According to Chomsky (1998, 1999, 2001) UG makes available a set of functional features, from which different languages select subsets for their feature inventories. Features can accordingly be classified as either universal features (i.e., selected by all languages) or parameterized features (i.e., selected by some languages but not by others). This paper compares the L2 acquisition of the two types of feature and suggests that this theoretical distinction is also reflected in the acquisition data. It reports on the results of a number of experimental tasks carried out by very advanced L2 Spanish speakers of various L1s and indicates that parameterized features of the L2 not present in the L1 are problematic for adult learners, even at very advanced levels. In these studies the two functional features present in the subjects' L1s - structural Case and uninterpretable number - appear to be fully acquirable. By contrast, only the performance of those subjects whose L1 has uninterpretable gender features is native-like. The asymmetries found are compatible with the view that L1 functional feature specification can be a good predictor of L2 attainment.
Proceedings of the 6th Generative Approaches to Second Language Acquisition Conference (GASLA 2002): L2 Links
edited by Juana M. Liceras, Helmut Zobl, and Helen Goodluck
Table of contents