
LI-3 and LI-4) according to their levels of linguistic intelligence from low to high as
shown in Table 4-13 to see the relationship between the participants’ LI and ANX
scores. In this way, each PER and LI group consisted of 350 participants. Although
there were the 1,400 participants, the participants of each PER and LI group were not
100% overlapped. That is to say, the participant of the PER-1 group might be categorized into the LI-4 group simply based on his or her personality/linguistic intelligence levels.
Table 4-11: Relationship between Personality and Anxiety Scores PER Group
5N ANX
Mean SD
PER-1 350 3.25 .64
PER-2 350 3.16 .68
PER-3 350 3.13 .64
PER-4 350 3.05 .72
It is clearly seen in Table 4-9 that the participants’ ANX level gradually decreased when their PER scores increased (PER-1 to PER-4). The ANOVA indicated a significant main effect for the participants’ PER on their ANX scores (F=5.284, p=.001). However, by means of scheffe post hoc comparisons, except for the mean
difference between PER-1 and PER-4 (p=.002 < .05), among all the other comparisons, i.e., PER-1 with PER-2, PER-1 with PER-3, PER-2 with PER-3, PER-2 with PER-4, PER-3 with PER-4, no significant differences were found as shown in Table 4-12 below:
5 Groups PER-1 to PER-4 were the participants with the bottom 25% of the PER (extroversion) level to those with the top 25% of the level. The PER mean score of PER-1, PER-2, PER-3, and PER-4 were 2.53, 3.07, 3.44, and 3.95 respectively.
Table 4-12: A Scheffe Post Hoc Multiple Comparison of Each PER Group (I)
PER_Group
(J) PER_Group
Mean Difference
(I-J)
SD Sig.
1 2 .08952 .05 .374
3 .12571 .05 .105
4 .19810(*) .05 .002
2 1 -.08952 .05 .374
3 .03619 .05 .917
4 .10857 .05 .205
3 1 -.12571 .05 .105
2 -.03619 .05 .917
4 .07238 .05 .564
4 1 -.19810(*) .05 .002
2 -.10857 .05 .205
3 -.07238 .05 .564
* The mean difference is significant at the .05 level.
* Dependent Variable: ANX mean
This suggested that although there was a decreasing tendency of the participants’ ANX scores, only their PER differences at the two extreme points (the introverted vs. the extroverted) showed a clear difference.
Table 4-13: Relationship between Linguistic Intelligence and Anxiety Scores LI Group
6N ANX
Mean SD
LI-1 350 3.42 .67
LI-2 350 3.14 .63
LI-3 350 3.08 .62
LI-4 350 2.96 .69
The participants’ ANX scores were also found to decrease when their LI scores
6 Group LI1 to LI4 were the participants with the bottom 25% of linguistic intelligence level to the ones with the top 25% of the level. The LI mean score of LI-1, LI-2, LI-3 and LI-4 were 2.14, 2.67, 3.00 and 3.58 respectively.
increased (from LI-1 to LI-4) as presented in Table 4-13. The ANOVA also indicated a significant main effect for the participants’ LI on their ANX scores (F=32.077, p=.000). With the scheffe post hoc comparisons, it was found that except for the mean
differences between LI-2 and LI-3, and between LI-3 and LI-4, all the other differences reached a statistically significant level as shown in Table 4-14 below:
Table 4-14: A Scheffe Post Hoc Multiple Comparison of Each LI Group (I)
LI_Group (J) LI_Group
Mean Difference
(I-J)
SD Sig.
1 2 .27943(*) .05 .000
3 .34229(*) .05 .000
4 .46667(*) .05 .000
2 1 -.27943(*) .05 .000
3 .06286 .05 .653
4 .18724(*) .05 .002
3 1 -.34229(*) .05 .000
2 -.06286 .05 .653
4 .12438 .05 .095
4 1 -.46667(*) .05 .000
2 -.18724(*) .05 .002
3 -.12438 .05 .095
* The mean difference is significant at the .05 level.
