Abstrakt
The role of canonical word order in language acquisition has been studied extensively. According to Brown (1973), children show implicit awareness of word order of the language they acquire. Research suggested that children speaking languages with flexible word order treat SVO as the default in their comprehension until a relatively late age (Dittmar et al., 2008; Smolík, 2015; Slobin, Bever, 1982). However, it is not clear whether comprehension of word order reflects children's productions. Assuming the usage-based approach to language development, one would expect that children's comprehension reflects their use of language, and thus that limited comprehension of word order should be accompanied by a dominant usage of the default word order.
The present study examined the use of word order in 65 children who participated in a two-wave longitudinal study when they were 2;5 - 2;7 (Time 1) and 4;1 - 4;6 (Time 2). During each lab visit, children engaged in a free play session with mothers and their language was recorded and transcribed. In the current study, we extracted all sentences with transitive verbs and coded them for the relative order of the agent and patient arguments and the verb. We also separately coded whether the argument was realized by a noun or a pronoun or omitted. The coding was performed for children as well as mothers. Overall, we coded 1062 maternal utterances, 344 child utterances in Time 1, and 452 in Time 2.
The results are summarized in Tables 1 and 2. Overall, we found that children use all types of word orders in their conversations with mothers. The canonical SVO word order was not most frequent, showing in 8.72% and 12.38% cases in T1 and T2, respectively, and 6.96% in mothers. It thus does not seem to be the dominant word order in productions.
Table 2 demonstrates that the pattern of word orders depends heavily on whether the argument is a noun or a pronoun. Objects are quite much more common in pre-subject and preverbal position if they are pronouns. In pro-drop utterances, preverbal objects are also more common if realized as pronouns. On the other hand, preverbal nominal objects are far from absent in children as well as mothers. The overall pattern suggests that SVO may be the default word order in the system, but it is definitely not the majority order in what children hear.
Our findings provide interesting theoretical implications. If we observe strong preference for SVO in early comprehension but quite variable word orders in production, mere frequency of the input word orders or usage in children cannot explain the comprehension pattern. However, if we assume that children have a strong tendency to create general rules from partial and noisy input, the pattern can be more easily understood. The results are thus support the view that children tend to form early abstractions in their linguistic representations.