why children omit function morphemes: metric vs. syntactic structure

It is an interesting and noticeable fact that in the early stages of language production, children omit certain elements from their speech. Upon closer inspection, it becomes clear that children regularly omit certain function morphemes, as opposed to content words. Function morphemes (including determiners, pronouns, auxiliary verbs, and inflection morphemes) form a natural class from both a phonological and a syntactic perspective: phonologically speaking, in many languages (including English), function morphemes tend to be unstressed and monosyllabic; syntactically speaking, function morphemes are the heads of functional projections, which appear notoriously late in children's speech (Radford 1988). Thus, we can ask whether children omit functional elements for phonological or syntactic reasons. If children omit function morphemes for reasons related to metrical stress, we should find similarities between unstressed function m rphemes and unstressed lexical syllables, and, depending on our assumptions about metric structure, we might find asymmetries in the omission rates of subject vs. object determiners. 1 On the other hand, if children omit function morphemes for syntactic reasons, according to some accounts (e.g. Hoekstra, Hyams & Becker 1997, Clahsen et al. 1996) we should find correlations between omissions in the subject and the predicate, following the structural contingency known as spec-head agreement. In fact we find both of these patterns, which suggests that both metrical and syntactic processes constrain children's output forms, but in different ways.