The fourth research question aims to investigate task effects on the second language acquisition of Japanese kanji characters. The overall comparison between the PM task and the TM task and a general discussion of the major findings are as follows. First, in the previous research, it was found that the participants performed better on the kanji-context mix text (Mori, 2003). The participants in the previous research also showed that they felt more confident when doing the task with kanji characters and the texts. Wesche and Paribakht (2009) also found that it was useful to infer the meaning of unknown words from the textual cues and that the utilization of lexical inferencing might result in incident vocabulary learning1. However, not much research explored the effect of the pictorial cues in the acquisition of L2. The pictorial cues can be a factor helping L2 learners interpret
kanji characters. To understand the participants’ interpretation of Japanese kanji characters
in two different ways, two comprehension tasks (the PM task and TM task) were employed in the present study, as shown in Table 4-20.Table 4-20 Examples of the Test Items in the PM Task and the TM Task
Task PM Task TM Task
Example
______とは、個人的に行う裁縫、
刺繍、編み物などの創作活動を いいます。
Answer 愛情 ‘love’ (Part 2, Q10) 手芸 ‘handcraft’ (Part 2, Q8)
As shown in Table 4-21, the result of one-way ANOVA indicated that there was a significant difference in interpreting kanji characters in the PM task and in the TM task (F (1, 92) = 5.889b, p < .05). It was showed that most of our participants performed better on the
1 The incident vocabulary learning means that L2 learners do not learn vocabulary by rote learning. They learn the vocabulary by inferring the meaning correctly from the texts.
TM task (M = 0.84) than the PM task (M = 0.81).
Table 4-21 Participants’ Overall Performances on the PM Task and the TM Task
Task M SD F
p-value
PM 0.81 0.10 5.889b .017*
TM 0.84 0.15
As can be seen in Table 4-22, the participants performed significantly better on the TM task than on the PM task in the I group (F (1, 46) = 4.405, p < .05), Group A (F (1, 46) = 4.467, p < .05) and the native baseline group (F (1, 46) = 20.125, p < .001). As for Group B (F (1, 46) = 2.508, p > .05), no significant result was found.
Table 4-22 The p-values for the within-group Differences on the PM Task and the TM Task
B I A N
F
2.508 4.405 4.467 20.125P
.120 .041* .040* .000***In addition, the mean scores of our participants’ responses to the PM task and the TM task for each group are presented in Figure 4-4.
Figure 4-4 Overall Performances of Each Group on the PM Task and the TM Task
B I A N
PM-Mean 0.73 0.78 0.83 0.92
PM-SD 0.09 0.06 0.08 0.05
TM-Mean 0.67 0.83 0.87 0.98
TM-SD 0.16 0.11 0.06 0.04
0 0.2 0.4 0.6 0.8 1 1.2
As can be seen in Figure 4-4, Group B performed better on the PM task than on the TM task (B: M = 0.73 > 0.67) while the other three groups (i.e., Groups I and A, and the native baseline group) performed better on the TM task than on the PM task (I: M = 0.83 > 0.78, A:
M = 0.87 > 0.83, native baseline group: M = 0.98 > 0.92).
According to these results, it was found that the participants’ performance was rather satisfactory that the mean scores of all proficiency groups were above 0.7. Their performance on the TM task was better than on the PM task although Group B performed on the PM task better than on the TM task. This implies that for our higher-level participants, texts indeed affected their judgment, supporting the utility of the textual cues as stated in Wesche and Paribakht (2009). However, the performance of Group B also showed that the pictorial cues could influence the participants’ interpretation of kanji characters. From the perspective of L1 acquisition of vocabulary, children’s performance on meaning inferencing was better when the pictorial cues were available to them (Holmes, 1987; Bialystok, 1981). This in turn can be a possible explanation for our result that Group B in the present study behaved more like children that they need more pictorial input. As Gahungu (2010) and Haththotuwa Gamage (2003) suggested, learners would have certain learner strategies, such as using the context and grouping kanji characters with other kanji characters which have the same pronunciation. Creating associations between pictures and
kanji is a strategy as well. Although Haththotuwa Gamage (2003) did not discuss the
influence of L2 proficiency, her participants reported that the picture association was one of their learner strategies.Table 4-23 The p-values for the PM Task and the TM Task
PM TM
F
27.855 36.988P
.000*** .000***Table 4-24 The Results of Scheffé Post Hoc Analysis (Comparison within Two Tasks)
Std. Error Sig. 95% Confidence Interval Lower Bound Upper Bound
*. The mean difference is significant at the 0.05 level.
