This chapter introduced the backgrounds of the participants, the methods, and materials used in the two tasks (i.e., the PP task for production and the GJ task for comprehension). In addition, the procedures (a pilot study, formal study, and scoring policy) were described.
CHAPTER FOUR
RESULTS AND DISCUSSION
In this chapter, the results of the experiment will be presented, and answers to the research questions proposed in Chapter One will be discussed. Section 4.1 examines the hierarchical difficulty of the four constraints of the ka construction, and section 4.2 further discusses the difficulty of each sub-property. Section 4.3 compares the subjects’ performances on the two tasks, and other patterns other than the target sentences will be examined in section 4.4. In section 4.5, age effects will be studied.
Finally, a summary of the results is given in section 4.6.
4.1 Property Effects
This section addresses the first research question, that is, the effects of the four constraints on L1 acquisition of the Taiwanese ka construction: the Referentiality constraint, the Omission constraint, the Dynamicity constraint, and the Progressive constraint. The results of the four constraints will be presented first, followed by a discussion about the subjects’ performances on the four constraints. Table 4.1 shows the general results of the experimental group and the control group.
Table 4.1 Subjects’ Correct Responses on the Four Constraints of the Taiwanese ka Construction (Average Means)
Constrain Group
Referentiality Omission Dynamicity Progressive M SD M SD M SD M SD
Experimental 0.57 0.20 0.42 0.15 0.64 0.24 0.51 0.20
Control 0.72 0.07 0.53 0.16 0.95 0.08 0.64 0.09
As illustrated in Table 4.1, the mean score of Omission was the lowest for both the experimental group (M=0.42) and the control group (M=0.53). Regarding the other three properties, both groups performed the best on Dynamicity (Experimental:
M=0.64; Control: M=0.95), followed by Referentiality (Experimental: M=0.57;
Control: M=0.72), and Progressive (Experimental: M=0.51; Control: M=0.64).
Namely, the performances of the two groups on the four constraints pattern alike and the degree of difficulty of the four constraints for both groups can be ranked from easy to difficult as Dynamicity > Referentiality > Progressive > Omission (p<0.05).
Furthermore, an independent samples test was executed to see if there was a significant difference in the performances of the two groups on the four constraints.
The results displayed that the performances of the adult group were significantly better than the experimental group on the four constraints (p<0.05), showing that there exists a gap in the acquisition of the Taiwanese ka construction. The above results correspond to the previous studies in that the average scores of our experimental group were lower than those of our control group, indicating that the ka construction is similar to the ba construction in that it has not been fully mastered at the initial stage of children’s language development (Erbaugh 1982, Cheung 1992, Fahn 1993).
To see if the same developmental tendency can be seen in each age group’s acquisition of the ka construction, one-way ANOVA and a paired samples test were applied with different age groups as an independent variable.
Figure 4.1 Each Age Group’s Correct Responses on the Properties of the ka Construction (Average Means)
A general developmental tendency was displayed in Figure 4.1. The performances of our subjects increased with age. Furthermore, Group 3 (age 6) had gained the adult-like ability (p>0.05). By contrast, the performances of the two younger groups were observed to be significantly worse (p<0.05) than those of the adult group, except for Omission, where no difference was found (p>0.05) (Group 1 vs. Group 4:
Referentiality: p=0.000, Dynamicity: p=0.000, Progressive: p=0.001; Group 2 vs.
Group 4: Dynamicity: p=0.000, Progressive: p=0.021). Group 2 (age 5) also performed like the control group on Referentiality (p>0.05). In addition, a comparison among the three younger age groups showed that a significant difference existed, except for Omission, where no difference was found. The result showed that Group 1 (age 4) attained significantly lower scores (p<0.05) than Group 3.
However, the performances of Groups 1 and 2, and Groups 2 and 3 patterned similarly (p>0.05), showing that our subjects performed better gradually from ages 4 to 6. The result may also show that Group 2 was a transitional point for the acquisition of the ka construction.
The finding that our children gained an adult-like ability at the age of six corresponds to the previous findings on the children’s acquisition of the Mandarin ba construction. In examining the children’s acquisition of the ba construction, Cheung (1992) found that the six-year-olds produced more ba sentences than the five-year-olds; Fahn (1993) also found that the Compound Verb and the Definiteness constraints were not acquired until the age of six. Moreover, Fahn’s (1993) findings also showed that age five seemed to be a transitional point for the acquisition of the ba construction since the five-year-olds had mastered the Verb Selection, Modifier and Progressive constraints but not the Compound Verb and the Definiteness constraints in her study.
