To the best of our knowledge, no previous studies have empirically investigated the influence of construction effects within different cleft types on L2 learners’ cleft and
pseudo-cleft performances. In addition, the markedness differential hypothesis (Eckman 1977) predicts that a difficulty hierarchy can be found in L2 learners’ language
performance. Such a difficulty hierarchy reflects the marked and unmarked areas for L2 learners. As discussed in previous chapters, the hypothesis is supported by various studies (Liao 2014, Xie 2008). However, these studies have not presented a clear and thorough
picture of the difficulty hierarchy of distinct Mandarin cleft types. Without this difficulty hierarchy, the effect of cleft type differences on L2 learners’ cleft performance cannot be
clearly revealed. On the basis of these concerns, our second research question addressed two main issues: (1) whether cleft type differences influence L2 learners’ preference for certain constructions and (2) whether a consistent difficulty hierarchy can be observed in Japanese-speaking learners’ performances on clefts and pseudo-clefts. To answer these
66
questions, we observed Japanese-speaking learners’ performances on five cleft types, namely NP, VP, PP, ADJP, and ADVP. The following subsections present and discuss our subjects’ performances on these cleft types.
4.2.1 Overall Results
In this section, we first examine our subjects’ overall performances on the five cleft types as can be seen in Table 4-7.
Table 4-7 Subjects’ Overall Performances on the Five Cleft Types
Type Constructions F %
Chi-square p-value
NP Cleft 341 52% .029 .986
Pseudo-Cleft 308 48%
VP Cleft 335 58% 9.429 .009**
Pseudo-Cleft 241 42%
PP Cleft 312 54% .314 .855
Pseudo-Cleft 269 46%
ADJP Cleft 260 53% 1.868 .393
Pseudo-Cleft 231 47%
ADVP Cleft 260 53% 1.833 .400
Pseudo-Cleft 238 47%
The results revealed that only the performance on the VP cleft type reached significance (p = .009 < .05). Such a result was not observed for other cleft types, indicating that cleft type differences did not significantly affect our subjects’ preference for Mandarin clefts and pseudo-clefts.
67
Table 4-8 P-values for the within-Group Differences among the Five Cleft Types
NP LI HI Control
Table 4-8 indicates that no significant differences were present within the L2 group in the cleft types other than the VP cleft type. In the control group, no significant differences were found in any of the cleft types. These results further verify that construction differences were not a factor influencing our subjects’ cleft performances. Different cleft constructions did not pose specific difficulties to any of the groups.
The results revealed the following frequency counts for each cleft type in cleft construction, as observed in the LI group: NP, 103; VP, 106; PP, 100; ADJP, 82; and ADVP, 90. In the HI group, the acquisition order is outlined as follows: NP = PP = VP >
ADVP = ADJP (F = 113, 116, 113, 88, and 93, respectively). In the control group, the observed performance on the five cleft types was NP = VP = PP > ADJP = ADVP. Similar
68
results were found in the subjects’ performances on the pseudo-cleft construction. The
findings reveal that the NP, VP, and PP cleft types were the most highly acceptable cleft types. The ADJP cleft type was accepted only by the control group. Japanese-speaking learners did not accept such a cleft type. The ADVP cleft type was the least acceptable.
Compared with the other cleft types, none of the groups highly accepted ADVP clefts.
A more detailed examination revealed no clear proficiency effects.
Table 4-9 P-values for the Five Cleft Types
Cleft NP VP PP ADJP ADVP
Chi-square
2.135 .794 1.173 .400 1.669p-value
.344 .672 .556 .819 .434Pseudo-Cleft
NP VP PP ADJP ADVP
Chi-square
1.974 21.145 .387 5.922 .437p-value
.373 .000** .824 .052 .804Table 4-9 shows that the subjects’ performances on the five cleft types did not reach significance (p < .05) for the two constructions. For cleft constructions, none of the cleft types reached significance (p = .344, .672, .556, .819, .434 > .05), demonstrating the absence of proficiency effects. For pseudo-cleft constructions, only the VP cleft type revealed a significant difference (p = .000). Other cleft types did not present such a result.
On the basis of the overall performances on the five cleft types within the two constructions, L2 language proficiency was clearly not a factor regarding cleft type differences.
69
In summary, this section demonstrates that cleft type differences did not influence the subjects’ cleft performances. Furthermore, the following difficulty hierarchy was
determined:
(3) NP = VP = PP > ADJP = ADVP
We also determined no clear proficiency effects. The results of the three groups were similar.
