This section aims to examine the general patterns our subjects made in the PP task. There were three major types of patterns found other than the target sentences:
1) No elicitation, 2) Non-ka sentences, and 3) Ka sentences. No elicitation refers to the situation where nothing was elicited; the case of non-ka sentences is related to the situation where the subjects responded with the non-ka counterparts; the case of ka sentences is concerned with the situation where ka sentences were uttered, but they were not target sentences or they were ungrammatical. Non-ka sentences can be further divided into five sub-types: a. Key Words, where only a related verb was produced; b. S+V[-telic]+O, where a non-ka sentence with an atelic verb was produced; c. S+V[+telic]+O, where a non-ka sentence with a telic verb was produced;
d. Code Switching, where the Taiwanese ka was replaced with the Mandarin Chinese ba or gei; e. Substitution of ka, where Taiwanese tui or ho was used instead of ka.
Also, Ka sentences can be categorized into four sub-types: a. Without Teh, where the target sentence was a ka sentence with a progressive marker, but only a ka sentence without a progressive marker was elicited; b. Without Omission, where the target sentence was a ka sentence with a syntactically absent ka NP, but only a ka sentence was produced; c. Indefinite Ka-NP, where an indefinite ka-NP was used; d.
Teh+V[+telic], where a progressive marker co-occurred with a telic ka-verb. Table 4.10 presents the three major types and their sub-types. Also, the frequency of each type and each sub-type was calculated.
Table 4.10 Other Patterns Elicited in the PP Task
Key Words Khioh khilaih. (G1S4)
‘Pick up.’
Mama ba pangking chit chingkhi.
(G2S13)
‘The mosquito stung the little sister.’
38
‘Mother is hitting the little sister.’
26 (7.49%)
9 (12.16%)
Without Omission Koko ka (Abin) me. (G2S8)
‘The brother scolded Abin.’
*Aing teh ka cit e pangking chit chingkhi a.
As shown in Table 4.10, a total of 347 unintended responses were found in the experimental group and 74 in the control group. For the 74 unexpected sentences, 43 were classified into Non Ka Sentences and 31 Ka Sentences. Also, the top three
sub types of error patterns in the control group were as follows: S+V [-telic]+O (56.76%), followed by Without Omission (18.92%) and Without Teh (12.16%).
Among the unexpected responses of the experimental group, 21 sentences were in the category of No Elicitation, 231 sentences were of Non-Ka Sentences, and 95 sentences were of Ka Sentences. In addition, three of the most common error types were as follows: S+V [-telic]+O (23.63%), followed by S+V [+telic]+O (14.41%) and Without Omission (13.54%). It was found that S+V [-telic]+O and Without Omission were the types most frequently produced by the experimental group and the control group. Furthermore, the two types of ungrammatical ka sentences, (*Indefinite ka-NPs and *Teh + V [+telic]), were found in both the experimental group and the control group.
To see if the same preferences and tendencies apply to each age group, the frequencies of the error patterns made by each age group are examined. Table 4.11 shows the frequencies of the error types produced by the four age groups.
Table 4.11 Other Patterns of Each Group Elicited in the PP Task
Table 4.11 illustrates that the number of error responses decreased with age (G1: 160;
G2: 107; G3: 80; G4: 74), indicating that our subjects’ production of ka sentences got better as they grew older. In addition, there was a greater decrease from Groups 1 to 2 than from Groups 2 to 3, suggesting that the production of the ka construction improved a lot at the age of 5. Among the error patterns, the number of No Elicitation was used more frequently by Group 1 (G1: 12) than by the other groups (G2: 4; G3: 5; G4: 0). Key words was found highest in Group 1 (26), dropped dramatically in Group 2 (7) and vanished in Groups 3 and 4 (0). S+V [-telic]+O was
used relatively frequently by each group (G1: 44; G2: 15; G3: 23; G4: 42). S+V [+telic]+O was produced frequently by Group 1 (35) and rarely used by the other groups (G2: 7; G3: 8; G4: 1). Code Switching and Substitution of Ka were two patterns commonly used by Group 2 (18; 24) but seldom produced by the other groups. Without Teh were seldom produced by each group (G1: 10; G2: 7; G3: 9; G4:
9). Without Omission came with an increasing trend in the three younger groups (G1: 11; G2: 16; G3: 20) and decreased a little in Group 4 (14). Indefinite ka-NPs (G1: 0; G2: 3; G3: 3; G4: 5) and teh + V [+telic] (G1: 4; G2: 6; G3: 6; G4: 3) were not commonly produced by each group, and between them, teh + V [+telic] was used more often than Indefinite ka-NPs.
With regard to the error type of each group, it was found that the preference for each group was similar. What follows are the top three preferred patterns for each group:
(8) Group 1: S+V [-telic]+O > S+V [+telic]+O > Key Words
Group 2: Substitution of Ka > Code Switching > Without Omission Group 3: S+V [-telic]+O > Without Omission > Without Teh
Group 4: S+V [-telic]+O > Without Omission > Without Teh
As illustrated in (8), S+V [-telic]+O and Without Omission were commonly found in each group. Also, S+V [-telic]+O was the pattern most frequently produced for each group except for Group 2 where Substitution of Ka took the first place.
