Recent researches (Tomasello, 2003) have proposed than children's syntactic knowledge starts with simple constructions and grows progressively towards more complex and abstract constructions. We propose that subject dislocations (e.g. la fille elle dort) could be a marker of such development in children with normal language development and in children with SLI. This development would go through a four-step process. [1] First, children would produced words (X) that correspond to bare forms (e.g. a word with no morphosyntactic markings) such as ‘dort'. [2] Second, children would produce constructions with one open slot (aX) or two open slots (aX bY) that are each made of a word and a highly frequent morphosyntactic marker which is produced automatically, without integrating its meaning and functional value in the utterance. Morphosyntactic markers will not necessary agree in gender or number (e.g. la fille i dort). [3] Third, children would produce the same type of constructions as before, but they will take into account the meaning and functional value of the markers and morphosyntactic markers will agree in gender or number (e.g. la fille elle dort). [4] Four, children would produce complex constructions with two open slots (aXY) without dislocations (e.g. la fille dort).
Our prediction is that children can produce together forms that are close in the developmental process (for example forms [1] and [2], or [2] and [3]), but not forms that are very different (for example forms [1] and [3], or [2] and [4]). This hypothesis will be tested with children with SLI and with control children in a repetition task.