Exploring the conceptual origin of a linguistic behavior: Event endstate representation in infancy

By 6 months of age, infants are sensitive to goals and results of actions (e.g., Woodward, 1989), and words denoting results (e.g., 'up', 'allgone') are among the earliest acquired. However, there is a puzzling paradox with older children: preschoolers allow noncompletion readings of change-of-state verbs--e.g., interpreting 'fill' as if it meant 'pour' rather than something being made full (e.g., Wittek, 2002). In many domains there is a homology between the learner's conceptual system and linguistic system. Therefore, in this study, we explore the conceptual origin of preschoolers' linguistic behavior. We ask: do infants conceptualize event endstate as a critical event component, such that a change in endstate leads to a change in the event category? For example, do they perceive an event in which a particular endstate comes about (e.g., paper falls to cover a spoon) as belonging to a different event category from events in which the same event occurs, but to a different degree (e.g., paper falls and partially occludes, but does not fully cover, a spoon)? This study differs from most earlier work investigating infants' understanding of goals and changes of state because it focuses on events that (a) are naturally occurring physical events with no human agent, and (b) come to different degrees of completion, rather than completely different endpoints. In a habituation-switch study, 14-month-olds were habituated to a change-of-state event--half viewed an event with full occlusion, and the other half viewed an event with partial occlusion. When infants reached a preset habituation criterion, the test phase began. At test, those who viewed the full event during habituation were shown the partial one (Full-then-Partial Condition), and those who viewed the partial one before were shown the full one (Partial-then-Full Condition). We expect infants to be more surprised at the change in the Full-then-Partial Condition than in the Partial-then-Full Condition, if they conceptualize endstate as a critical event component. Specifically, we hypothesized that in the Partial-then-Full Condition, upon seeing the partial event, infants would categorize it as an "occluding" event, and be unsurprised to see another occluding event at test, even though the degree of occlusion was different. In the Full-then-Partial Condition, however, we hypothesized that infants would categorize it as a "covering" event, and would be surprised to see the test event, which is not a good exemplar of a "covering" event. See Figure 1. Results (n = 18, in progress) are consistent with our prediction: infants looked longer at the test event in the Full-then-Partial Condition but not in the Partial-then-Full Condition; see Figure 2. We take this result to suggest that the difference between the full and partial occlusion events is not merely perceptual, but a conceptual one; in other words, infants categorize events on the basis of the degree to which they achieve a particular endstate. We do not, therefore, find evidence that preschoolers' linguistic behavior (e.g., interpreting 'fill' as 'pour') might be explained by event conceptualization. We interpret the results in light of hypothesized differences between linguistic and conceptual encoding.