The dissertation: Phonetic Category Learning
The stimuli consist of alveopalatal [ɕ] and retroflex [ʂ] sibilant fricatives paired with the vowel [a]. This distinction was chosen to be the stimulus set for these experiments for several reasons. First, this distinction is difficult for English speakers to discriminate (subsumed generally under English [S]), allowing for meaningful learning to take place. Further, these sounds are common to several languages from different families, such as Polish and Mandarin, which also have a dental/alveolar sibilant fricative. This allows for a comparison of perceptual systems having different sibilant inventories, a two-way contrast (English) versus a three-way contrast (Polish, Mandarin). More generally, a sibilant fricative place contrast was chosen because the perceptual cues for these sounds are relatively well understood and for simplicity in creating continua from natural stimuli.
The modified, naturally produced
stimuli were created by blending productions of [ɕa]
and [ʂa] produced by a native
speaker of Polish. One token of each syllable was selected based on
similarity of vowel quality and similarity to typical measurements of
Polish fricatives as described in Nowak (2006). Each syllable was
split at the fricative/vowel boundary and separate 10 step continuua
were made from [ɕ] to [ʂ]
and between each vowel. These continua were then recombined into CV
syllables in every C+V combination resulting in a 10 x 10 matrix.
The image to the right describes the stimuli. Click on a
circle to hear an individual syllable from the matrix. Click on [ɕa]
and [ʂa] to hear the unmodified
original sounds that were used to create the matrix. Click on
Fricative to hear the fricative continuum and click on Vowel
to hear the vowel continuum.