The Trouble with /h/: Evidence from ERPs

Recent work on second language (L2) phonological acquisition has argued that the relative ease and difficulty that adult L2 learners encounter with particular novel segments can be attributed to the underlying inventory of phonological features required to represent contrasts in the learner’s first language (L1): segments which present difficulty in acquisition are precisely those whose representations require features that are not supplied by the L1 (Brown, 1997, 2000). Brown proposes that this inability to construct target-like representations for the new segments results in an inability to perceive contrast. In Brown’s view, the difficulty that native Japanese speakers encounter with the novel English /l/ vs. /r/ contrast (Yamada, 1995) is due to the absence of the feature [coronal] from underspecified inputs in the Japanese grammar; [coronal] is required for the representation of /l/ in English (Brown, 1997, 2000). Native Chinese speakers, on the other hand, show improvement with increasing proficiency on this novel contrast as the Chinese grammar does supply all the necessary features for the representation of /l/: crucially, the Chinese phonemic inventory contains sub-coronal contrasts, thereby implicating the presence of [coronal] in the grammar (Brown, 1997, 2000). Under this approach, new segments whose representations require a subset of available L1 features are predicted to be acquirable. This seems to be at odds, however, with the observation that native French speakers encounter a great deal of difficulty with the phoneme /h/ in acquiring English: evidence from production shows that it may be either omitted or inserted, as in (1) (Janda & Auger, 1992). The presence of both types of errors for any given speaker suggests that a perceptual problem is at fault; we will return to this in section 2.1 below.