In this paper, we investigate how L2-learners whose L1s lack articles acquire the semantics of articles, with special focus on article use by L1-Korean learners of L2-English. Based on our experimental studies, we argue that article choice in L2-acquisition reflects systematic access to universal semantic features: definiteness, specificity, and partitivity. It is shown that L2-learners associate the with [+specific] and [+partitive] features, which results in non-random error patterns: overuse of the with specific indefinites and partitive indefinites. Our findings provide further evidence that L2 learners' errors are traceable to parameter settings that are not necessarily instantiated in their L1 or L2 – but which are available through UG. The paper also presents evidence that the interlanguage grammar of L2-learners has important parallels to that of child L1-learners in the domain of article semantics, supporting our conclusion that L2-acquisition, like L1-acquisition, is guided by UG.
The fact that primary linguistic data under-determine unconscious knowledge of language has been considered strong evidence for postulating Universal Grammar (UG) (Chomsky, 1965, 1986b). While the argument was originally made for first language (L1) acquisition, it also holds for second language (L2) acquisition, with some caveats: if L2-learners' linguistic knowledge is not traceable to either transfer from the L1 or explicit instruction, this provides evidence for the guiding role of UG (for arguments of UG access in L2-acquisition, see Schwartz and Sprouse, 1994, 1996, among others).