* Dependent Variable: ANX mean
Thus, the results indicated that a gradual tendency of the participants’
decreasing ANX scores was found in this case but the tendency was clear when the participants’ linguistic intelligence levels were lower.
Correlations of the ANX scores with the PER scores in each PER group and
with the LI scores in each LI group were also done, as shown in Table 4-15 and Table
4-16 below:
Table 4-15: Correlations between PER and ANX Scores of Each PER Group
PER Group ANX
1 PER Pearson Correlation -.19(**)
Sig. (2-tailed) .000
N 350
2 PER Pearson Correlation -.04
Sig. (2-tailed) .498
N 350
3 PER Pearson Correlation -.03
Sig. (2-tailed) .636
N 350
4 PER Pearson Correlation -.04
Sig. (2-tailed) .441
N 350
** Correlation is significant at the 0.01 level (2-tailed).
Table 4-16: Correlations between LI and ANX Scores of Each LI Group
LI Group ANX
1 LI Pearson Correlation -.24(**)
Sig. (2-tailed) .000
N 350
2 LI Pearson Correlation -.09
Sig. (2-tailed) .096
N 350
3 LI Pearson Correlation -.14(**)
Sig. (2-tailed) .008
N 350
4 LI Pearson Correlation -.10
Sig. (2-tailed) .061
N 350
** Correlation is significant at the 0.01 level (2-tailed).
It was found that all the correlations obtained here were negative. In terms of
the correlation of the participants’ ANX scores with their PER scores, only the ANX
scores of PER-1 correlated significantly with their PER (p< .01). For the other groups,
PER-2, PER-3, and PER-4, the significant results were not found. This implied that
only the very introverted participants (as those in PER-1) manifested their anxiety to be closely related to their introverted characteristics. With respect to the participants’
LI scores, a significant correlation was also found between the ANX and LI scores of LI-1 (p< .01), showing that for those participants who possessed low LI scores, their apprehensive feelings toward the LC test were more clearly shown to be affected by their LI scores. What’s more, a significant correlation was also found between the ANX and LI scores of LI-3 (p< .01), but not of LI-2 and LI-4 groups. However, in terms of the three significant cases mentioned above, the correlation coefficients turned out to be very low from -.14 to -.24 indicating the participants’ ANX scores and their LI scores were not in a strong relationship so were their ANX scores and their PER scores
The interpretations of the results presented above are discussed in the following.
First, the participants in the present study were found to be slightly extroverted since
the mean score was 3.25. The participants’ personality was found to be a main factor
affecting their listening comprehension anxiety via the ANOVA and correlation
analyses. Specifically speaking, the more extroverted participants were less anxious
when they engaged in English listening comprehension activities and their more
introverted counterparts showed the opposite results. These results basically supported
Eysenck’s (1961) proposal of the personality type in associating with anxiety. In
addition, the results also complemented MacIntyre and Charos’ (1995) suggestions
that the personality should be a factor affecting the students’ language learning
anxiety. Also, aside from the explanation from the neuroticism perspective (Eysenck,
1961) that anxiety was caused from the interaction of over-reactivity and strong
condition-ability, another possible explanation may be the more functional strategies
the extroverted persons dare to adopt like social-affective strategies (Wakamoto, 2000)
which can be used to help alleviate the apprehensive feelings by seeking for the outer
supports. However, with a closer examination, the findings further suggested that only when the personality differences were in the two extremes, can the significances in their listening comprehension anxiety be revealed. These conditions may be explained by considering the “fuzzy” area which makes it hard to clearly consider a person extrovert or introvert. As Wakamoto (2000) indicated, personality type can only be judged by locating a person in the extroverted or introverted continuum because in reality the two types of personality can co-exist. Therefore, a big gap of the personality assessed in the scale is needed to relate to a clear difference in their listening comprehension anxiety.
