Chapter 9---9.2.2~9.2.2.3

9.2.2 Testing perception of co-variation

Hypothese 1:
- listeners formulate equivalence categories in which the two sites of a lowered velum, N and V(nasal), are perceptually equivalent

Hypothese 2:
the range of variants of V(nasal) and N that listeners treat as perceptually equivaletn will differ depending on the voicing of the coda consonant


9.2.2.1 Methodological approach

1. co-varying acoustic properties
- trading with each other is taken as evidence of th ecoherence among parts of the acoustic signal that belong together

2. wavefrom-edeiting techniques
(bed, bend, bet, bent)
-three groups of pairs
a. N-only pair: /n/ duration was the only difference between pair members
b. cooperating paris: the stimulus with the shorter /n/ had less vowel nasalization than did the stimulus with the longer /n/
c. conflicting pairs: the stimulus with the shorter /n/ had more vowel nasalization than did the on with hte longer /n/


9.2.2.2 predictions

1. conflicting pairs, despite large acoustic differences between pair members, should be difficult to dscriminate--possibly more difficult than the acoustically less distinct N-only pairs

2. cooperating pairs, whose member have large acoustic differences and alrge differences in total nasalization, should be correctly judged as different

3. the expected influence of coda voicing is that the perceptual judgments of listeners will broadly reflect the distribution of V(nasal)N measures found for th eproduction of VNC(voiced) and VNC(voiceless) words, such that vowerl nasalization will have a greater influence on judgments in the voiceless than in the voiced context.


9.2.2.3 Results

(expected)
1. discrimination was most accurate for coopearating pairs, whose members differd substantially in total nasalization across the V(nasal)N sequence

(unexpected)
2. listeners also showed the expected greater sensitvity to vowel nasalization in the [t] than in athe [d] context

3. listeners who consistently discriminated the conflicting trials more poorly tha n the acoustically less distinfct N-only trials, and listeners whose overall accuracy on conflincting trials was similar to that on cooperating trials.

4. nasa murmurs are more likely to be detected when followed by silence (the voiceless closure) than when followed by glottal pulsing (the voiced closure)

5. Diffierent listeners have diffierent levels of respondence to the simuli