Processing studies with adults have documented on-line effects of referential context in parsing syntactic ambiguities (e.g. Altmann & Steedman, 1988) and referential indeterminacies (e.g. Sedivy et al., 1999). By contrast, children have shown much less sensitivity to referential factors in resolving syntactic ambiguities (e.g. Trueswell, Sekerina, & Hill, 1999). The current study investigates children's use of discourse factors in resolving referential indeterminacies that do not involve syntactic ambiguity. There is some evidence from the word-learning literature that children interpret adjectival modifiers as discourse contrastive, and use this information as a constraint on word learning (e.g. Prasada & Cummins, 2000). These findings mirror adults' on-line preference to generally interpret prenominal adjectives as marking a contrast between two referents of the same kind (Sedivy et al.).
In the current study, we used a procedure similar to Prasada's involving novel nouns, in which children (ages 3-6) were shown three objects (e.g. a small jar, a big jar, and a big mug) and were given an instruction that either contained a prenominal adjective (e.g. "I'm thinking of a big fep-can you give it to me?") or an adjective in predicate position (e.g. "I'm thinking of a fep. It's big-can you give it to me?"). Children could either respond contrastively, using the adjective to distinguish one object from another of the same kind (the big jar) or noncontrastively (the mug). Unlike Prasada's study, prenominal adjectives did not show more contrastive responses, suggesting that children were not linking modification with a discourse contrastive function. However, in Prasada's study, but not in ours, prenominal adjectives received primary stress, which may facilitate or drive the effect. In addition, we were interested in whether children's ultimate responses reflected initial biases in the interpretation of adjectives (i.e. whether their on-line biases would be similar to those observed for adults). To this end, we conducted an eye-movement experiment, using adjectives in predicate and prenominal position, with prenominal adjectives occurring as unstressed, or receiving primary stress.
A pilot study with adults showed that while stress resulted in somewhat of an increase in contrastive responses, even unstressed prenominal adjectives received significantly more contrastive responses than adjectives in predicate position. Adults' behavioral responses were also reflective of their on-line biases as indicated by eye movements to objects; thus, both the off-line responses and on-line data for adults capture a tendency to interpret prenominal adjectives as signaling a discourse contrast.
We are currently collecting both off-line response data and eye-movement data from young children to determine whether they understand the contrastive function of prenominal modifiers, or whether earlier evidence along these lines was confounded by stress. Results of the study will have implications for the development of discourse constraints in language processing.
References
Altmann & Steedman, (1988). Interaction with context during human sentence processing. Cognition, 30, 192-238.
Prasada & Cummins, (2000). Structural Constraints on the Interpretation of Novel Count Nouns. Paper presented at the 25th Annual Boston University Conference on Language Development.
Sedivy, Tanenhaus, Chambers & Carlson (1999). Achieving Incremental Semantic Interpretation through Contextual Representation. Cognition, 71(2): 109-147.
Trueswell, J. C., Sekerina, I., Hill, N. M., & Logrip, M. L. (2000). The kindergarten-path effect: studying on-line sentence processing in young children. Cognition, 73, 89-134.