4.2.2 General Discussion
To address our second research question, this section discusses whether Japanese-speaking learners’ cleft performances are influenced by cleft type differences. The overall results indicated no significant differences among the subjects’ performances within each
cleft type, demonstrating an absence of clear construction effects.
Regarding the difficulty hierarchy, all three groups highly accepted the NP, PP and VP cleft types, but not the ADVP and ADJP types, within the two constructions.
According to Lee (2005), the NP and PP types are commonly observed in Japanese clefts;
the VP type is also found in Japanese clefts (Ishihara 2012). However, the ADVP and ADJP types are not found in Japanese clefts. This might explain the lower acceptability for the ADVP and ADJP clefts compared with the other cleft types. Hence, for JSLs, the
70
NP, PP and VP clefts are the unmarked cleft types, whereas the ADJP and ADVP are the marked types. The organized difficulty hierarchy is presented as follows:
(4) NP = VP = PP > ADJP = ADVP
This finding confirms that L2 learners did not acquire all the cleft types at the same time.
Some cleft types such as the NP, VP, and PP clefts were easy to acquire, whereas other cleft types such as the ADJP and ADVP clefts were difficult. This further indicates that the NP, VP, and PP cleft types belong to the unmarked cleft types, whereas the ADJP and ADVP are considered to be marked types. These findings support the claim of the Markedness Differential Hypothesis (Eckman 1977) that a difficulty hierarchy can be found in learners’ performances, signifying the markedness of different cleft types. Hence,
this hypothesis is validated in the current study and the cleft type effects were attested.
In the difficulty hierarchy presented by Liao (2014), the NP and PP clefts were the easiest cleft types. The ADVP and ADJP cleft types were difficult to acquire.
Nevertheless, Xie (2008) reported a different difficulty hierarchy, in which the NP and ADVP clefts were the easiest for L2 learners. These were followed by the VP clefts, and the ADJP clefts were the most difficult to acquire. The difficulty hierarchies of Liao (2014) and Xie (2008) are presented as follows.
(5) a. Liao (2014): NP = PP > ADVP > ADJP b. Xie (2008): NP = ADVP > VP > ADJP
71
Our difficulty hierarchy is similar to that of Liao (2014). If the VP cleft type is not considered, our difficulty hierarchy also indicates that the NP and PP clefts were acquired first and that the ADJP cleft type was the most difficult to acquire. This demonstrates that the difficulty hierarchy for cleft sentences is similar between Mandarin-speaking learners of English clefts and Japanese-speaking learners of Mandarin clefts. Thus, we consider that the difficulty hierarchy of clefts are universal. More study can be conducted in the future to show more support for such claim.
Moreover, a detailed examination of the performances of the three groups indicated that the subjects did not perform differently within each cleft type, revealing the lack of proficiency effects.
In summary, this section shows that construction effects are not obvious and cleft type differences hindered the subjects’ acquisition of cleft sentences. Moreover, the
difficulty hierarchy provided by the subjects cogently supports the markedness differential hypothesis. Our Japanese-speaking learners’ language proficiency did not influence their cleft performances.
4.3 Contextual Effects
Contextual effects have been a major concern in previous cleft research (Declerck 1984, Li 2008, Liao 2014, Mai & Yuan 2016, Prince 1978). As mentioned in Chapters
72
One and Two, Liao (2014) and Mai and Yuan (2016) have both discussed issues regarding contextual effects. However, they did not clearly investigate whether contextual effects affected L2 learners’ language performances. Whether contextual effects influence L2 learners’ cleft acquisition remains unknown. To investigate this question, we examined
whether contextual factors influence Japanese-speaking learners’ cleft performances. The results and relevant discussions are elaborated in the following subsections.
4.3.1 Overall Results
We initially investigated our subjects’ overall performances on the two types of
contexts. Table 4-10 shows the overall scores of the two constructions within the two contexts.