Second, the linguistic intelligence score of the participants in the present study showed a moderate to slightly weak intensity with a mean score of 2.85. That is to say, the participants were the learners who were at a moderate to slightly weak linguistic intelligence level. Besides, their linguistic intelligence level was found to be a main effect accounting for their listening comprehension anxiety according to the ANOVA, correlation, and regression analyses. Those who were found to be higher at the linguistic intelligence level were less anxious listeners and those who were lower at the linguistic intelligence level were more anxious listeners. Because there are few previous studies of the effect of linguistic intelligence on listening comprehension anxiety, it is hard to compare the present results with the previous findings.
Nevertheless, the results obtained here are in accordance with the basic concept of the MI theory proposing the benefit of MI into language related aspects (Hall-Haley, 2001;
Hall-Haley, 2004). The rationale for this is simple and straightforward. As indicated
by Armstrong (2000), linguistic intelligence is closely related to language learning
which helps to master structures, sounds, and meanings of language. Those who
possess a higher degree of linguistic intelligence are good at dealing with language
and the intelligence allows them to have more processing capacity to tackle with the
given task resulting in lower anxiety (MacIntyre, 1999; Tobias, 1986).
Finally, concerning the relationship of personality and linguistic intelligence with listening comprehension anxiety, the correlation coefficients of the participants’
personality and linguistic intelligence with their listening comprehension anxiety were small (r=-.13 and r=-.28 respectively). By comparing the values, the negative correlation coefficient of the participants’ listening comprehension anxiety with their linguistic intelligence was smaller than that with the participants’ personality traits.
This shows that the linguistic intelligence may be in a stronger relationship with the anxiety than the personality traits with the anxiety.
7The regression analysis also indicated a stronger predicting power of the linguistic intelligence than the personality on the listening comprehension anxiety. As a result, the present study confirms that with the two independent factors (personality and linguistic intelligence), linguistic intelligence is a more dominant factor affecting the learners’ listening comprehension anxiety. Like the reasons mentioned earlier, this may be explained by the born-nature of insufficient and difficult conceptualization of personality. Moreover, linguistic intelligence is definitely a more language-oriented construct by nature. Thus, it is more dominant than personality to account for a language-related construct, listening comprehension anxiety.
4.2.2 General Performance of the Two Anxiety Groups
With regard to the relationship of the participants’ PER and LI scores with the ANX scores, Table 4-17 presents the descriptive results of the HI-ANX and LOW-ANX participants’ PER and LI levels:
7 The correlation coefficient found here was negative. That meant the smaller the value was, the stronger the relationship was.
Table 4-17: Personality and Linguistic Intelligence Scores of the Two Anxiety Groups
Group Mean SD
PER 3.18 .61
HI-ANX
(n=350) LI 2.66 .62
PER 3.38 .55
LOW-ANX
(n=350) LI 3.05 .57
As shown above, the mean score of the participants’ PER level obtained from the LOW-ANX group, 3.38, was higher than that of the HI-ANX group, 3.18, and the difference between the two groups reached a statistically significant level (t=4.541, p<.01) via the t-test comparison, showing that those who were less anxious were more
extroverted than those who were more anxious. The PER level of both groups revealed the slightly extroverted characteristics similar to the overall findings found in the previous section. In the same vein, the mean score of the LI level obtained from the LOW-ANX group, 3.05, outnumbered that of the HI-ANX group, 2.66. The t-test comparison also yielded a significant difference between the two groups (t=8.484, p<.01), indicating that the LI level of the LOW-ANX group was indeed higher than
the LI level of the HI-ANX group. In other words, the less anxious participants were more likely to possess a higher degree of linguistic intelligence and the more anxious participants, in contrast, were more likely to have a lower degree of linguistic intelligence.
