4.2 Results of GJT
4.2.2 RT of GJT
4.2.2.1 Data-cleaning process
A one-way ANOVA was performed to see whether there was a significant difference between the two experimental groups and a control group in RT of GJT.
Regarding the process of data-cleaning, the following process was taken: (1) exclude RTs with incorrect answers, (2) exclude RTs of fillers, (3) exclude RTs longer than 1500ms and faster than 100ms, and (4) data points of RTs which is out of 2.5 standard deviations are replaced with the cutoff point.
First of all, RT data points of incorrect answers were removed, followed by those of fillers. After removing the RT data points of incorrect answers and fillers, outliers were excluded. To remove outliers, RTs that took too much time were taken off. In this study, RT data points that are over 1500ms were removed as a cutoff point.
When deciding the cutoff, the present study referred to Iseki’s experimental psycholinguistic study (2003), whose research design was close to the present study in terms of the design of software for yes/no judgment and the number of lexical units consisting of a sentence. Using the design of his yes/no judgment, participants of native Japanese were asked to answer whether or not a sentence presented is meaningful as quickly as possible with the time limit up to 1500ms. The number of lexical units consisting of a sentence in Iseki’s study was close to that of the present study, ranging from two to five lexical units. He assumed that there was a higher chance that spending over 1500ms is so long that learners can afford to process a sentence with the reliance on explicit knowledge. This means that learners do not
point in this thesis. Likewise, RT data points which are faster than 100ms were removed. As claimed by researchers, psychological process requires at least 100ms (Ratcliff, 1993; Whelan, 2008). It is possible that RTs which were faster than 100ms would be that learners accidentally pressed a button without processing a target language, and, therefore, that these RTs do not result from genuine online processing of morphosyntactic knowledge. For this reason, RT data points that are below 100ms were removed as outliers. Finally, RTs that are 2.5 standard deviations from the mean of the same participants were replaced with the cutoff point. This was done because previous studies examining how foreign learners of Japanese processed its word order adopted this standard deviation as a cutoff (e.g., Iseki, 2003; Koizumi et al., 2006). To sum up, the RT data point of incorrect answers, fillers, the ones over 1500ms and below 100ms were excluded. Also, the ones that are out of 2.5 standard deviations were replaced with the cutoff point.
4.2.2.2 RT of GJT (Japanese sentences in the Korean word order)
This section illustrates (1) the comparison of means of RT among Taiwanese learners with advanced Korean, Korean learners with advanced Chinese, and native Japanese speakers in Japanese sentences in the Korean word order, and (2) its statistical results of a one-way ANOVA and a post hoc analysis in each type of sentence complexity, the 3- to 6-word sentences. For host hoc analysis, Tukey HSD was used. The mean difference is significant at the 0.05 level. At the end of the section presented a summary of all the results of RT of GJT. The following presented the results of the 3- to 6-word sentences.
The mean of RT of the 4-word sentences were given in Figure 17. Mean RTs of the experimental group have attained 449.77ms for Taiwanese learners, 450.49ms for Korean learners while Japanese natives’ mean RT is 314.625ms.
Figure 17. Means of RT: 3-word sentences (Korean word order)
The result of ANOVA revealed a significant difference among the three groups as presented in Table 39 (F(2,372) = 14.693, p < .05). A post hoc tests revealed that, as presented in Table 40, significant difference existed between Taiwanese learners with advanced L2 Korean and native Japanese, and between Korean learners with advanced L2 Chinese and native Japanese.
0 100 200 300 400 500
Taiwanese learners Korean learners Native Japanese
ms
Table 39
Result of RT with ANOVA: 3-word sentences (Korean word order)
df SS MS F
p
Between Groups 2 1627317.088 813658.544 14.693 .000**
Within Groups 372 20600637.621 55378.058
Total 374 22227954.709
Note.* p<.05. **p<.01
The same result was found in RTs of the 4-word sentences (in Figure 18).
ANOVA also showed that there was a significant difference among Taiwanese
Table 40
Tukey HSD analysis of RT: 3-word sentences (Korean word order)
(I) group (J) group
Mean Difference (I-J)
Std.