Table 4-10 Subjects’ Overall Performances on the Two Types of Contexts
Type Constructions F %
Chi-square p-value
Discourse Opener Cleft 19 11% 17.771 .001**
Pseudo-Cleft 161 89%
Thematic Continuity Cleft 94 52% 2.182 .336
Pseudo-Cleft 86 48%
As demonstrated in Table 4-10, in the discourse opener context, the performances on clefts and pseudo-clefts differed significantly (p = .001 < .05), implying that the subjects strongly associated the function of discourse openers with pseudo-cleft sentences. In the thematic continuity context, the performances on clefts and pseudo-clefts were not
73
significant (p = .336 < .05). Nevertheless, our subjects tended to use more clefts than pseudo-clefts in these contexts. This finding reveals that the thematic continuity context was more difficult for our subjects to acquire than was the discourse opener context because they did not strongly associate this context with cleft sentences. These results indicate that in the discourse opener context, pseudo-cleft sentences may be easily elicited, whereas in the thematic continuity context, cleft sentences may be triggered. Thus, the different contexts influenced the subjects’ preferences, revealing that their cleft
performances were affected by contextual factors.
Table 4-11 P-values for the within Group Differences in the Two Contexts
Discourse Opener LI HI Control
Chi-square
17.067 41.667 51.066p-value
.000*** .000*** .000***Thematic Continuity LI HI Control
Chi-square
2.400 .067 .067p-value
.121 .796 .796By observing the preferences for the two constructions in the two contexts within each group, we found that in the discourse opener context, the subjects had a strong preference for one construction, but such a result was not found in the thematic continuity context.
We further observed the subjects’ preferences for cleft sentences in the two contexts.
74
Table 4-12 Overall Cleft and Pseudo-Cleft Performances in the Two Contexts
Type F %
Chi-square p-value
Table 4-12 shows that in the thematic continuity context, all groups preferred to use cleft sentences. All three groups used pseudo-clefts in the discourse opener context. According to this finding, the subjects strongly considered that pseudo-clefts have a discourse opener function, and they considered that clefts may be used to continue a topic from previous contexts. Regarding proficiency effects, the results demonstrated that only the p-values for cleft sentences in the discourse opener context reached significance (p < .05). In the discourse opener context, the LI group performed differently from the other two groups (p = .030, .004), but no differences were observed between the HI and control groups (p
= .463). In the thematic continuity context, comparisons among all groups showed no significant differences (p = .391, .775 .737). The overall findings indicate that the LI group did not perform differently from the other two groups. The results of the HI group
75
are similar to those of the control group. Therefore, proficiency effects were not observed.
In summary, this section demonstrates that the subjects’ performances were
affected by contextual effects. Moreover, the subjects preferred pseudo-clefts in the discourse opener context, but they preferred clefts in the thematic continuity context.
Furthermore, no clear proficiency effects were observed.
4.3.2 General Discussion
As mentioned in Chapters One and Two, contextual effects have been widely discussed in previous studies (Declerck 1978, Liao 2014, Mai & Yuan 2016 Prince 1978).
Mai and Yuan (2016) demonstrated that contextual effects were crucial factors influencing cleft acquisition. In the current study, the discourse opener and thematic continuity contexts were examined to address this issue. As described in the results, in the discourse opener context, the subjects preferred pseudo-clefts, whereas in the thematic continuity context, they might select cleft sentences. Accordingly, context clearly affected the subjects’ preferences for certain cleft sentences. Our results thus confirm the presence
of contextual effects. As Mai and Yuan (2016) reported, contextual effects are crucial factors influencing L2 learners’ cleft acquisition.
Prince (1978) noted that clefts and pseudo-clefts have different discourse functions.
Pseudo-clefts can function as discourse openers and clefts can be used to continue the topic from the previous contexts. According to the current study, the functions claimed in
76
Prince (1978) can be found in Mandarin clefts. Our findings prove that Mandarin pseudo-clefts function as discourse openers. Moreover, we demonstrate that Mandarin pseudo-clefts can be used in the thematic continuity context. Such findings strongly accord with the claims of Prince (1978).
Regarding proficiency effects, the performances of the LI group did not differ substantially from those of the HI and control groups, signifying that the learners’
performances were not affected by their language proficiency. Hence, no clear proficiency effects were observed. The results also reveal that the discourse opener context was easy for the Japanese-speaking learners to acquire, possibly because discourse openers are a universal feature that can be found in nearly all languages. Universal features are regarded as general senses and have a wide distribution. If a feature is universal (or unmarked), L2 learners can acquire at an early stage (Greenberg 1991). Thus, the discourse opener context is assumed to be easy for L2 learners to acquire. The thematic continuity context requires strong connections with discourse cohesion. Scholars have considered discourse cohesion such as backgrounding and subordinating information to be important in assisting learners with the pragmatic usages of clefts. To realize the pragmatic functions of clefts in the thematic continuity context, L2 learners must analyze the information given from previous contexts, accurately interpret each of the given sentences, and then choose the correct one. This procedure requires some time for Japanese-speaking learners
77
to process. Hence, the thematic continuity context is more difficult than the discourse opener context is.
