Avant-garde Dutch 367Figure 15.5 reveals the same state of affairs with respect to the (normal-
ized) magnitude of the spectral change in the diphthongs. The women gener-
ally have a larger differ ence between onset and offset of the diphthongs than
the men, t(15) = 2.93 (p = 0.005, one-tail).
Figures 15.4 and 15.5 together indicate that the phonetics of the sound
change in progress are best characterized as a combined lowering and spec-
tral enhancement of the low-mid di ph thong: the onset changes from low-mid
to fully low but the offset remains more or less stationary, such that a larger
spectral distance has to be covered between onset to offset. The correlat ion
between onset lowering and strength of diphthongization is signi¿ cant for the
female speakers, r = 0.48 (p = 0.030, one-tail) but not for the males, r = 0.17
(ins.). This ¿ nding strengthens the claim that the sound change in progress is
predominantly found with female speakers.
3.5 Relationship between perception and acoustic measures
Both raters obtained moderate, but highly signi¿ cant, correlations between
their perceptual openness scores for the /͑i/ onsets and the normalized height
index computed from the acoustic formant measurements, with r = -0.69 (N =
32, p < 0.001) for RB and r = -0.62 for VH (N = 32, p < 0.001). The correlation
coef¿ cients are negative since 0% vowel height corresponds to the maximum
openness score on the perceptual scale from 0 to 10. The correlation coef¿ -
cients improve substantially, however, if only the ratings and acoustic indices
obtained for the 16 female speakers are taken into account, with r = -0.80 (N =
16, p < 0.001) for VH and r = -0.79 (N = 32, p < 0.001) for RB.
These correlations illustrate that the relationship between perception and
acoustics is quite strong for the female speakers but poorer for the males.
Although the male speakers are concentrated towards the non-low extreme of
the acoustic onset height dimension, there are a few men who have acousti-
cally low /͑i/ onsets, but who are not perceived as having particularly open
onsets. Closer inspection of the acoustic data revealed that these men did not
feature the stronger diphthongal glide that would be expected to accompany
the low /͑i/ onsets. In fact, the male speaker with the perceptually most open
/͑i/ onset had virtually no diphthongal glide at all. The women with extremely
low /͑i/ onsets, however, did have the corresponding stronger glides. It would
appear, therefore, that our perception of onset height in the diphthong /͑i/ is
co-determined by the size of the closing gesture.
In view of the preceding ¿ ndings it is dif¿ cult to claim that our relative
acoustic measure (even after normalization) corresponds in a straightforward
fashion with perceived vowel openness. It would appear that the acoustic