In order to sketch the directionality effect on Mandarin complex sentences, we
173
put the data of the three complex sentence types altogether, including both comprehension and production data. The results obtained from a two-way ANOVA (2 Branching Direction x 6 Age Groups) with repeated measures revealed a significant main effect of Branching Direction (F(1, 102) = 45.192, p < .001) and a significant interaction (F(5, 102) = 2.536, p < .05). When we entered the accuracy data from each group into separate independent T-tests, we discovered that the accuracy of the left-branching structure was significantly higher than that of the right-left-branching structure in the 7-year-old and adult groups (7y: t(34) = 2.276, p < .05; adults: t(34) = 4.304 p
< .001). The trend was only numerically suggestive in other age groups (3y: t(34) = 2.017, p > .05; 4y: t(34) = 1.463, p > .05; 5y: t(34) = .829, p > .05; 6y: t(34) = 1.087, p > .05), as shown in Figure 5.1. This finding is compatible with the fact that Mandarin is an essentially left-branching language.
Figure 5.1 The Mean Scores and Standard Deviations of Branching Directions by Age Group
Concerning the directionality effect in each complex sentence type, we reviewed the results from previous chapters and compiled Figures 2.1, 3.1 and 4.1 in Figure 5.2, showing that left-branching structures were found to be significantly
174
preferred in complex sentences with RCs and complex sentences with ACs, while right-branching structures were observed to be significantly predominant in complex sentences with CCs. The branching direction asymmetry discovered in each clause type was also reported in each age group although for some age groups the difference was only numerically suggestive. Additionally, the asymmetry was more obvious in the AC construction when children reached 6 years old than in the CC and RC constructions by any age groups.
BD of CCs BD of RCs
BD of ACs
Figure 5.2 The Mean Scores and Standard Deviations of CCs, RCs and ACs in each Branching Direction by Age Group
With respect to the developments in acquiring CCs, RCs and ACs of the same branching position, the reported results in Sections 2.4.1, 3.4.1 and 4.4.1 suggested that almost all 6- and 7-year-olds, that is, the school-aged children, could give significantly more correct responses than children aged from 3 to 5 years old, that is,
175
preschoolers. In addition, except for left-branching CCs, which seemed to be acquired by children reached 6 years old, the acquisition of all complex sentence types with either left- or right-branching clauses was still in progress.
5.1.2 Discussion
The results relevant to the directionality effect partially support the notion that the complex left-branching structure is preferred in a left-branching language, and vice versa (Bever, 1970; Dryer 1992; Goodluck, 1991) in that RCs and ACs taking branching positions are preferred in Mandarin, generally considered a left-branching language, yet CCs taking right-left-branching positions are favored. This inconsistency with the directionality hypothesis may be attributable to the essence of Mandarin. Mandarin, although an essentially left-branching and head-final language, has the VO sequence as unmarked construction, which allows head-initial constructions, including the objects of verbs, objects of prepositions and a limited number of complements (Huang & Li, 1995).1 In line with the unmarked VO sequence, CCs, which function as arguments of the matrix clause predicates, tend to occur to the right of the head, viz. the verb. On the contrary, according to the essentially left-branching characteristic, RCs, which serve as modifiers of noun phrases, and ACs, which serve as modifiers of an associated main clause or a verb phrase, tend to be left branching. This disharmonic headedness in Mandarin is well captured by the X’-structure of Chinese proposed by Huang (1982). As shown in (1), Huang divided Chinese structure into two forms: a non-nominal category, as in (1a), and a nominal category, as in (1b). At the X’ level, a non-nominal category has its head preceding its complement, e.g., [V’ V CP], which suggests that the unmarked
1 Mandarin is typologically mixed, having both VO and OV sequences (Hawkins, 1990). Nevertheless, the VO sequence is unmarked.
176
position of a CC is right branching. On the other hand, at the X’ level of a nominal category, the head must follow the non-head elements, e.g., [N’ CP N’], which implies that the unmarked position of an RC is left branching. Since RCs and ACs are both adjuncts to their modifying elements, it is also inferable that ACs tend to be left branching.
(1) The X’-structure of Chinese is of the form a. [Xn Xn-1 YP*] iff n = 1, X ! N
b. [Xn YP* Xn-1] otherwise (Huang, 1982: 41 (20)) As to the larger gap between the two branching directions of ACs found in child groups aged 6 or more and the adult group, the ambient language might provide an interpretation. As many researchers have noted (Li, 2008; Lin, 2005; Tsao, 1990;
Wang, 1996, 1999, 2002, 2006), the majority of complex sentences with ACs in Mandarin typically consist of the ACs preceding the main clauses. To be more specific, as reported by Wang (1996), in spoken Mandarin, about 90.9% of temporal ACs, 87% of conditional ACs and 92.9% of concessive ACs were left-branching2, an extreme distribution which is not found in CCs and RCs. This biased preference might be acquired when children get older due to the fact that they have got enough input.
Finally, from the results relevant to the Age effect in each complex sentence type as well as in branching directions of each complex sentence type, we may reasonably conclude that children acquire each complex sentence type in the similar way that school-aged children have acquired significantly better than preschool-aged
2 According to Wang (1996), 73.9% of the causal ACs are located to the right of the associated main clauses in spoken Mandarin.
177
children although they may not perform the complex sentences equally well as adults do.
5.2 The Construction Effects