The results of the paired samples test illustrated that Omission was the most difficult property for all groups, and Dynamicity the easiest one. The hierarchies of the four constraints for each group were summarized as follows:
(1) Group 1: Dynamicity = Referentiality = Progressive = Omission Group 2: Dynamicity = Referentiality > Progressive = Omission Group 3: Dynamicity > Referentiality = Progressive > Omission Group 4: Dynamicity > Referentiality > Progressive > Omission
The results as in (1) supported our earlier finings that Dynamicity was the easiest to be acquired, followed by Referentiality, Progressive and finally Omission. In examining the children’s acquisition of the Mandarin ba construction, Fahn (1993) found that the Verb Selection (=Dynamicity), Modifier, and Progressive constraint were acquired before the Definiteness (=Referentialtiy), and Compound Verb constraints. The result of the present study confirmed Fahn’s study in that our subjects also scored higher on Dynamicity than on Referentiality. However, contrary to Fahn’s findings, the Progressive constraint was not easier than the Definiteness constraint (=Referentiality) for our subjects.
The earlier acquisition of Dynamicity than Referentiality, and Progressive can be explained by the theory of Conservativism (Pinker 1989) and its implication of Rapid Learning among language learners. The theory of Conservativism predicts that children have to acquire first the contrast distinguishing the verbs that are permitted in the ka construction from that are prohibited and then they can apply the contrast to the ka pattern. In other words, the children’s acquisition of the contrast must precede their acquisition of the constraint.
Regarding Dynamicity and Referentiality, the relatively late acquisition of Referentiality than Dynamicity can be attributed to the relative difficulty of the contrast among generic, definite and indefinite markers compared to the contrast between non-stative and stative verbs. Also, the degree of difficulty of the contrasts may be explained by the early acquisition of verbs than referring expressions.
Studies in Chinese (Tardif 1996, Tardif et al. 1996) and Korean (Gopnik and Choi 1990, 1995, Choi and Gopnik 1995) have shown that verbs may be more frequently used than nouns in the early language acquisition. Furthermore, based on findings for English (Bates et al. 1988, 1994) and Italian (Bates et al. 1979), Caselli et al.
(2001) propose a four-stage model of lexical development and suggest that verbs are mastered prior to grammatical function words. The early acquisition of verbs before grammatical function words indicates that children may acquire the contrast between non-stative and stative verbs earlier than the contrast among definite, indefinite, and generic markers. According to the theory of Conservativism, the acquisition of a certain contrast must precede to the application of the contrast to the other syntactic structures. Since the contrast between non-stative and stative verbs may be mastered before the contrast among definite, indefinite, and generic markers, it is highly possible that the contrast between verb meanings is applied to the ka construction earlier than the contrast among referring expressions. Therefore, Dynamicity is
expected to be mastered prior to Referentiality.
The same argumentation may be applied to the later acquisition of Progressive than Dynamicity. Again, for children to acquire the constraints, they have to acquire the contrasts between non-stative and stative verbs, and telic and atelic verbs. The finding that Dynamicity was easier to be mastered than Progressive may be attributed to the late acquisition of the contrast between telic and atelic verbs than between non-stative and stative verbs. The degree of difficulty of the contrast may be explained by the Semantic Feature Hypothesis (SFH) (Clark 1971). Assuming Clark’s (1971) Semantic Feature Hypothesis (SFH), and adopting Vendler’s (1967) classification of verb types, we may hypothesize that verbs in Taiwanese can be classified as follow:
(2) Feature Representations of Verbs in Taiwanese (adopted from Vendler 1967)
According to Clark (1971), superordinate features in feature hierarchies are filled in first in children’s representation of word meanings, followed by subordinate features.
In our case, we may hypothesize that children had to acquire the features of [+/-telic]
for them to distinguish telic verbs from atelic verbs. In addition, the distinction of
[+/-telic] might be subsumed under the feature of [+dynamic]. Hence, it is predicted that children would acquire the distinction of stative and non-stative verbs before they acquire that of telic and atelic verbs. The late acquisition of the distinction between telic and atelic verbs by our subjects then accounted for the finding that Progressive was acquired later than Dynamicity.