As for the HI-ANX and LOW-ANX groups’ performances regarding their ANX
correlating with the PER and LI scores, the results are shown in Table 4-18 below:
Table 4-18: Correlation between Personality and Linguistic Intelligence with Anxiety Scores of the Two Anxiety Groups
ANX_Group PER LI
Pearson
Correlation -.03 -.15(**)
HI-ANX
(n=350) ANX
Sig. (2-tailed) .605 .007 Pearson
Correlation -.10 -.22(**)
LOW-ANX
(n=350) ANX
Sig. (2-tailed) .053 .000
** Correlation is significant at the 0.01 level (2-tailed).
It can be seen from Table 4-18 that the ANX scores of the HI-ANX and LOW-ANX groups were significantly and negatively correlated with their LI scores (HI-ANX: p< .01; LOW-ANX: p< .01), although the r values were low (HI-ANX: r=
-.15; LOW-ANX: r=-.22). These results showed that the ANX scores of both anxiety groups would decrease when they possessed a higher degree of linguistic intelligence and vice versa. In addition, it was found that the correlation coefficient of the ANX and LI scores in the LOW-ANX group was lower than that of the HI-ANX group, implying that the LOW-ANX participants’ listening comprehension anxious feelings might be more closely related to their LI level. Nevertheless, though negative correlations were found between both anxiety groups’ ANX and PER levels, the results did not reach a significant difference.
Taking the perspective of the learners into consideration, it was found that some interviewees reported that their PER traits were related to their ANX scores, but only nine (5 interviewees of the HI-ANX group; 4 interviewees of the LOW-ANX group
8) had this opinion (see HI-ANX4 for example). The other 11 interviewees (5 of the HI-ANX group; 6 of the LOW-ANX group) considered that there was no relationship
8 The two groups interviewed, were randomly selected from the two anxiety groups, HI-ANX and LOW-ANX.
between their ANX scores and their PER scores (see HI-ANX1 and LOW-ANX6 for example). Different viewpoints of the association of the PER with the ANX are illustrated in Excerpt 4-1 below:
Excerpt 4-1
91. I don’t think my extroverted personality has anything to do with my listening comprehension anxiety since they are two different things. (taken from HI-ANX1).
2. I am a more extroverted person. This characteristic (being extroverted) helps me not be so anxious when listening to English because extroverted people dare to seek for help by asking for clarity… (taken from HI-ANX4)
3. I think the personality would not affect your listening comprehension anxiety because I don’t consider there is relationship between them. (taken from LOW-ANX6)
As for the LI factor, most of the interviewees reported that their LI did help reduce their ANX. Fifteen (8 interviewees of the HI-ANX group; 7 interviewees of the LOW-ANX group) mentioned that their ANX was influenced by their LI. Some examples can be seen from Excerpt 4-2 in the following (see HI-ANX9 and LOW-ANX5 for example). However, the other five (2 of the HI-ANX group; 3 of the LOW-ANX group) stated that their LI level was not relevant to their ANX.
Excerpt 4-2
1. I consider my lower level of linguistic intelligence as a factor influencing my listening comprehend anxiety because this characteristic leads me to be a poor listener of English. (taken from HI-ANX9)
2. I consider my linguistic intelligence level is higher and this higher level of linguistic intelligence influence the listening comprehension anxiety… a person with higher level of linguistic intelligence can easily get the gist of what is heard and will not be very anxious in listening. (taken from LOW-ANX5)
9 Since the interviews were conducted in Chinese, the excerpts were translated into English by the researcher.
Moreover, it was found that in terms of the PER factor, among the ten HI-ANX interviewees, five of them reported that they were extroverted and five, introverted.
The same pattern was reported by the LOW-ANX interviewees with five saying that they were extroverts and the other five, introverts. As for the LI factor, four of the HI-ANX interviewees considered themselves as persons with a high level of LI, and six of them with a low level of LI. However, for the LOW-ANX interviewees, except for one interviewee, all of them reported that they had a high level of LI. These results showed that the participants’ LI seemed to be more related to their ANX than their PER since more interviewees perceived LI as something related to anxiety. Also, more interviewees pointed out that their PER did not seem to influence their ANX. The group effect also confirmed these results that both the HI-ANX and LOW-ANX interviewees included the extroverted and introverted participants in a half-and-half manner, but for the LI, the LOW-ANX interviewees tended to be formed by the participants with a higher level of LI only.
