236 Bartáomiej Plichta and Brad Rakerd
The response patterns were very similar for the two talkers, with the /a/-
to-/æ/ cross-over point occurring at steps 4.3 for Talker UP, and at step 4.5 for
Talker LM. There was no statistical difference between these two values (t(8)
= 0.96; p > 0.05). This strongly suggests that the respondents from the UP did
not take NCCS inÀ uences into account when interpreting vowel cues.
4.9 Results for the LM respondents
The LM respondents had resided in Southeast Lower Michigan throughout
most of their lives and they were highly familiar with speech marked by
NCCS. Figure 9.9 shows the LM respondents’ results for the sociophonetic
test. There is strong evidence that they were sensitive to the dialectal dif-
ference between the two talkers. For Talker UP (upper panel) the cross-over
from /a/ to /æ/ occurred at step 4.3. For Talker LM (lower panel), the result
was signi¿ cantly different (t(8) = 4.34; p < 0.002), with the cross-over occur-
ring at step 5.4, more than a f ull step f ur ther front (i.e., at a higher value of F2)
along the continuum than for Talker UP.
Another point to be made about the response pro¿ le for the LM respon-
dents is that it was very similar to that of the UP respondents when the talker
was from the UP and had a speech pattern that showed little evidence of
NCCS. This can be seen by comparing the upper panels of Figures 9.8 and 9.9,
both of which show a crossover value of 4.3. But a cross-over difference of 0.9
continuum steps arose when the talker was from Lower Michigan and exhib-
ited NCCS (compare the results for Talker LM shown in the lower panels of
the ¿ gures; cross-over = 4.5 for UP respondents and 5.4 for LM respondents.)
Apparently the LM respondents formed a perceptual representation of NCCS
speech—a representation that included /a/ fronting—and when they encoun-
tered a talker whose speech pattern was consistent with that representation it
modulated their perceptual interpretation of the available vowel cues. The UP
respondents, on the other hand, had no such representation (and showed no
vowel shift), presumably owing to their much more limited experience with
speakers who exhibit NCCS.
- The role of the speech community
The results of the present study showed: (1) that information about a talker’s
dialect can play a signi¿ cant role in vowel identi¿ cation; and (2) that this effect
is constrained by listener-dependent sociolinguistic factors. Speci¿ cally, it