78 Rebecca Roeder
The second column of Table 3.1 lists normalized F1 means for /æ/ and
shows that only two speakers have a normalized F1 above 750 Hz. 700 Hz
or lower (normalized) is considered Northern Cities Shifted by Labov, Ash
and Boberg (2006), indicating that most of these speakers are shifted or close
to shifted in their pronunciations of /æ/. Comparison of the F1/F2 means for
/æ/ to the F1/F2 means for /͑/ in the speech of each individual further distin-
guishes the respondents.
The third column in Table 3.1 gives the position of the average for tokens
of /æ/ relative to the position of the average for tokens of /͑/ in the vowel space,
based on t-tests done in Plotnik. For example, Lucy has a mean F2 for /æ/ that is
different from her mean F2 for /͑/ at a probability level of .05, such that /æ/ is
signi¿ cantly further front in the vowel space than /͑/. But her F1 means for the
two vowels are not signi¿ cantly different. The difference in means for /æ/ and
/͑/ is signi¿ cant in both F1 and F2 for only two speakers, Judith and Jose, who
are at the top of the list. In these speakers, /æ/ is fronted and raised in relation
to /͑/, which is typical of an advanced NCS system and suggests that Judith and
Jose are more advanced into the NCS than any of the other speakers.
The next six speakers have an /æ/ that is fronted in relation to /͑/ but is not
raised above /͑/. This is not surprising given that many Anglo NCS speakers,
including the young Detroit women used as a control group in the larger study,
do not raise /æ/ above /͑/ either. The F1/F2 means for /æ/ and /͑/ in the next four
speakers on the list are not signi¿ cantly different between vowels. Overlapping
means are common in the speech of people who exhibit only incipient or con-
servative accommodation to the NCS, as may be the case with these speakers.
Finally, the last four respondents listed in Table 3.1 have conservative non-NCS
vowel con¿ gurations, in which /æ/ is backer or lower than /͑/.
In the following analysis of environment, the top eight speakers in Table
3.1, who appear to be more affected by local Michigan norms than the others,
are grouped together and referred to as Group A. Both gender and level of
education have been strongly correlated with accommodation to local stan-
dards in previous studies, and it is notable that, out of these sixteen partici-
pants, the three who have college degrees (Jose, Solana, Edmund) and all but
one of the women under the age of 25 are included in Group A. Ralph, the
only Group A respondent who is neither a woman nor a college graduate, is a
monolingual speaker of English who works in sales at an upscale department
store and presumably must conform to accepted norms to ful¿ ll the require-
ments of his occupation.
To summarize, Table 3.1 illustrates that—with regard to the vowels /æ/
and /͑/—these second and third generation speakers fall along a continuum
from fairly typical Northern Cities shifters (Group A) to unshifted (Walter,