Spelling suggestions: "subject:"acousticarticulatory mismatch"" "subject:"acousticarticulatory dismatch""
1 |
Associations Between Linguapalatal Contact Patterns and Spectral Moments for /s/Bennett, Leslie Leatham 02 March 2009 (has links) (PDF)
Both acoustic and palatographic measures have proven to be useful in speech science research. However, it is not known how closely or consistently these two measures are associated with each other. Therefore, this study investigated the association between changes in tongue-to-palate contact patterns and simultaneous changes in acoustic spectral moments for the fricative /s/. Twenty adults were fitted with pseudopalates and repeated VCV nonsense syllables consisting of an initial schwa followed by the target consonant /s/ and ending with one of three corner vowels (/i, ɑ, u/). EPG (electropalatography) data were quantified using three custom numerical indices (s-narrow, s-wide, and asymmetry) derived from specified zones on the pseudopalate which loosely reflected dimensional differences in the fricative groove. These indices produced general details about changes in tongue contact over time, but index values were not unique to specific contact patterns. The EPG numerical index values were then compared with differences in spectral moments (spectral mean and variance) from the time-aligned acoustic signal. On the whole, all combinations of spectral mean and variance and EPG indices resulted in some weak but significant correlations across all vowel contexts and participant groupings. The majority of these correlations were negative, meaning that as EPG index values increased, spectral mean and variance decreased. Some of the strongest of these correlations were present between s-narrow and spectral mean and variance. Therefore, in order to give a clearer picture of the link between lingual physiology and spectral moments, these variables were correlated for each individual speaker. Stronger significant correlations between s-narrow and both spectral mean and variance were identified in some participants. The majority of these correlations were also negative, suggesting that as the s-narrow index increased, the spectral mean and the variance decreased. A few participants' results that showed interesting lingua-palatal contact patterns are discussed in more detail. Generalization based on specific correlations from this study must be undertaken with considerable caution due to desynchronization of EPG data and the acoustic signal found in several tokens.
|
Page generated in 0.0829 seconds