The maintenance and loss of reflexive anaphors in L2 English
This thesis is a generative-based investigation of second language (L2) attrition. L2 attrition research to now has measured the loss of lexical items, morphology, word order, and so on. However, none to my knowledge has examined attrition from within an established theoretical framework such as Chomsky's theory of Government and Binding (GB). In particular, this study considers the loss of reflexive binding in proficient L2 English speakers. Informants are six Japanese university students who spent their junior (third) years abroad in the United States. These six informants consisted of two groups: three who had childhood exposure to English, and three whose first exposure to English in the L2 environment was as adults during their university stay overseas. In order to observe attrition which might occur shortly after exposure to the L2 ceased, an important aspect of this research was to begin data collection as soon as possible after the informants' returns to Japan from studying abroad. Data collections occurred at various intervals for each informant and lasted up to 16 months. Data for this longitudinal study were collected via two tests: a truth value judgment test and a grammaticality judgement test. This study is unique in that it uses generative-based SLA research tools and methods to investigate L2 attrition. Furthermore, the truth value judgment test and the grammaticality judgment test provide results which support the hypothesis that principles of reflexive binding attrite in a manner not inconsistent with UG constraints. The general pattern exhibited by all six test subjects initially shows varying but high levels of knowledge of reflexive binding. Over the course of their data collection periods, the informants' knowledge of reflexive binding in English becomes unstable in the face of zero exposure to the target language. In particular, reflexives in finite subordinate clauses tend to remain bound grammatically to local antecedents to a greater degree than in nonfinite clauses. Reflexive binding in tensed clauses thus appears more resistant to attrition. Age at first exposure to the L2 was also considered as a factor in determining ultimate level of attrition. Evidence was found of a sensitive period up to age eight for the successful acquisition and long-term maintenance of knowledge of the principles of reflexive binding, even upon loss of exposure to the L2.