Error Sig. 95% Confidence Interval Lower Bound Upper Bound
learners who have attained advanced L2 Korean, Korean learners with advanced Chinese, and native Japanese speakers (F(2,373) = 9.787, p < .05). as in Table 41 The results of a post hoc analysis also identifies that significant difference in RT was found between Taiwanese learners with L2 Korean/L3 Japanese and Japanese natives, and between Korean learners with L2 Chinese/L3 Japanese and Japanese natives as in Table 42.
Figure 18. Means of RT: 4-word sentences (Korean word order)
0 50 100 150 200 250 300 350 400 450 500
Taiwanese learners Korean learners Native Japanese
ms
Table 41
Result of RT with ANOVA: 4-word sentences (Korean word order)
df SS MS F
p
Between Groups 2 1254079.791 627039.895 9.787 .000**
Within Groups 373 23897852.677 64069.310
Total 375 25151932.468
Note.* p<.05. **p<.01
With reference to the results of RTs of complex sentences, the results are consistent with the 3- to 4-word sentences. Means of RTs of the 5-word sentences
Table 42
Tukey HSD analysis of RT: 4-word sentences (Korean word order)
(I) group (J) group
Mean Difference (I-J)
Std.
Error Sig. 95% Confidence Interval Lower Bound Upper Bound
being 473.1ms (Taiwanese learners), 421.4ms (Korean learners), and 355.8ms (Japanese natives) respectively (see Figure 19), ANOVA showed that there is a significant difference among the three groups shown in Table 43 (F(2,346) = 7.936, p
< .05 ). The results of a post hoc in Table 44 revealed that significant difference exists between Taiwanese learners with L2 Korean and native Japanese as a control group, and between Korean learners with L2 Chinese and native Japanese as a control group.
Figure 19. Means of RT: 5-word sentences (Korean word order)
0 50 100 150 200 250 300 350 400 450 500
Taiwanese learners Korean learners Native Japanese
ms
Table 43
Result of RT with ANOVA: 5-word sentences (Korean word order)
df SS MS F
p
Between Groups 2 1627317.088 813658.544 14.693 .000**
Within Groups 372 20600637.621 55378.058
Total 374 22227954.709
Note.* p<.05. **p<.01
Finally, RTs of the 6-word sentences among Taiwanese learners, Korean learners, and native Japanese were found to be significantly different (F(2,322) =
Table 44
Tukey HSD analysis of RT: 5-word sentences (Korean word order)
(I) group (J) group
Mean Difference (I-J)
Std.
Error Sig. 95% Confidence Interval Lower Bound Upper Bound
12.879, p < .05), with means of RT being 488.4ms, 459.3ms, and 353.3ms respectively (see Figure20). Descriptive statistics also illustrated that RTs of the three groups were significantly different as shown in Table 45. A post hoc tests as in Table 46 exhibited that the significant difference was found between Taiwanese learners with L2 advanced Korean and native Japanese, and between Korean learners who have reached advanced level of Chinese and native Japanese.
Figure 20. Means of RT: 6-word sentences (Korean word order)
0 100 200 300 400 500 600
Taiwanese learners Korean learners Native Japanese
ms
Table 45
Result of RT with ANOVA: 6-word sentences (Korean word order)
df SS MS F
p
Between Groups 2 1439470.329 719735.164 12.879 .000**
Within Groups 322 17995030.274 55885.187
Total 324 19434500.603
Note.* p<.05. **p<.01
In summary, a one-way ANOVA showed that there was a significant difference between Taiwanese learners, Korean learners, and Japanese natives in all
Table 46
Tukey HSD analysis of RT: 6-word sentences (Korean word order)
(I) group (J) group
Mean Difference (I-J)
Std.
Error Sig. 95% Confidence Interval Lower Bound Upper Bound
the four types of sentence complexity. A post hoc analysis identified that there was a significant difference between Taiwanese learners with advanced L2 Korean and native speakers of Japanese, and between Korean learners with advanced L2 Chinese and native speakers of Japanese in all the four types of sentence complexity. The results of RT of GJT are summarized in Table 47 and Table 48 (p value and Tukey HSD respectively).
