8 Dennis R. Preston and Nancy Niedzielski
back and forth between the two choices. Such experiments as these and
others reported by Niedzielski in this chapter reveal a sophisticated merger
of speech science and sociophonetic investigation in studying implicit and
explicit knowledge of language variation.
Chapter 12 is the ¿ rst of two that focus on group or ethic identi¿ cation
from speech alone by nonspecialist listeners. Here Thomas, Lass, and Car-
penter ask if African American English can be identi¿ ed, and, if so, if there
are demographic differences among those who make such identi¿ cations with
different degrees of reliability and if there are speci¿ c phonetic details that
permit the identi¿ cation. They focus on two regional groups and European
American and African American listeners’ perceptions of the fronting of /o/
(as in “hope”), a feature associated with European American but not African
American speakers in the US South, and on the raising of /æ/ (the vowel of
“bag”), a feature associated with southern African Americans but not Euro-
pean Americans. Sample sentences were presented unmodi¿ ed, monoton-
ized, and with all vowels converted to schwa. The local groups were superior,
regardless of ethnicity, but considerably hampered in correct identi¿ cation
when the vowel qualities were changed to schwa, and female voices were
identi¿ ed less correctly than male. In a second experiment, the same two vow-
els were presented along with sentences that contained no samples of vowels
known to be distinctive in the varieties under investigation in an attempt to
directly contrast prosodic clues with those of vowel quality. The results show
that both cues are important, making ethnic voice identi¿ cation dependent on
complexes of features rather than single elements.
Part III: Production and Perception
Chapter 13 continues the question of untrained listeners’ ability to identify
groups and includes new work on the characteristics of such groups. Purnell
¿ rst characterizes potentially salient linguistic features of German heritage
speakers of English in Wisconsin and African American English. His gen-
eral goal is to determine what relationship there is between the strength of
distinctiveness in these various aspects of the speech signal and the relative
salience of those features to identi¿ cation. In the ¿ rst experiment, a study is
made of the acoustic correlates to ¿ nal consonant devoicing, ¿ rst showing
that younger speakers in the German heritage area are more aligned with the
oldest speakers in this community, a relationship often uncovered in such sec-
ond language background speech communities. The results show the effec-
tiveness of a trading model for devoicing, in which glottal pulse is inversely