According to the results presented above, we can see that the ka construction is complicated for our subjects because the number of error responses did not go down till the age of six. Also, the huge number of S+V[-telic]+O and Without Omission supported our previous claim that the use of the Taiwanese ka construction is highly related to the discourse context. When our subjects encountered an atelic verb, they tended to avoid using the ka construction, supporting the analysis that the function of
the ka sentence with a bare verb is to make the verb become the semantic focus (Hung 1995, Tsao 2003). In addition, with regard to Omission, our subjects may think that the ka-NP is an old information mentioned in the previous discourse and thus the NP is understandable even if it is syntactically absent.
There were some differences in using the top three preference types by each group. For example, Group 1, they did not use the ka construction not only when the verb was atelic but also when the verb was telic. Their reluctance to use the ka construction in general may suggest that they were still at a “no contrast” stage, where the distinction between the element exhibited in the construction and that prohibited had not been acquired at all. Hence, they generally avoided producing the ka sentences. Furthermore, the high frequency of Key Words used in Group 1 showed that their general control over their mother tongue was still not good. As for Group 2, it is interesting to see that their preference types were different from the other three groups. They preferred greatly to Substitution of Ka and Code Switching where ka was replaced by Taiwanese tui ‘toward’ or Mandarin ba, indicating that age five might be a transitional point for the acquisition of the ka construction. Group 2 entered a
“partial contrast” stage, where the contrast between permitted and prohibited elements was only partially mastered. Also, this also showed that our subjects treated ka as a syntactic equivalent of Taiwanese tui ‘toward’ or Mandarin ba. The reason might lie in the similar meanings or distributional contexts they have. As pointed out in Chapter Two, except for being a patient marker, ka can also serve as a goal marker.
As a goal marker, ka marks the entity which is the termination of a monetary transaction (Teng 1975). Similarly, tui ‘toward’ in Taiwanese can mark the entity toward which an action is performed. Hence, because of the semantic similarity between ka and tui our subjects tended to use them interchangeably. With regard to the substitution of Mandarin ba for ka, the high similarity of their semantic meaning
and syntactic distribution accounted for our subjects’ performances. Both Taiwanese ka and Mandarin ba are analyzed as a disposal structure and it is generally agreed that when ka occurs with a telic verb, it is a counterpart of Mandarin ba (Teng 1982, Hung 1995, Tsao 2003). Hence, it is not surprising that our subjects substituted Mandarin ba for Taiwanese ka. With regard to the preference types of Group 3, it was found that they were the same as those of Group 4, showing that at the age of six, our subjects had an adult-like grammar and had entered the “full contrast” stage.
In addition, the two types of ungrammatical ka sentences, Indefinite ka-NP and teh + V [+telic], produced by our subjects again suggested that our subjects might consider the indefinite ka-NP in the present study as specific and the thematic role of the ka-NP as benefactive, hence the result.
4.5 Age Effects
Generally speaking, our children performed better with age. The four-year-olds performed worse than the other age groups, showing that they had not yet mastered the construction. At the age of 5, our subjects showed a dramatic improvement, suggesting that age 5 might be a transitional point for the acquisition of the ka construction. Finally, the six-year-olds performed similarly to the control group, exhibiting that children of age 6 had reached the adult-level grammar. In addition, the performances of age 4 seemed to be similar to those of age 5, and the performances of age 5 were similar to those of age 6, showing that there was a gradual improvement from ages 4 to 6.
With regard to the responses to the main constraints, each group was found to perform the best on Dynamicity and the worst on Omission. Besides, each group’s responses showed a trend. What follows are the hierarchies of the main properties found in each group repeated from section 4.1:
(9) Group 1: Dynamicity = Referentiality = Progressive = Omission Group 2: Dynamicity = Referentiality > Progressive = Omission Group 3: Dynamicity > Referentiality = Progressive > Omission Group 4: Dynamicity > Referentiality > Progressive > Omission
In general, Dynamicity was the least challenging for each group, followed by Referentiality, then Progressive and Omission.
As for the performances of the four groups on the sub-properties of the four constraints, the results displayed that statistical differences did exist. The four-year-olds showed a significantly higher preference for Definite/Generic than Indefinite (p<0.05) in response to Referentiality; regarding Omission, they performed better (p<0.05) on 1st/2nd Person than on 3rd Person; in the case of Dynamicity, they preferred Non-stative to Stative (p<0.05); finally, with respect to Progressive, a preference for either Atelic or Telic was not found (p>0.05). In response to Referentiality, the five-year-olds performed significantly better on Definite/Generic than Indefinite (p<0.05); in the case of Omission, 1st/2nd Person was of the same difficulty as 3rd Person (p>0.05) for them; regarding Dynamicity, they scored higher on Non-stative than on Stative (p<0.05); as for Progressive, they showed no significant preference for either Atelic or Telic (p>0.05). In the case of our six-year-olds, their responses to Referentiality showed a significantly higher score on Definite/Generic than on Indefinite (p<0.05); as for Omission, 1st/2nd Person was of the same difficulty as 3rd Person for them (p>0.05); regarding Dynamicity, they showed a higher preference for Non-stative (p<0.05); regarding Progressive, the 6-year-olds performed better on Atelic than on Telic (p<0.05).
In the case of task effects, it was found that generally speaking, each age group performed better on the GJ task than on the PP task except for Referentiality, where an opposite result was obtained.