Perception of Indexical Features in Children’s Speech 3455.6 Results by articulation rate
The data are arranged in Figure 14.6 according to our auditory categorization
by articulation rate. As was the case with voice quality, the number of stimuli
in the non-modal categories is fairly small (n=13 in each case). The emergent
trend in this particular analysis was variable across the listener groups. Both
British groups gave fewer “girl” responses to fast stimuli than they did slow
stimuli. For the Americans the reverse pattern was found.
Following the regression analysis we took an objective measure of articula-
tion rate, in terms of syllables per second for each token (Künzel 1997). The
results of this more detailed analysis failed to clarify the picture, however, with
no signi¿ cant correlation emerging between articulation rate and responses.
5.7 Results for word-medial variants
No variant was returned as a signi¿ cant factor in the regression analysis for
stimuli with medial /p t k/. However, when we compare the listener groups we
do ¿ nd evidence for a signi¿ cant difference in responses.
Figure 14.7 shows the proportion of “girl” responses for medial tokens con-
taining plain and laryngealized variants separately. For both the non-local UK
Figure 14.6 Percentage of “girl” responses by articulation rate.