Pronouns as elsewhere elements: Implications for language acquisition

Date of Completion

January 2005


Language, Linguistics




This dissertation investigates the acquisition of A- and A'-bound pronouns in Brazilian Portuguese and English. Previous studies on the acquisition of pronouns have found that children behave at chance level when pronouns with local antecedents in A and A' positions are tested. ^ The hypothesis under investigation here is that children performed poorly in tests with locally A- and A'-bound pronouns because the source of the problem is the same. There are several reasons to pursue a unifying approach. First, both constructions involve pronouns. Second, results of studies in several languages indicate that children perform similarly on both tests, incorrectly accepting these cases at chance level. And third, the age-range when this chance performance is detected is the same in both cases, that is, around 4 and 5 years of age. ^ Following Hornstein (2001), I assume that (A- and A'-) bound pronouns are elsewhere elements, that can only be inserted in a derivation if needed for convergence. Adopting Grodzinsky and Reinhart's (1993) hypothesis, I claim that such a condition is too demanding for young children, as their limited working memory cannot handle complex computations, such as those required in order to assess if bound pronouns are licit in a derivation. ^ Using the grammaticality judgment task, the same Brazilian Portuguese- and English-speaking children were interviewed on two experiments, one involving A-bound pronouns and another involving A'-bound pronouns. The data revealed that the majority of the children performed at chance level (50% correct responses) when A'-bound pronouns placed in extractable positions were tested (* the frog that he is skating is happy ). Children also behaved at chance level when pronouns locally A-bound by both referential and quantified antecedents were tested (*the dog i/every dogi is scratching him i). Importantly, children did not behave at chance in control cases where the kind of computation mentioned above is not necessary. In these cases, children behaved at ceiling. These results support the hypothesis under investigation here, indicating that children's problems with pronouns is related to processing problems rather than to the lack of some linguistic knowledge. ^