346 Paul Foulkes, Gerard Docherty, Ghada Khattab, and Malcah Yaeger-Dror
and American listeners the proportion of “girl” responses is approximately equal
when they hear stimuli with plain or laryngealized tokens. For the Tynesiders,
however, there is a clear difference according to variant. With plain tokens the
proportion of “girl” responses (58%) is similar to that from the non-locals and
Americans, and with no signi¿ cant difference according to listener group. But
with laryngealized tokens the Tyneside listeners gave signi¿ cantly fewer “girl”
responses (39%) than they did for plain tokens (chi sq = 21.289, df = 1, p < .001).
The Tynesiders furthermore gave signi¿ cantly fewer “girl” responses to the
laryngealized tokens than either of the control groups (Tynesiders versus non-
local UK: chi sq = 15.803, df = 1, p < .001; Tynesiders versus Americans: chi sq
= 22.608, df = 1, p < .001; no difference between the two control groups).
5.8 Results for pre-pausal variants
Finally, Figure 14.8 illustrates responses to stimuli with pre-pausal /p t k/.
The variant was shown to be signi¿ cant for all three listener groups in the
regression analyses. The data in Figure 14.8 enable us to explore these pat-
terns more fully.
Figure 14.7 Percentage of “girl” responses to word-medial tokens by listener group
and variant.