The results gained from the two anxiety groups can be discussed as follows.
The group effect regarding the personality and linguistic intelligence of the two
groups was ensured since it was found that the higher or lower anxiety participants
would possess different levels of personality and linguistic intelligence. More
specifically, for the more anxious listeners, they were found less extroverted and also
with poor linguistic intelligence. The other group of the less anxious listeners was the
opposite. They were found more extroverted and also to be with better linguistic
intelligence. These results were then confirmed by the correlation analyses showing
that there was a negative correlation between the participants’ listening
comprehension anxiety and their personality and between their listening
comprehension anxiety and their linguistic intelligence. Nevertheless, considering the
intensity of the listening comprehension anxiety with the participants’ personality and
linguistic intelligence, it was found that both anxiety groups yielded more significant results of correlating their listening comprehension anxiety with their linguistic intelligence than personality level. The interview data also showed that a more consistent association of the anxiety with the participants’ linguistic intelligence than with their personality level. More interviewees voiced for the influence of the linguistic intelligence level than the influence of the personality on their listening comprehension anxiety. These results basically echoed with the findings of the present study mentioned earlier that there was a relationship between listening comprehension anxiety with personality and linguistic intelligence level. And the participants’
linguistic intelligence was found to be more dominant than personality in affecting the anxiety. In addition, the present findings highlighted again the previous claims calling attention to be given to language learning anxiety and personality (Eysenck, 1961;
MacIntyre & Charos, 1995) and possible influences of intelligence on the language learning since listening comprehension anxiety can surely be defined as a language learning related construct (Armstrong, 2000; Hall-Haley, 2001; Hall-Haley, 2004).
Summing up, the participants’ personality and linguistic intelligence were found
to be related to their listening comprehension anxiety. With the results obtained from
two anxiety groups, the same findings were confirmed. Specifically speaking, the
more anxious listeners were less extroverted and possessed a lower level of linguistic
intelligence; the less anxious listeners were more extroverted and had a higher level of
linguistic intelligence. In addition, compared the participants’ personality and
linguistic intelligence with their anxiety, linguistic intelligence was found to be a
more dominant factor affecting their listening comprehension anxiety than
personality.
4.3 Other Factors and Solutions of Listening Comprehension Anxiety
To address the third research question, some other factors and solutions in relation to the participants’ listening comprehension anxiety will be reported in this section. These results were obtained from the interviews with the participants. The factors affecting the participants’ listening comprehension anxiety will be described first, followed by their suggested ways to reduce listening comprehension anxiety.
Finally, discussions will be made during the course of presenting the interview data in attempt to frame the results obtained into a clear assertion.
4.3.1 Other Factors Affecting Listening Comprehension Anxiety
In addition to personality and linguistic intelligence, the interviewees reported some other factors that might affect their listening comprehension anxiety. Below are seven factors that were reported to affect the participants’ listening comprehension anxiety.
Factor 1: Inherent Features of Testing
The most influential factor mentioned by the interviewees was the testing itself.
This finding is consistent with the studies of general language anxiety (e.g. Horwitz et al.’s, 1986; Young, 1991) and some studies of listening anxiety (e.g. Cheng, 2005;
Vogely, 1998). Several interviewees (11 out of 20) indicated that they experienced
listening comprehension anxiety when they thought it was a test (see HI-ANX3,
LOW-ANX5 and LOW-ANX8 for example). They suggested that they felt anxious
about listening due to the pressure and anxiety-provoking nature in a testing setting
(see HI-ANX3 for example). In addition, the reason why they were afraid of tests was
mainly due to the fear of not being able to understand what they were going to hear
(see HI-ANX6 and LOW-ANX8 for example). Interviewees of the HI-ANX group
and the LOW-ANX group were both anxious about test-related events or activities.
