Various models of code-switching analysis suggest that closed-class and open-class words are not processed equally, and that the nature of such a distinction plays a central role in code-switching production. Such models, however, make different predictions based on their respective theories of the representation of closed-class words. The Matrix Frame Model (MFM---Myers-Scotton, 1993) assigns a central role to system morphemes (closed class words that do not assign or receive theta-roles). Alternatively, the Functional Head Constraint Model (FHC---Belazi et al., 1994) assigns a central role to functional heads (closed class items that are head of a functional phrase). A minimalist framework (MacSwan, 2000), however, explains code-switching constraints in terms of feature matching and switches below X0.; This study investigates (1) whether the representation of the distinction between functional and lexical elements in the three models of code-switching analysis could account for the data on code-switching constraints; and (2) whether language dominance has any effect on the predictions of the three models.; 52 bilinguals (14 English-dominant females, 14 Spanish-dominant females, 12 English-dominant males, 12 Spanish dominant males) participated in three experiments (a sentence production task, a self-paced reading task, and a judgment verification task) that were carried out in regard to this study.; Experiment I (sentence production task) did not yield a main effect in terms of the production of an obligatory embedded island. This outcome does not support the predictions made by the MFM and the FHC, and was consistent with the predictions of MacSwan's model. However, the self-paced reading task (Experiment II) showed a significant effect in terms of slower response time for embedded system morpheme switches. The results from Experiment I and II suggest that although embedded islands may not be obligatory, there is an additional processing cost for embedded system morpheme switches. Experiment III (judgment verification task) produced a main effect when the embedded system morpheme did not match a focused referent. Such a result suggests that system morphemes are active within an embedded island.
展开▼