Table 47
Summary of RT of GJT (ANOVA for Korean word order)
Number of words p value
3 words
p = .000**
4 words
p = .000**
5 words
p = .000**
6 words
p = .000**
Note.* p<.05. **p<.01
Table 48
Summary of RT of GJT (Tukey HSD analysis for Korean word order)
Tukey HSD Taiwanese-
Japanese control group
Korean-
Japanese control group
3 words
p =.000** p =.000**
4 words
p =.000** p =.007**
5 words
p =.000** p =.037*
6 words
p =.000** p =.017*
Note.* p<.05. **p<.01
4.2.3.3 RT of GJT (Japanese sentences in the Chinese word order)
This section presents (1) mean RTs of Taiwanese learners, Korean learners, and native Japanese in Japanese sentences in the Chinese word order, starting from three through six words, and (2) descriptive analysis of the results in Japanese sentences in the Chinese word order.
Figure 21 illustrates means of RTs in the 3-word sentences. While Taiwanese learners’ mean RT is 567.95ms and Korean leaners’ mean RT is 464.94ms, native Japanese mean RT is 363.97ms.
Figure 21. Means of RT: 3-word sentences (Chinese word order)
In terms of descriptive analysis, Table 49 showed that significant difference was detected among Taiwanese learners, Korean learners, and Japanese natives (F(2,329) = 20.767, p < .05). Furthermore, a post hoc analysis, as presented in Table 50, has identified that there was a significant differences between Taiwanese learners and Japanese natives, and between Korean learners and Japanese natives. In addition, the post hoc analysis also detected the significant difference between Taiwanese learners and Korean learners.
0 100 200 300 400 500 600
Taiwanese learners Korean learners Native Japanese
ms
Table 49
Result of RT with ANOVA: 3-word sentences (Chinese word order)
df SS MS F
p
Between Groups 2 2548323.628 1274161.814 20.767 .000**
Within Groups 329 20185910.264 61355.350
Total 331 22734233.892
Note.* p<.05. **p<.01
Table 50
Tukey HSD analysis of RT: 3-word sentences (Chinese word order)
(I) group (J) group
Mean Difference (I-J)
Std.
Error Sig. 95% Confidence Interval Lower Bound Upper Bound Taiwanese Korean 103.005* 35.701 .012 18.95 187.06
Japanese 203.975* 31.684 .000 129.38 278.57 Korean Taiwanese -103.005* 35.701 .012 -187.06 -18.95
Japanese 100.971* 34.249 .010 20.34 181.60 Japanese Taiwanese -203.975* 31.684 .000 -278.57 -129.38
Korean -100.971* 34.249 .010 -181.60 -20.34
Note. p < .05
The 4-word sentences exhibited the identical results. Their mean RTs are 530.45ms for Taiwanese learners and 452.17ms for Korean learners while native Japanese RT is 343.54ms as shown in Figure 22.
Figure 22. Means of RT: 4-word sentences (Chinese word order)
A one-way ANOVA showed that there was a significant difference in means of RTs among the three group shown in Table 51 (F(2,332) = 19.992, p < .05). A post hoc analysis showed that the means RTs between Taiwanese learners and native Japanese, and between Korean learners and native Japanese were significantly different as Table 52 presented. In addition, the result of RTs with the 4-word sentences found that RTs between the two experimental groups are slightly significantly different.
0 100 200 300 400 500 600
Taiwanese learners Korean learners Native Japanese
ms
Table 51
Result of RT with ANOVA: 4-word sentences (Chinese word order)
df SS MS F Sig.
Between Groups 2 2241152.711 1120576.355 19.992 .000**
Within Groups 332 18609299.182 56052.106
Total 334 20850451.893
Note.* p<.05. **p<.01
Regarding the results of the 5-word sentences, it exhibited the same patterns.
But a post hoc analysis did not detect a significant difference between the two
Table 52
Tukey HSD analysis of RT: 4-word sentences (Chinese word order)
(I) group (J) group
Mean Difference (I-J)
Std.