Excerpt 4-3 below shows some examples obtained from the HI-ANX and LOW-ANX interviewees:
Excerpt 4-3
1. I consider listening is definitely anxiety-provoking especially when it is related to testing. I dislike the evaluations related to test stuffs because … tests are not really practical when they are applied to daily communication. (taken from HI-ANX3) 2. Yes! I experience listening comprehension anxiety for I am afraid of not getting
what I hear. This may lead to the poor performance in my grades. But, if the listening has nothing to do with the tests, it doesn’t matter whether I comprehend it or not. (taken from HI-ANX6)
3. Generally speaking, I don’t experience listening comprehension anxiety unless I am engaged in the tests …. (taken from LOW-ANX5)
4. I feel only a little anxious when listening to English comprehension tests.
Because I am afraid of being not be able to understand the given message in listening tests. (taken from LOW-ANX8)
Factor 2: Fast Speed
Another listening comprehension anxiety provoking factor frequently mentioned is fast speed. Many interviewees reported that their anxiety stemmed from the fast speed of listening to English, which often led to their incomprehensibility or misunderstanding about the listening texts (see HI-ANX1 and LOW-ANX9 for example). Cheng (2005) also found that her subjects’ listening comprehension anxiety was aroused by the fast speed of the listening materials. A total of 7% and 17% of her subjects’ anxiety were caused by the fast speed of audio and video input respectively.
The examples of the anxiety of the participants in the present study related to fast
speed are listed in Excerpt 4-4 below:
Excerpt 4-4
1. I don’t quite understand the meaning of the listening text especially at its fast speed … and this makes me feel anxious. (taken from HI-ANX1)
2. Another reason for my listening comprehension anxiety is the fear of incomprehension due to the fast speed. (taken from LOW-ANX9)
Factor 3: Unclear articulation
As found in Cheng (2005), unclear articulation was one of the reasons that attributed to her subjects’ listening comprehension anxiety. Unclear listening, junctures and accents of foreigners were also perceived to be causes to make listening unclear to our participants, and thus led to their listening comprehension anxiety (see HI-ANX4 and HI-ANX10), as shown in Excerpt 4-5:
Excerpt 4-5
1. …accents and junctures of foreigners increase my anxious feelings when I listen to English…they make listening hard to be comprehend… (taken from HI-ANX4) 2. … sometimes their articulation may not be so clear for me. Or they may have
some junctures which make their talk incomprehensible to me. (taken from HI-ANX10)
Factor 4: Difficulty Level
Some interviewees also mentioned the difficulty level of listening texts.
Lengthy texts (see HI-ANX7 for example), texts beyond their current levels (see
LOW-ANX7 for example), or difficult activities or texts (see LOW-ANX6 for
example) were all the causes for their listening comprehension anxiety. Cheng (2005)
also indicated the difficulty of listening as one of the sources of her subjects’ listening
comprehension anxiety and the difficulty of input as she pointed out were unfamiliar
words or expressions and difficult or unfamiliar topics. In Vogely (1998), the level of
difficulty was called for attention. 11% of the sources reported by her subjects
attributed their listening comprehension anxiety to this cause. Some examples
regarding the difficulty level mentioned by our participants are presented in Excerpt 4-6 below:
Excerpt 4-6
1. In my opinion, the shorter sentences allow me adequate time to process their meanings. Thus, I feel more relax and not anxious! … As for the longer sentences, they are difficult! … it’s hard for me to put all the information contained together to comprehend the given message, which may lead to my anxiety… (taken from HI-ANX7)
2. I don’t feel quite anxious when listening. Only the very difficult activities or listening texts which may lead to my incomprehension would I feel a bit anxious.
(taken from LOW-ANX6)
3. If the difficulty level of the listening is beyond my level… this can easily lead to more pressure for me… (taken from LOW-ANX7)
Factor 5: Inappropriate Strategy Use
With regard to the wrong strategy use, the interviewees reported that sometimes, they might focus on comprehending a single vocabulary word, which often made them miss other important information (see HI-ANX5 and HI-ANX7 for example).
