Many recent generativist models attribute grammatical knowledge to young children on the basis that children's language patterns the same way as the target adult language. It has been proposed that the child acquires this knowledge early on in development by a process of parameter setting. Wexler (1996) presents the VERY EARLY PARAMETER SETTING HYPOTHESIS (VEPS), which suggests that basic language-specific parameters are set correctly at the earliest observable stages, that is, from the onset of multi-word speech. However, the patterns predicted by VEPS are necessarily a characteristic feature of the input language and could therefore be just as readily explained in more limited scope terms. A good example of this is French negation where, in order to perform correctly, the child simply has to register that certain lexical forms precede the negative particle pas and certain other forms follow it. One implication of such a limited scope account is that there should be a strong relation between the lexical specificity of children's and mothers' use of verbs before and after pas. We present data for one subject, aged from 2;1 to 2;8, that are compatible with this hypothesis. While these data do not count directly against VEPS, they do suggest that the very low frequency of errors in children's use of French negation does not in and of itself constitute evidence for early syntactic knowledge.