Error Sig. 95% Confidence Interval Lower Bound Upper Bound
experimental groups (i.e., Taiwanese learners and Korean learners). The means of RTs in the 5-word sentences are as follows: 484.97ms for Taiwanese learners, 420.40ms for Korean learners, and 373.22ms for native speakers of Japanese as Figure 23 shows.
Figure 23. Means of RT: 5-word sentences (Chinese word order)
Table 53 showed that there is a significant difference among the three groups (F(2,327) = 5.979, p < .005). A post hoc analysis also revealed that RTs between Taiwanese learners and Japanese natives. However, no significant difference was unexpectedly found between Korean learners and Japanese natives (see Table 54).
0 100 200 300 400 500 600
Taiwanese learners Korean learners Native Japanese
ms
Table 53
Result of RT with ANOVA: 5-word sentences (Chinese word order)
df SS MS F
p
Between Groups 2 737506.557 368753.279 5.979 .003**
Within Groups 327 20168445.567 61677.204
Total 329 20905952.124
Note.* p<.05. **p<.01
Table 54
Tukey HSD analysis of RT: 5-word sentences (Chinese word order)
(I) group (J) group
Mean Difference (I-J)
Std.
Error Sig. 95% Confidence Interval Lower Bound Upper Bound Taiwanese Korean 64.570 35.235 .161 -18.39 147.53
Japanese 111.749* 32.335 .002* 35.62 187.88 Korean Taiwanese -64.570 35.235 .161 -147.53 18.39
Japanese 47.179 33.836 .345 -32.48 126.84 Japanese Taiwanese -111.749* 32.335 .002* -187.88 -35.62
Korean -47.179 33.836 .345 -126.84 32.48
Note. p < .05
Finally, the 6-word sentences were shown below. The mean RTs of the 6-word sentences, as shown in Figure 24, were 484.84ms for Taiwanese learners, 491.81ms for Korean learners, and 368.37ms for native Japanese.
Figure 24. Means of RT: 6-word sentences (Chinese word order)
ANOVA has confirmed that there is a significant difference among the three groups (F(2,353) = 10.099, p < .05) as in Table 55. To locate the significant differences, a post hoc was implemented. It showed that RTs between Taiwanese learners and native Japanese, and between Korean learners and native Japanese were significantly different in Table 56.
0 100 200 300 400 500 600
Taiwanese learners Korean learners Native Japanese
ms
Table 55
Result of RT with ANOVA: 6-word sentences (Chinese word order)
df SS MS F
p
Between Groups 2 1219095.221 609547.611 10.099 .000**
Within Groups 353 21307114.765 60360.098
Total 355 22526209.986
Note.* p<.05. **p<.01
In summary, in exploring the significant difference between the means of RTs of L1 Taiwanese with L2 Korean and those of native Japanese as well as L1 Korean
Table 56
Tukey HSD analysis of RT: 6-word sentences (Chinese word order)
(I) group (J) group
Mean Difference (I-J)
Std.
Error Sig. 95% Confidence Interval Lower Bound Upper Bound
with L2 Chinese and those of native Japanese, ANOVA and a post hoc analysis revealed that RTs of native Japanese were significantly faster than those of Taiwanese learners with L2 advanced Korean and Korean learners with L2 advanced Chinese across all the four types of sentence complexity despite a minor exception. However, one difference in the statistical results between RT of GJT presenting Japanese sentences in the Korean word order (4.2.3.1) and those in the Chinese word order (4.2.3.2) is that, in 4.2.3.2, RTs of the two experimental groups were significantly different while 4.2.3.1 did not show such a significant difference. The statistical outcomes were summed up in Table 57 and Table 58 (p value and Tukey HSD respectively).
Table 57
Summary of RT of GJT (ANOVA for Chinese word order)
Number of words p value
3 words
p = .000**
4 words
p = .000**
5 words
p = .003**
6 words
p = .000**
Note.* p<.05. **p<.01
Table 58
Summary of RT of GJT (Tukey HSD analysis for Chinese word order)
Tukey HSD