Song (2005) also generated similar results with that learners’ causes of listening comprehension anxiety came from their failing to grasp each word heard. Aside from missing some information/details, some other inappropriate strategies like note-taking and occasional incomprehension were reported in Cheng (2005) and were found in the present study as well, as shown in excerpt 4-7 in the following:
Excerpt 4-7
1. Sometimes, when I miss one single word, I get panic and lose all the following easy English. (taken from HI-ANX5)
2. When I encounter an unknown vocabulary, I tend to focus on it and want to
figure out its meaning. As a result, I miss all the coming information and can not
comprehend the message. (taken from HI-ANX7)
Factor 6: Lack of Confidence
Some interviewees had negative or false perception about themselves (see LOW-ANX2 for example) by concerning themselves inferior in listening (see HI-ANX6 and HI-ANX8 for example). As documented in the literature, low self-esteem may arouse learners’ listening comprehension anxiety (Song, 2005) or foreign language learning anxiety (Liu, 2006). Some examples of this type are given in Excerpt 4-8:
Excerpt 4-8
1. I always consider myself as a poor English learner. So when it comes to listening, I presuppose that the task is too difficult for me. Therefore, I am always not very serious about English listening. (taken from HI-ANX6)
2. Because I am afraid that I am inferior to other classmates. (taken from HI-ANX8)
3. I would feel nervous and worried about what if I don’t have the ability to comprehend the heard passage at the beginning of the listening. (taken from LOW-ANX2)
Factor 7: Inattention
Regarding the causes for the participants’ inattention, the interviewees indicated that sometimes their physical condition made them hard to concentrate on listening comprehension tasks, and naturally caused their listening comprehension anxiety (see HI-ANX5 and HI-ANX9), as shown in Excerpt 4-9:
Excerpt 4-9
1. If I don’t get enough sleep, it is hard for me to be attentive to the listening comprehension tasks. (taken from HI-ANX5)
2. My listening ability is very poor, so I always don’t get what the listening tasks are all about. Especially when I am distracted by other things, I can hardly get the meanings from listening passages. (taken from HI-ANX9)
To sum up, most of the participants interviewed indicated their anxiety
stemmed from the testing itself (11 out of 20). In general language learning, Young (1991) claimed that testing was one of the main sources of language learning anxiety.
Similar results were found in Cheng (2005), who explored EFL Taiwanese college students’ sources of listening comprehension anxiety and found that testing/evaluation was the source most frequently referred for anxiety which went up to 30% and 33%
(for audio and visual activities respectively) of all the sources. On the contrary, Vogely (1998) in a western context found that among the entire source categories for her subjects’ listening comprehension anxiety, testing was not influential with only 2%. This discrepancy is quite understandable since in Taiwan, unlike in the western context, we are under a test-dominant system. Students here are under pressure of numerous tests for that these tests are always highly valued. Scoring high is often viewed as a promising ticket for life success in this increasingly competitive society.
Since our participants just graduated from senior high schools, they might still perceive some leftover of tests filling in every blank of their high school classes in order to perform well on the Joint College Entrance Examination. Another reason accounted for this discrepancy may be due to the culture difference between the oriental and western students including their perception about tests, personality traits, and learning habits as mentioned earlier in this chapter. Comparing the two anxiety groups, we found that the HI-ANX interviewees provided more entries
10than the LOW-ANX interviewees (HI-ANX: 15 entries; LOW-ANX: 12 entries). In addition, the HI-ANX students’ factors affecting their anxiety were loaded in a wider manner than the LOW-ANX students’ (HI-ANX: all the 7 factors; LOW-ANX: 4 factors).
This result can be explained by the fact that the HI-ANX interviewees felt more anxious, and thus they might have more factors to report. The LOW-ANX students
10 The number of the entries regarding the other types of factors and solutions for listening comprehension anxiety was roughly calculated by counting the idea units found. In addition, only the idea which was mentioned more than twice by the interviewees were reported here.
