266 Erik R. Thomas, Norman J. Lass, and Jeannine Carpenter
African American voices from European American voices is, in general,
yes. In fact, accuracy rates in all but a few of these studies fell in the range
of 70% to 90%.
A number of studies also have af¿ rmatively answered the second ques-
tion, that of whether demographic differences among speakers and/or listen-
ers affect identi¿ cation rates. A few studies have reported that some listeners
have dif¿ culty identifying African Americans whose English sounds quite
standard (Buck 1968; Tucker and Lambert 1969; Abrams 1973; Baugh 1996).
Similarly, African Americans whose speech shows features typical of local
European American vernaculars also confound listeners (Wolfram 2001;
Mallinson 2002; and Thomas and Reaser 2004a). Some studies (Bailey and
Maynor 1989; Haley 1990; Thomas and Reaser 2004a) have reported that
younger African Americans can be easier to identify than older African
Americans, though it should be noted that the listeners in these experiments
were college-aged. Demographic features of listeners also affect discrimi-
nation ability. Alvarenga (1971), Haley (1990), Hawkins (1992), and Fore-
man (2000) all found differences among listener groups in the accuracy of
their identi¿ cations; in each case, the differences probably had to do with
the listeners’ familiarity with speakers of both ethnicities. Foreman (2000)
reported most explicitly that the best identi¿ cation was by listeners with
extensive contact with both groups.
Investigators have approached the third question, what features listeners
can access for identi¿ cations, in a piecemeal fashion. Two earlier studies,
Roberts (1966) and Bryden (1968a), reported that reading errors cued identi¿ -
cations of a speaker as African American. Subsequent inquiries have focused
on phonetic attributes. Vowel quality was found to inÀ uence identi¿ cations by
Lass, Almerino, Jordan, and Walsh (1980), Graff, Labov, and Harris (1986),
Purnell et al. (1999), and Thomas and Reaser (2004). The latter three studies
found that quality differences of particular vowels—/o/ as in coat, /au/ as in
how, and the /͑/ in hello—affected identi¿ cations. Some counterevidence is
provided by Bryden (1968a), though he examined only /u/, as in too. Foreman
(2000) performed an experiment to demonstrate that intonation could serve
as a cue, though Koutstaal and Jackson (1971) had found earlier that intona-
tion could not be the only cue that listeners used. Both Hawkins (1992) and
Walton and Orlikoff (1994) report that overall F0 level serves as a cue, with
lower F0 associated with African Americans; both of these studies focused on
male speakers. Walton and Orlikoff (1994) also reported that two voice qual-
ity features, jitter and shimmer, were correlated with ethnic identi¿ cations,
and Purnell et al. (1999) noted the same for the harmonics-to-noise ratio (itself
a cover feature for other voice quality factors). The conclusion emerging from