The role of frequency and distributional regularity in the acquisition of word order

Studies of spontaneous speech (e.g. Pine, Lieven, & Rowland, 1998) have highlighted how children's use of grammatical markers, such as the use of word order to mark agent-patient relations, is often restricted to specific lexical items, rather than being generalized as much as one might expect. This has led to the suggestion that grammatical knowledge is generalised from lexically specific constructions. A major implication of this constructivist view is that grammatical development will be affected by the frequency of individual lexical (and larger syntactic) items and the probabilistic relations between these items. In particular a greater degree of distributional regularity may facilitate the acquisition of a given lexical item. Intuitively this appeals to the idea that the more a child hears something and the more it occurs in the same place (absolutely or relative to some other marker) the more likely it is that s/he will be able to use it. The following two experiments set out to investigate these ideas. In the first experiment we tested the effect of verb frequency on the use of word order as a grammatical marker (this study is reported in detail in Matthews, Lieven, Theakston, & Tomasello, in press). In the second experiment we looked into the role of distributional regularity on the use of word order. More specifically we considered the role of case-marked pronouns, in structuring early grammatical knowledge by comparing English with French, a language where pronoun (clitic) and lexical objects do not share the same distribution relative to the verb. The experiments employed the WWO methodology, developed by Nameera Akhtar (1999) to test English-speaking children's productive control of word order. In Akhtar's experiment children were taught novel verbs (and known nouns) in a novel, ungrammatical word order (e.g. Subject Object Verb, as in " Ernie the spoon dacking "). The idea was to test whether, the children would themselves talk about the actions with the non-canonical word orders when asked, " What's happening? " or whether they would prefer to use the novel verb in the canonical order of their language. Akhtar found that children aged 4;4 preferred to use canonical SVO word order with the novel verbs. Whereas, children aged 2;8 and 3:6 were just as likely to adopt the SOV and VSO word orders with novel verbs as they were to switch to using them in SVO order. Importantly, however, the same children who had used the …