The learning and generalization of contrasts consistent or inconsistent with native biases.

This study investigates the process of generalizing a learned sub-lexical contrast across speakers of different non-native accents of English. We examine the generalization of a novel cue (voicing-cued release) that is non-contrastive in English, but contrastive in the manipulated speech of our L2 speakers of English, to a speaker with the same or different L1 as our training speakers (Exp. 1). We then examine performance when the learned contrastive cue is paired with native contrastive vowel duration (Exp. 2), and when the native contrast is present, but the learned contrast is not contrastive in the speech of the new speaker (Exp. 3). We find that learned cues are dominant enough to additively improve word recognition when paired with congruent native cues, although the performance is inhibited when native and learned cues conflict. These data illuminate the intricate balance between native contrasts and recently learned information, where learned information overrides, but is still affected by, native contrasts.