4.2 Results of GJT
4.2.1 Scores of GJT
This subsection demonstrated the results of GJT by dividing into four sections according to 3- to 6-word sentences. Each of them includes the mean scores and descriptive analysis. To reiterate the scoring method of GJT, there were five questions consisting of the Korean word order and five questions consisting of the Chinese word order. The maximum score of the Korean word order and the Chinese word order is five each in all the four sentence complexity. If the learners chose a Japanese sentence taking Korean word order as correct, one point was added to Korean word order. If they chose a Japanese sentence taking the Chinese word order as correct, one point was added to Chinese word order. The following section provided the statistical results of each type of sentence complexity, starting from 3-word sentences.
As for the result of 3-word sentences, the mean scores of L1 Taiwanese, L1 Korean, and native Japanese are shown in Figure 13. In the scores of Korean word order, each group gained 4.61, 3.2, and 4.93 respectively, while the scores of the Chinese word order in each group dropped to 1.00, 1.49, and 0.1 were respectively.
Note. Max = 5
Figure 13. The mean scores of GJT: 3-word sentences
In terms of the between-subjects effects, p value was found just p = .05 as shown in Table 25 (F(1,90) = 3.091, p = .05). In the within-subjects effects, Table 26 indicates that there is a significant difference in the scores of the Korean word order and the Chinese word order both in the two experimental groups (F(1,90) = 188.693,
p < .05). The interaction effect was revealed (F(1,61) = 10.173, p < .05).
Table 25
Between-subjects effects of GJT: 3-word sentences
Source
df
SS MS Fp
Table 26
Within-subjects effects of GJT: 3-word sentences
word order
df
SS MS Fp
word order 1 561.933 561.933 188.693 .000**
word order * group
1 60.590 30.295 10.173 .000**
Error(source) 61 268.023 2.978
Note.* p<.05. **p<.01
The results of a post hoc analysis, as presented in Table 27, showed that there was a slightly significant difference between L1 Chinese learners and native Japanese while no significant difference was found between L1 Korean learners and native Japanese and between the two experimental groups. In sum, the results of the within-subjects and between-within-subjects largely exhibited the similar pattern to OT.
The same tendency was observed in the 4-word sentences. Figure 14 exhibited that both L1 Taiwanese and L1 Korean were more likely to choose Japanese sentences with the Korean word order as correct answers (4.73 and 3.14 respectively) than those with the Chinese word order (0.58 and 1.46 respectively) while native Japanese received 4.97 and 0.07 in the Korean word order and the Chinese word order respectively.
Table 27
Tukey HSD comparison analysis of GJT: 3-word sentences
95% Confidence Interval (I) group (J) group
Mean Difference (I-J)
Std.
Error Sig. Lower Bound Upper Bound
Taiwanese Korean .07 .119 .829 -.21 .35
Japanese .29* .119 .048* .00 .57
Korean Taiwanese -.07 .119 .829 -.35 .21
Japanese .22 .122 .183 -.07 .51
Japanese Taiwanese -.29* .119 .048* -.57 .00
Korean -.22 .122 .183 -.51 .07
Note. *p < .05
Note. Max = 5
Figure 14. The mean scores of GJT: 4-word sentences
When it comes to the between-subjects effects, no significant difference was observed as in Table 28 (F(1,90) = 1.273, p > .05). Table 29 indicates that the scores of the Korean word order significantly differed from those of the Chinese word order both in L1 Taiwanese and L1 Korean (F(1,90) = 22.128, p < .05). The interaction was also found (F(1,90) = 12.760, p < .05). A post hoc analysis confirmed that no significant difference was detected as in Table 30.
0
Table 28
Between-subjects effects of GJT: 4-word sentences
Source
df
SS MS Fp
Intercept 1 1271.455 1271.455 3420.799 .000**
Group 2 .473 .473 1.273 .285
Error 90 .372 .372
Note.* p<.05. **p<.01
Table 29
Within-subjects effects of GJT: 4-word sentences
word order
df
SS MS Fp
word order 1 625.970 625.970 228.128 .000**
word order * group
2 70.024 35.012 12.760 .000**
Error(source) 90 246.955 2.744
Note.* p<.05. **p<.01
Even in the scores of complex of sentences, similar patterns were found. In terms of the mean scores of the 5-word sentences, native Japanese have achieved 4.77 and 0.27 in the Korean word order and Chinese word order respectively. As for L1 Taiwanese and L1 Korean learners, the Korean word order (4.79 and 3.09 respectively) was higher than in the Chinese order (1.06 and 1.49 respectively) as shown in Figure 15 below.
Table 30
Within-subjects effects of GJT: 4-word sentences
(I) group (J) group
Mean Difference (I-J)
Std.
