| | |
Summary: Copyright 2008 Psychonomic Society, Inc. 1064
Spoken word recognition requires that the perceptual
system cope with a noisy signal, sequential inputs that
persist for only a fleeting moment, and temporary ambi-
guity as words unfold. Two particularly challenging as-
pects of this requirement are that phonemic and lexical
contrasts are rarely instantiated along a single dimension
(cue) and information from disparate cues often arrives
asynchronously. Therefore, at any point in time, the word
recognition system rarely has all of the relevant informa-
tion for a phonetic distinction.
Integration of asynchronous cues has not been a focus of
prior theories of spoken word recognition. One approach
would be for the system to store early cues in a temporary
buffer until the remaining cues have arrived. This would
minimize premature (incorrect) commitments but would
force the system to delay making even a partial decision.
Alternatively, each cue could provide partial evidence for
higher level units (e.g., features, phonemes, or words) as
soon as it arrived.This would speed up recognition if early
|
