| |
Abstract:
Phonological theory shows that the phonemes of a language are
not primitives , but are composed of subphonemic units known as
features. The phonemes /b/, /d/, and /g/ differ in place of
articulation but share the feature [+voice], which distinguishes
them from the phonemes /p/, /t/, /k/, which are all [-voice].
Existing research has shown that the generator of the Magnetic
Mismatch Field (MMF) in human auditory cortex accesses
representations of both the discontinuous acoustic sensitivity
which underlies speech perception (Näätänen et al.
1997), and phonemic categories themselves (Phillips et al. 1998).
The current study investigates whether the generator of the MMF can
also access subphonemic feature representations, specifically the
feature [voice]. Evoked magnetic fields were recorded using a
64-channel whole head biomagnetometer while subjects passively
listened to sequences of standard and oddball stimuli. [+voice]
stimuli were chosen randomly from a set of 9 synthetic syllables,
consisting of /ba/, /da/, and /ga/ each with 3 different voice
onset times (VOT); [-voice] stimuli were chosen randomly from a
corresponding set of examples of /pa/, /ta/ and /ka/. There was no
many-to-one ratio among stimuli at either the acoustic or phonemic
level, but there was at the level of phonological features.
Comparison of responses to standard and deviant stimuli revealed a
MMF component which could only be accounted for if the generator of
the MMF has access to phonological feature representations.
|
