As stated in the literature, referential complexity of renjia in Chinese may cause difficulties for children in their acquisition of this pronominal. Researchers have pointed out patterns according to the number of interpretations in the target sentences (cf. Wei, 2001).
One is with only one interpretation and the other is with ambiguous interpretations. Therefore, this section mainly addresses the first research question: Do Chinese children show similar responses to different interpretation patterns in their acquisition of the pronominal renjia?
Two interpretation patterns (single and multiple) were designed to explore whether our children felt easier to comprehend and produce the single-interpretation patterns.
4.1.1 Overall Findings
The mean scores of our subjects’ responses to the single- and multiple- interpretation patterns are shown in Figure 4-1.
59
Figure 4-1 Subjects’ Responses to Different Interpretation Patterns
As can be seen in Figure 4-1, one-way ANOVA indicates that there was a significant difference among six age groups in the two interpretation patterns (single: F (5, 90) = 51.316, p < .001; multiple: F (5, 90 ) = 32.705, p < .001). All the groups performed better on the
single- than the multiple- interpretation patterns (G1: M = 0.48 > 0.25, G2: M = 0.55 > 0.40, G3: M = 0.66 > 0.55, G4: M = 0.76 > 0.64, G5: M = 0.82 > 0.78, Control: M = 0.90 > 0.88).
Table 4-1 further presents the within-group differences between the two interpretation patterns for each group.
Table 4-1 The p-values between Different Interpretation Patterns
G1 G2 G3 G4 G5 Control
F 11.178 19.944 7.252 7.986 1.431 2.141
P .004* .000* .017* .013* .250 .164
As can be seen in Table 4-1, a significant difference between the single- and the multiple- interpretation patterns was found in G1 (F (1, 30) = 11.178, p < .01), G2 (F (1, 30)
= 19.944, p < .001), G3 (F (1, 30) = 7.252, p < .001) and G4 (F (1, 30) = 7.986, p < .001), while G5 (F (1, 30) = 1.431, p > .05) and the control group (F (1, 30) = 2.141, p > .05) did not respond significantly differently to the two interpretation patterns. Thus, the within-group comparisons indicated that the single-interpretation patterns were significantly easier than the
60
multiple-interpretation patterns for the children (G1-G4), indicating that referential ambiguity indeed influenced children’s comprehension and production of the pronominal renjia.
As for a between-group comparison, in the single-interpretation patterns, G3 performed significantly better than G1 (p < .01) and G2 (p < .05), and G1-G2 performed statistically similarly (p > .05). G5 significantly outperformed G1-G3 (p < .001), but they did not yield a significant difference compared with the control group (p > .05). In the multiple-interpretation patterns, G4 significantly outperformed G1-G2 (p < .01), and G5 did significantly better than the other experimental groups, except for G4 (p > .05) and showed no significant difference with the control group, showing that they had acquired the adult-like grammar of this pronominal.
4.1.2 General Discussion
The major finding concerning the pattern effects are addressed as follows. Firstly, our subjects demonstrated a similar tendency that the single-interpretation patterns were easier than the multiple-interpretation patterns, which is consistent with Wei’s (2001) L2 study.
Although no within-group comparison was made between the two interpretations patterns in her study, both English and Japanese groups had more difficulties responding to the multiple- than single-interpretations.
As stated in Goodluck (1991), learnability deals with how grammatical principles, input in rule formation and errors that have been made contribute successful learning of a linguistic system in a limited time span. Based on the terminology deriving from Wexler & Culicover (1980), syntactic complexity will influence the degree of difficulty in acquiring grammatical constructions. In their terminology, different constructions are degree-n learnable. For example, if one construction is degree-0 learnable and the other is degree-1 learnable, the later will be more difficult for learners to acquire. The learnability theory is also proposed with an attempt to explain the speed with which language is acquired and to prevent the
61
learners from making errors (Goodluck, 1991). For instance, with regard to the Subjacency principles, the number of embedded sentences is taken as the measure of syntactic complexity in particular constructions. The ability to deduce whether a grammatical structure is derived by movement will be degree-1 learnable (Wexler & Culicover, 1980). By analogy, if we take the number of interpretations of the pronominal to be the measure of semantic complexity of the target sentences, the single-interpretation patterns would carry a lower degree of learnability than the multiple-interpretation patterns. In the single-interpretation patterns, there was only one intended referent fitting the semantic features of the pronominal in the target sentences, while the two potential referents in the multiple-interpretation patterns increased semantic complexity of the sentences. Therefore, more errors may occur in the multiple-interpretation patterns, since they involve in a higher degree of learnability.
Generally speaking, the correct responses to the pronominal renjia in both interpretation patterns increased with age. There were no significant differences between the 4-year-olds (G1) and the 5-year-olds (G2) in response to both patterns. The 6-year-olds (G3) significantly outperformed the 4-year-olds (G1) and the 5-year-olds (G2) on the single-interpretation patterns, implying that our children’s acquisition of the pronominal renjia with a single interpretation in sentences abruptly rose at age 6 (G3). However, not until age 7 (G4) did the children significantly outperform the 4- and 5-year-olds on the multiple-interpretation patterns, indicating that renjia with a multiple interpretation in sentences was acquired at the period of ages 6-7.
Our younger children’s lower rate of correct responses to the multiple-interpretation patterns might be due to their inability to recognize referential ambiguity. As claimed by Asher (1976), children would settle on the first possible referent they see without comparing messages with all the possible referents in the referential field. Their referential choices reflect cognitive inabilities (Asher, 1976, 1979; Bearison & Levey, 1977), or immature meta-communicative abilities (Robinson, 1981; Singer & Flavell, 1981) in the
62
comprehension of referential communication. Echoing to Jackson & Jacobs (1980), the way children chose a referent for the pronominal was more arbitrary, idiosyncratic and fortuitous.
Moreover, our findings were also similar to Markman’s (1977) results that the children of his study tended to employ passive processing and seldom checked the adequacy of their judgment. In sum, the results of our younger children1 (4- and 5-year-olds) were in accordance with the previous results that the younger children had more difficulty dealing with referential ambiguity.