Rutgers University AND

A recently developed method of head-mounted eye-tracking (Trueswell et al., 1999) has been employed in two experiments to investigate differences between adults and children in on-line processing of ambiguous short-distance pronouns (Tenny, 1999) in English. Sixteen adults and 16 five-to-seven-year-old children viewed a pair of pictures with two characters and listened to sentences describing the pictures which contained either an unambiguous reflexive (himself) or an ambiguous pronoun (him). They had to choose a picture that corresponds to the sentence. For adults, all three types of analyzed data — responses, reaction times, and eye movements — indicate that pronouns are referentially ambiguous (20 % sentence-external referent). Adults were capable of making inferences on-line with respect to less accessible referent, showing a competition between the looks to two potential referents shortly after encountering the pronoun in the sentence. Children’s response data reveal that they overwhelmingly prefer the sentence-internal referent for the ambiguous pronoun (93%). However, their eye movement data are qualitatively similar to adults’, showing their emerging adult-like awareness of the referential ambiguity in sentences with pronouns. But since children’s processing resources are more limited than these of adults (Chien and Wexler, 1990; Avrutin, 1994), children need more time for accessing the sentence-external referent and are reluctant to revise their initial sentence-internal referent commitment in the absence of information that would require such a revision in globally ambiguous sentences. This supports the hypothesis that children’s processing mechanism works like adults ’ (Thornto