| Summary: | 博士 === 國立中正大學 === 心理學研究所 === 86 === It is well known that dysarthric speech is often more
intelligible to listeners familiar with the speaker than to
those who are not. The underlying mechanisms for this speaker
familiarity effect, however, are not fully understood. In this
study, two experiments were conductedto investigate effects of
speaker familiarity and laboratory training onperception of
dysarthric speech. In the first experiment, a Chinese-
speakingwoman with cerebral palsy produced lists of bisyllabic
words, bisyllabic pseudowords and isolated monosyllables.
Familiar listeners (the speaker''s parents and brothers) and
naive listeners listened to and transcribed theseitems. The
results indicated that familiar listeners yielded
substantiallyhigher intelligibility scores than naive listeners
and that the effect was more pronounced for meaningful words
than isolated syllables and pseudowords.It was shown that
familiarity effect resulted from the interaction betweenspeech
decoding and word knowledge. Error analyses and acoustic
measurementsshowed that familiar listeners could use the
acoustic information available to identify stops and affricates.
In the second experiment, an intensivelaboratory training was
carried out to discover the dynamics of the familiarity effect.
In Exp2A, feedbacks about the target words were or werenot given
to the listeners. In Exp2B, a forced-choice paradigm was
appliedin training task, and confusability among syllable pairs
was manipulatedso that top-down and bottom-up influences could
be differently observedduring the training course. The results
indicated that laboratory trainingwas effective and it can in
fact mimic the development of the familiarityeffect in the
naturalistic environment.
|
|---|