Error Sig. 95% Confidence Interval Lower Bound Upper Bound
Taiwanese Korean .07 -.03 .109 .954 -.29
Japanese .29* .13 .109 .433 -.12
Korean Taiwanese -.07 .03 .109 .954 -.23
Japanese .22 .17 .111 .297 -.10
Japanese Taiwanese -.29* -.13 .109 .433 -.39
Korean -.22 -.17 .111 .297 -.43
Note. p < .05
Note. Max = 5
Figure 15. The mean scores of GJT: 5-word sentences
The between-subjects effects showed that there was no significant difference in the scores among L1 Taiwanese, L1 Korean, and native Japanese shown in Table 31 (F(1,90) = 2.957, p > .05). In contrast, Table 32 presented the within-subjects effects. It indicates that both of the two experimental groups picked up more Japanese sentences taking the Korean word order as correct than those consisting of the Chinese word order (F(1,90) = 192.395, p < .05). The same holds true in the interaction effect (F(1,90) = 10.100, p < .05). A post hoc analysis showed that the significant difference was found between L1 Chinese and native Japanese, though very slightly, while no significant difference was found between L1 Korean and naïve Japanese, and between L1 Chinese and L1 Korean as presented in Table 33.
0
Table 31
Between-subjects effects of GJT: 5-word sentences
Source
df
SS MS Fp
Intercept 1 1355.739 1355.739 1489.134 .000**
Group 2 5.385 2.692 2.957 .057
Error 90 81.938 .910
Note.* p<.05. **p<.01
Table 32
Within-subjects effects of GJT: 5-word sentences
word order
df
SS MS Fp
word order 1 525.358 525.358 192.395 .000**
word order * group
2 55.158 27.579 10.100 .000**
Error(source) 90 245.756 2.731
Note.* p<.05. **p<.01
Finally, Figure 16 shows the results of the 6-word sentences. L1 Taiwanese learners chose more Japanese sentences with the Korean word order as grammatically correct than those with the Chinese word order (4.61 and 0.64 respectively). The same holds true for L1 Korean learners as illustrated in Figure 16 below (3.31 and 1.34 respectively).
Table 33
Tukey HSD analysis of GJT: 5-word sentences
(I) group (J) group
Mean Difference (I-J)
Std.
Error Sig. 95% Confidence Interval Lower Bound Upper Bound
Taiwanese Korean .26 .170 .290 -.15 .66
Japanese .41* .170 .049* .00 .81
Korean Taiwanese -.26 .170 .290 -.66 .15
Japanese .15 .174 .666 -.27 .57
Japanese Taiwanese -.41* .170 .049* -.81 .00
Korean -.15 .174 .666 -.57 .27
Note. *p < .05
Note. Max = 5
Figure 16. The mean scores of GJT: 6-word sentences
Although there was no significant difference among the three groups as Table 34 showed (F(1,90) = .263, p > .05), the descriptive statistics supported this finding, revealing that there was a significant difference between the scores of participants as shown in Table 35 (F(1,90) = 199.677, p < .05) as well as the interaction effect (F(1,90) = 7.208, p < .05). Table 36, which presented the results of a post hoc analysis, also verified this result.
Table 34
Between-subjects effects of GJT: 6-word sentences
Source
df
SS MS Fp
Intercept 1 1321.500 1321.500 199.677 .000**
Group 2 .286 .143 7.208 .001**
Error 90 49.047 .545
Note.* p<.05. **p<.01
Table 35
Within-subjects effects of GJT: 6-word sentences
word order
df
SS MS Fp
word order 1 616.708 616.708 199.677 .000**
word order * group
2 44.526 22.263 7.208 .001**
Error(source) 90 277.968 3.089
Note.* p<.05. **p<.01
In summary, 4.2.1 presented statistical analysis of the mean scores of GJT.
Overall, the results showed that within-subjects effects exhibited significant difference across all the four types of sentence structures as well as the interaction effects. In contrast, between-subjects effects showed that no significant difference was identified between the two experimental groups and the native control with some minor exceptions. Summary of p values and Tukey HSD analysis in GJT were summarized in Table 37 and Table 38 respectively.
Table 36
Post hoc analysis of GJT: 6-word sentences
(I) group (J) group
Mean Difference (I-J)
Std.
Error Sig. 95% Confidence Interval Lower Bound Upper Bound
Taiwanese Korean -.10 .132 .749 -.41 .22
Japanese -.05 .132 .936 -.36 .27
Korean Taiwanese .10 .132 .749 -.22 .41
Japanese .05 .135 .927 -.27 .37
Japanese Taiwanese .05 .132 .936 -.27 .36
Korean -.05 .135 .927 -.37 .27
Note. p < .05
Table 37
Summary of p values in GJT (between- and within-subjects)
Number of words Between-subjects Within-subjects
3 words
p = .050 p = .000**
4 words
p = .285 p = .000**
5 words
p = .057 p = .000**
6 words
p = .770 p = .000**
Note.* p<.05. **p<.01
Table 38
Summary of p values in GJT (Tukey HSD analysis)
Tukey HSD Taiwanese-
Japanese control group
Korean-
Japanese control group
3 words
p = .048* p = .122
4 words
p = .433 p = .297
5 words
p = .049* p = .666
6 words
p = .936 p = .927
Note.* p<.05. **p<.01