246 Dennis R. Preston
These ¿ ve groups were entered as one factor group for the VARBRUL run,
and sex and the six vowels of the NCS were also tabulated as factor groups.
The dependent variable was accuracy of comprehension of the vowel.
Based on previous work in the area (e.g., Evans et al. 2000), I expected the
groups to be ordered as follows as regards comprehension, if comprehension
can be linked to the degree of participation in the shift: USM > RSM > MM
AASM > ASM.
Table 10.2 shows the overall results of the study, and these single word
tokens obviously cause considerable confusion. Two vowels, /ɬ/ and short /͑/
(= [ࣜ]), do not even reach the .50 level. Ash 1988 also found less than .50 com-
prehension by Chicagoans of radically shifted /ɬ/and /͑/.
Table 10.2 GOLDVARB Weights and Percentages for Factors InÀ uencing the Com-
prehension of NCS Vowels
Factor Group Weight %
Female n.s. 0.66
Male n.s. 0.62
USM n.s. 0.66
RSM n.s. 0.66
ASM n.s. 0.64
MM n.s. 0.61
AASM n.s. 0.57
͑ > @ 0.819 0.90
ľ 0.809 0.90
0.582 0.74
ࣜ 0.566 0.73
Ԍ 0.406 0.58
ɬ 0.290 0.46
͑ >ࣜ@ 0.093 0.18There is some rearrangement of the order of groups from the prediction
given previously, but the group that has completed or nearly completed the
NCS (USM) is the one which comprehends best, and one of the groups that
has been shown to participate least in the NCS (AASM) comprehends worst.
The rural speakers from southern Michigan (RSM), since they are closer to