Child first-language (L1) learners frequently use the definite article in referential indefinite contexts, that is, with nouns appearing in the discourse for the first time, where adults use the indefinite article. Adult second-language (L2) learners also overgeneralize the definite article. Research reported here shows 30 L2 learners use the in referential indefinite contexts at significantly higher rates than in nonreferential contexts. Thus, both L1 and L2 learners may share an initial hypothesis associating the with referential nouns. This evidence of a strategy common to L1 and L2 learners invites reinterpretation of both L1 and L2 acquisition data.

No abstract

No abstract

Several recent accounts of crosslinguistic variation in the properties of anaphors have rejected Manzini and Wexler's (1987) parameterized binding principles. Pica (1987), Battistella (1989), Cole et al. (1990) and Katada (1991), among others, claim that anaphors move in Logical Form (LF) such that 'long-distance' binding can be reduced to a series of local relationships. This article looks at research on adult second language learning in the light of the proposal that reflexives move in LF. A first approach to the issue reanalyses data from earlier research on the acquisition of the Japanese long-distance anaphor zibun, research conducted under the assumption that the binding principles are parameterized. Secondly, a new study of 58 adult learners of Japanese is presented, showing that learners' knowledge of zibun at a high-proficiency level is largely consistent with a key prediction of the movement in LF approach. Although relatively few high-proficiency learners in the subject pool seem to have arrived at the full native-speaker grammar of zibun, there is little evidence that the grammars they construct violate principles of Universal Grammar. On the other hand, data from lower-proficiency learners are less readily accounted for from the perspective of movement in LF.

Ninety-six second language learners of English responded to a 30-item multiple-choice questionnaire requiring them to identify the antecedent of a reflexive pronoun. Their judgments differ from those of a native-speaker control group in that they do not require that a reflexive take a clause mate antecedent, but both groups share a preference for subject over non-subject antecedents. The second language learners do not seem to transfer first language (L1) grammar into the second language (L2), nor do they recapitulate the course of L1 acquisition. An extension of Wexler and Manzini's (1987) parameter-setting model of L1 acquisition to L2 data may account for some of these results, but the high incidence of long-distance binding of reflexives remains problematic: these second language learners have set the governing category parameter too widely without positive evidence.