With regard to the result that Referentiality was acquired prior to Progressive, it could be resorted to the other function of ka, that is, ka as a benefactive marker. In the GJ task, our subjects were asked to judge the grammaticality of the item *Papa teh ka ia chit chingkhi ‘Father is wiping the chair clean.’ Semantically speaking, ka in this sentence carries more like a dual function, a benefactive and a patient marker.
Hung (1995) explores the four functions of ka, that is, ka as a patient, goal, source, and benefactive marker. She states that a progressive marker is compatible with ka as a benefactive marker, as shown in (3):
(3) a. Mama teh ka gua se sann.
mother PROG KA I wash clothes ‘Mother is washing clothes for me.’
b. Mama teh ka titi se singkhu.
mother PROG KA the little brother wash body
‘Mother is washing body for the little brother.’ (Hung 1995, pp. 121-122) Gua ‘I’ as in (3)a and titi ‘the little brother’ as in (3)b are analyzed as benefactive.
Also, though a progressive marker teh is added in (3)a and (3)b, the two sentences remain grammatical. Hence, a progressive marker is compatible with ka as a benefactive marker. For the unexpected lower score of Progressive that our subjects obtained in the present study, it is highly possible that our subjects treated ka in the test item *Papa teh ka ia chit chingkhi as a benefactive marker, yielding the present result. To confirm our claim, an item analysis was conducted. It was found that
our subjects did obtain a lower score in judging the grammaticality of the item *Papa teh ka ia chit chingkhi ‘Father is wiping the chair clean.’ They only scored 0.27 in this item while their average score of each item was 0.64. Also, for distinguishing the benefactive ka from the other kas, Hung proposes that the benefactive ka-NP is syntactically optional while the other ka-NPs are arguments of the verb. In (3), gua
‘I’ and titi ‘the little brother’ are not arguments of the verb se ‘wash’ because the object positions have been occupied by sann ‘clothes’ and singkhu ‘body.’ Thus, ka serves as a benefactive marker in (3). However, the result of the present study implies that using the syntactic factor to distinguish the benefactive ka from the other kas may not be correct. In the present study, the subjects were asked to judge the grammaticality of *Papa teh ka ia chit chingkhi ‘Father is wiping the chair clean.’
The ka-NP ia ‘chair’ is an argument of the sentences, and omission of the ka phrase in the sentence will result in ungrammaticality. Nevertheless, the results of the present study showed that our subjects tended to judge this sentence as grammatical, implying that they may consider that the ka-NP ia ‘chair’ benefits from the action of chit
‘wipe,’ and thus a progressive marker may be considered as compatible with the sentence. The result, then, suggests that a semantic factor may play an important role in deciding the thematic role of the ka-NP. Besides, the apparent lower score of our subjects can be explained better if the pragmatic factor is taken into consideration.
As discussed in Chapter Two, Hung (1995) argues that the ka sentence with a bare verb functions differently from the one with a verb and a verbal complement. The later can be treated as a counterpart of the Mandarin ba construction, both of which preposes the verbal object to make it the semantic focus. On the other hand, the ka construction with a bare verb is viewed as a way of preposing the verbal object to make the verb become the focus. The same argument was stated in Tsao (2003), who clearly pointed out that the difference between the ka construction with and
without a verbal complement lies in their discourse functions, that is, the ka construction with a complement emphasizes the affected result of the object that undergoes the action while the ka construction without a complement focuses mainly on the action itself. In examining Progressive, our subjects were supposed to use a progressive marker in the ka construction with a bare verb in response to the elicited question “What did X do to Y?” which emphasizes the affected object but not the action itself. In view of this, they might naturally refuse to use a ka expression. To see if the pragmatic factor plays a role in the present study, we may further divid our grammatical target items into two groups: one consists of items with a bare verb and the other with a verb and a verbal complement, and calculate the average means. It was found that a pragmatic factor did come into play since the average mean for the items with a bare verb was merely 0.24 while that for the items with a verbal and a complement was 0.76, showing that our subjects might perceive the inappropriateness of employing a ka expression under such a context and thus used an non-ka expression instead.
As for Omission, the apparently difficulty our subjects had in acquiring the constraint could be explained by the existence of the task effect which will be discussed in section 4.4.