The Development of Aspect
There is controversy in the field of language acquisition about the developmental time course of the acquisition of tense and aspect. The null hypothesis (argued in Smith, 1980; Weist et. al. 1984 among others) states that from the earliest stages, children use their tense morphology correctly and properly distinguish tense and aspect. The alternative hypothesis (argued in Antinucci and Miller, 1976; Bloom et. al. 1980 among others) states that early use of tense morphology actually marks aspectual distinctions and therefore that children acquire aspect before tense. The ``aspect before tense'' hypothesis comes in at least two forms. The weaker version (cf Antinucci and Miller, 1976) claims that children use past tense forms to mark events that involve a change of state and present tense forms to mark events that do not. The stronger version (cf Bloom et. al. 1980) makes the more specific claim that lexical aspect distinctions are mapped onto phrasal level tense and aspect morphology. Thus, for them, past tense indicates telicity, progressive -ing in English indicates an activity, and simple present tense (identified by 3rd singular -s) indicates a state.
The evidence in this debate revolves primarily around the kinds of phrasal level tense/aspect morphology that verbs of varying lexical aspectual types occur with. For Bloom et.al., the fact that two year olds always use the activity verb ``play'' in the (present) progressive, while they use the achievement verb ``find'' in the past tense, is taken as evidence that the phrasal morphology (in this case, the -ing or the -ed) is a marker of lexical aspect. Similarly counter evidence, from for example Smith, 1980, consists of tokens of atelic verbs occurring with past tense morphology, presumably showing that children (of aproximately age 2;5) are not locked into matching particular phrasal morphology with particular lexical aspects.
One difficulty with the experimental work on this question is that it sidesteps much of the linguistic complexity of aspectual systems. Aspect information comes from two sources: lexical aspect is a property of verbs and their phrases (lexical aspect refers to Vendler-esque categories such as activities, accomplishments, achievements and states); viewpoint aspect (following the terminology of Smith, 1991) consists of the imperfective, perfective (and possibly neutral) marking at the phrasal level. These two forms of aspect interact, most famously in the imperfective paradox. The imperfective paradox notes that telic and atelic VPs have different entaliments when they occur with imperfective veiwpoint aspect. Thus (1), with an atelic VP entails (2), but (3), with a telic VP does not entail (4):
(1) Mary was running.
(2) Mary ran.
(3) John was running to the store.
(4) John ran to the store.
The current study takes up an issue that has been largely ignored in this debate: what is the time course of the acquisition of viewpoint aspect? Does it pattern with lexical aspect on the grounds of semantic relatedness? or with other phrasal level morphology like tense? For example, one concrete prediction of the claim that there is a stage where children use phrasal morphology to mark lexical aspect is that children in this stage will be insensitive to the imperfective paradox. The combination of a telic VP with an imperfective phrasal marker is not a possibility for these children so we would expect them to have difficulty computing the entailment relations necessary for the paradox.
In this study, we tested adult controls using a story completion task. Subjects were given setences that varied with respect to both lexical and viewpoint aspect and asked to treat those sentences as story-beginnings and provide continuations. Statistical analyses have not been completed, but preliminary analyses appear to show that the adult completions vary reliably with the aspectual information given in the test sentence. Adults (not surprisingly) are sensitive to the aspectual entailments posited by standard linguistic accounts. Children aged 4 and 5 were tested with a similar task, as well as with the Frog story narrative task (Berman and Slobin 1994) and a truth value judgment task. The judgment task involved watching a video tape of short scenes and asking the children to evaluate sentences as valid descriptions of the video events. A child might have to evaluate a sentence like (4) above after watching the completion of the event, or after the video had been stopped in the middle of the event.
The children tested were clearly able to perform all three of the tasks. Though statistical analyses have not yet been completed, preliminary results suggest that these children are also able to reliably use both lexical and viewpoint aspect information. For example, one subject rejected the sentence in (5) given a scene in which a boy was still in the process of drawing a circle on the grounds that the circle was not yet complete.
(5) The boy drew a circle.
These results suggest that children by the age of 4 can distinguish phrasal level viewpoint aspect from lexical aspect, though it remains open as to whether these same children firmly distinguish viewpoint aspect from tense (cf Weist et. al. 1984). This finding seems to generally argue against the ``aspect before tense'' hypothesis, at least in the strongest version where children are presumed to use phrasal morphology to mark lexical aspect. Further testing with younger children (in the 2 to 3 year old range) is currently underway to establish when children are first able to make the lexical-viewpoint aspect distinction.