< Back to previous page


A quantitative and qualitative corpus study of the acquisition of topic constructions in child French

Book - Dissertation

In this PhD, we investigate the interaction between syntax and information structure (IS) in child French. Our case study is the analysis of dislocations, which are argued to indicate the topic of the sentence ("what the sentence is about"): (1) Les chats ils font miaou. 'Cats meow.' (Flavie, corpus TCOF, 6;1.4) (2) Où il est le papillon? 'Where's the butterfly?' (Marie, corpus Lyon, 2;1.13) The main contributions of this PhD to the field are the following: 1. Dislocation appears before age 2, at the very beginning of the production of two-word utterances. Until age 3, children produce more right dislocations (RD) than left dislocations (LD). Through a multinomial regression model, we were able to show that LD becomes more frequent when they start producing less elliptic utterances, which supports the hypothesis that syntactic constructions are acquired from the right edge (Freudenthal et al., 2010). 2. Dislocated constituents with a syntactic function within the sentence appear earlier than those with no syntactic link with the rest of the sentence (Hanging Topics), which appear from age 6. Dislocated subjects appear at age 2, before dislocated objects (at age 2;6). Our corpus data contain significantly more objects in RD than LD and more subjects in LD than in RD: children tend to overgeneralize the canonical SVO word order across constructions (Diessel & Tomasello, 2005). 3. Until age 5, pronouns are dislocated more frequently to the left, and lexical NPs to the right. Our hypothesis is that children have more difficulties to process lexical NPs than pronouns (Arnon, 2010). Since heavy constituents in SVO languages tend to be produced in sentence-final positions (Arnold et al., 2000), French children dislocate lexical NPs to the right rather than to the left. 4. Children produce dislocations to mark the topic, with active, deactivated, or physically present referents. This suggests that, like in adult language (Ashby, 1988), children use this construction to encode the topic and that they have acquired that topics must correspond to somehow accessible referents. 5. Through a logistic regression model, we were able to show that, just as adults, children use significantly more contrastive topics in LD than RD. Moreover, continuing topics are dislocated significantly more frequently in the right periphery. 6. We tested the hypothesis that the adult-like IS of dislocations by young children (age 2) is an effect of the production of item-based constructions with a low level of abstraction and fixed lexical items, as described by the usage-based literature on language development (Bannard & Lieven 2009). We found that early dislocations can be described in terms of concrete functions such as the expression of the location or possessor of an entity, which are associated with a specific syntactic constructional schema, such as Là X 'There X' or à X ça 'to X that'. This finding prompts two new (mutually exclusive) hypotheses: (i) the concept of topic is not present at the onset of language acquisition, but develops from more concrete schemas; or (ii) the concept of topic is present at the onset of language acquisition, but is first preferably realized through concrete schemas and then spreads to more general aboutness-topic contexts.
Publication year:2020