Predicting second language (L2) identification rates from first language (L1) mapping data: Similarity patterns for English and Korean obstruents in pre- and poststressed intervocalic, and postvocalic positions.