Acquiring Grammatical Aspect via Lexical Aspect: the Continuity Hypothesis

Previous research in child language acquisition identifies discrepancies between child and adult use of verbal inflectional morphology In this paper, we consider a case of partial undergeneralization where children's use of the-ed and-ing endings seems constrained by the aspectual class of the host verb. Children, more than adults, seem reluctant to generalize the imperfective morpheme to state verbs. Furthermore, they vastly undergeneralize the use of the-ed morpheme with atelic verbs. Many researchers observe similar generalizations: that children seem unwilling to mark activity verbs like walk or unbounded punctiliars like jump with an-ed ending, even though this is a tense marker in the adult language that applies to all types of events. They attribute the undergeneralization variously to lack of tense, confusion of tense with aspect, and skewed distribution in the adult input. These theories fail to cover the asymmetries in the CHILDES data we examine (MacWhinney, 1991). In a study of conversational tiers of three CHILDES file sets (Bloom, Brown, and Suppes), representing eight children, ages 1:4 to 5:1, MLUs 1.055 to 6.007, we find that children use the morphemes to mark restricted grammatical aspect in a way consistent with restrictions on grammatical aspect in other languages. This restriction is most parsimoniously accounted for by a rigorous application of the aspect/tense model in Olsen (1997) consistent with the continuity hypothesis and the subset principle (Berwick, 1985). Two important theoretical points emerge from this discussion: The first is that, in this model, aspectual features, rather than the verb classes they describe (STATE, ACTIVITY, ACCOMPLISHMENT, ACHIEVEMENT, SEMELFACTIVE), provide the appropriate level of granularity for describing the verb-morphology patterns. Speci f-ically, [+dynamic] licenses participation in the childs aspectual morphology and the corresponding imperfective-ing; this morpheme is therefore best analyzed as a progressive in the child language, parallel to similar event restrictions on the im-perfective in a variety of languages (Olsen, 1997). The feature [+telic] licenses the resultative perfective morphology realized in English as-ed, which is parallel to telicity restrictions on the perfective in other languages, e.g. Korean (Lee, 1995).