To conclude, this section reveals clear contextual factors. Mandarin pseudo-clefts were determined to function as discourse openers, and clefts had the feature of thematic continuity, as stated by Prince (1978). However, no clear proficiency effects were found.
The performances of the LI and HI groups did not differ significantly, demonstrating that L2 language proficiency did not strongly influence Japanese-speaking learners’
performances on clefts.
4.4 Proficiency Effects
In this section, proficiency effects are examined in terms of Japanese-speaking learners’ acquisition of Mandarin clefts and pseudo-clefts. According to the results and
discussions of the previous sections, the developmental stages of Mandarin clefts and pseudo-clefts can be illustrated as follows:
78
Stage 1: Low-Intermediate Level
1. Having acquired ungrammatical clefts and pseudo-clefts and grammatical clefts.
2. Having acquired the NP, PP, ADJP, and ADVP cleft types, but not the VP cleft type.
3. Understanding that pseudo-clefts can be used in the discourse opener context, but not in thematic continuity context.
4. Understanding that clefts can be used in the thematic continuity context.
Stage 2: High-Intermediate Level
1. Having acquired ungrammatical clefts and pseudo-clefts, as well as grammatical clefts.
2. Having acquired grammatical pseudo-clefts.
3. Having acquired the NP, PP, ADJP, and ADVP cleft types, but not the VP cleft type.
4. Understanding that pseudo-clefts can be used in the discourse opener context, but cannot be used in the thematic continuity context.
5. Understanding that clefts can be used in the thematic continuity context.
6. Understanding that clefts cannot be used in the discourse opener context.
Figure 4-3 Developmental Stages of Mandarin Clefts and Pseudo-Clefts
Initially, when Japanese learners’ Mandarin proficiency was low, the acquisition of
Mandarin clefts and pseudo-clefts was poorer than that of the group with higher Mandarin language proficiency. At this stage, the subjects could identify ungrammatical clefts and pseudo-clefts, in addition to grammatical clefts. Nevertheless, they had difficulty interpreting grammatical pseudo-clefts. Furthermore, the L2 learners at this stage acquired most of the cleft types, namely the NP, PP, ADJP, and ADVP types, except for the VP cleft type. According to the difficulty hierarchy presented in the previous section, the NP and PP cleft types were the easiest and the ADJP and ADVP cleft types were more
79
difficult. Finally, the Japanese learners at this stage understood that clefts could be used in the thematic continuity context and that pseudo-clefts could be used in the discourse opener context but not in the thematic continuity context. However, they did not understand that clefts could not be used in the discourse opener context.
At the second stage, when Japanese learners reached the HI proficiency level, they performed substantially higher than the learners at the first stage. They exhibited native-like performance levels on the interpretation of grammatical and ungrammatical Mandarin clefts and pseudo-clefts and on the pragmatic usages of Mandarin clefts and pseudo-clefts. However, they had not fully acquired the VP cleft type.
As mentioned in Chapter One, the L2 proficiency effects were supported by some researchers
(Carrell 1991, Chan 2004, Liao 2014, Mai &Yuan 2016, Xie 2008,)
. According to the findings of the present study, the HI learners outperformed the LI learners in terms of grammaticality effects and contextual effects. These results reveal that the performances of the HI group differed from those of the LI group. Thus, in the current study, the L2 proficiency effects could be confirmed.4.5 Summary
In this chapter, the overall findings and discussions are presented to address the research questions of the current study. The results reveal that the four effects, namely
80
grammaticality effects, construction effects, cleft type effects, and proficiency effects, did not strongly influence our Japanese-speaking learners’ interpretation of Mandarin clefts and pseudo-clefts, whereas the contextual effects did influence their cleft performances.
In the next chapter, the major findings of the present study are presented. The limitations of the current study, some pedagogical implications, and suggestions for future research are also provided.
81
C HAPTER F IVE
C ONCLUSION
In this chapter, the conclusion of the current study are provided. Section 5.1 summarizes the major findings of this study and Section 5.2 presents some pedagogical implications. Finally, Section 5.3 provides the study limitations and offers some suggestions for future research.