As for the between-group comparison shown in Table 4-23 and 4-24, the participants’
performance on the two tasks was compared. In response to the PM task, Group A performed significantly better than Group B (p < .01). The native baseline group significantly outperformed Group A (p < .01) and Groups I and B (p < .001). With regard to the TM task, Group I significantly performed better than Group B (p < .001) while Group A also performed significantly better than Group B (p < .001). The native baseline group
significantly outperformed Groups I and B (p < .001) and Group A (p < .01).
Generally speaking, L2 proficiency was found influential. We could see that the higher-level group’s performance was usually significantly better than the basic group’s performance on the PM task and the TM task. As found in Kondo‐Brown (2006), lower-proficiency learners could not successfully utilize the textual clues because of their low vocabulary size and lower-level learners might not have adequate breadth and depth of vocabulary (Wesche & Paribakht, 2009). Therefore, they struggled between the L1 Chinese characters and L2 kanji characters. Even though they could infer the meanings from the texts in the TM task, they might frequently encounter unknown kanji characters in the given texts and hence they could not get efficient help from textual cues.
Moreover, the interaction between forms and tasks (as shown in Figure 4-5) and between meanings and tasks (as shown in Figure 4-6) were observed.
Figure 4-5 Participants’ Performances on the Three Types of Forms of Japanese Kanji
Characters in the PM Task and in the TM Task
From Figure 4-5, we could see that the mean scores for the same form and the different form in the TM task were relatively higher than those in the PM task, whereas the
same form semi-same form different form
PM Mean 0.8 0.89 0.68
semi-same form in the PM task scored higher than those in the TM task, showing that picture coding is more useful on the semi-same form and textual coding is more useful on the same form and the different form.
The results of same forms and the different forms in the TM task could be explained in that forms might be more related to the formal syntactic aspect (Moris, 2003). The participants could select the correct item based on the textual cues, such as the particle markers. Thus, the textual environments provided supports and hence the participants’
performance of the TM task was enhanced. Nevertheless, their performance on the semi-same form showed a reversed pattern. When we examined the underlying mean scores of each group, only the native group performed better on the semi-same form in the TM task.
All the experimental groups performed better on the semi-same form in the PM task, indicating that our participants indentified the semi-same form more successfully by using pictorial cues. A possible explanation might be that these semi-same cases made them confused. Historically, it has been known that Japanese speakers borrowed Chinese characters and simplified those Chinese characters as kanji in order to develop Japanese writing system and make kanji characters more acceptable to the public (Kess & Miyamoto, 1999; Liu, 2009). However, the simplified forms of Chinese characters were reformed by Chinese speakers as well as Japanese speakers. On the one hand, Japanese speakers borrowed simplified Chinese characters. On the other hand, they simplified the selectively chosen characters by themselves (Liu, 2009). Thus, there are two kinds of simplified Japanese kanji characters and L2 learners might be confused with forms of kanji characters.
Instead of textual cues, the pictorial cues could provide our participants with immediate responses which help them infer better.
While forms were more dominant in the TM task, it was found that meanings became a more salient feature for the participants in the PM task.
Figure 4-6 Participants’ Performances on the Two Types of Meanings of Japanese Kanji
Characters in the PM Task and in the TM Task
The results presented in Figure 4-6 showed that the mean scores for each type of meaning in the PM task were higher than those in the TM task, indicating that it was more efficient to infer meanings of Japanese kanji characters with pictorial cues.
In the previous studies, Nassaji (2003) and Shahrokni (2009) argued that the text might not always be useful because the information from the texts could be unclear. Thus, the guessing from the textual cues may not be always easy and there is a possibility that they will make wrong choices. Shahrokni (2009) also found that their pictorial group outperformed the textual group. Therefore, our L2 participant’s performance seemed to be supported, that is, pictures provided more straightforward information. The participants could immediately infer the meaning and make more meaning-related choices. Although both the textual cues and the pictorial cues could help our L2 learners to infer the meanings of kanji characters, the pictorial cues gave them more meaning supports.
same meaning different meaning
PM Mean 0.97 0.93
PM SD 0.06 0.21
TM Mean 0.94 0.76
TM SD 0.1 0.22
0 0.2 0.4 0.6 0.8 1 1.2