This paper expands upon the complementary-systems model of speech perception by proposing that exemplar encoding is filtered by native language phonolgy through differential attentional weighting of particular acoustic cues. Phonetic processing in a second language is presumed to rely on an L2 phonological system that is not native-like due to the persistent effects of this attentional filtering. In learning to either re-weight attention or work around this filtering effect, individuals may vary with respect to their ability to exploit the exemplar store in L2 processing, leading to differences in long-term ability to develop native-like L2 phonologies.