A Reader in Sociophonetics

(backadmin) #1

382 Betsy E. Evans


also concluded that, while the imitations were generally accurate, there were
some features of the target voices (e.g., voice quality) that were more dif¿ cult
for the imitators to reproduce.
Rogers (1998) examined the possibility for non-native speakers to speak
with a less strong foreign accent than they usually do. This was prompted
by his forensic work with a threatening message recorded on an answering
machine. The recording appeared to be made by a Cantonese speaker of
English. Rogers compared the recording to the speech of the suspect. He
concluded through auditory and acoustic analyses that the suspect’s and
recorded voice were unlikely to belong to the same individual as the suspect
showed consistent differences in accent while speaking English from the
voice on the tape (e.g., differences in the spectral sections of [s]). He further
concluded that non-native speakers can imitate a stronger accent than they
normally have but not a weaker one. Lindsey and Hirson (1999) showed that
speakers with “disordered” /r/ could produce “standard” /r/ by measuring F3
of respondents’ /r/ in their normal speech and when imitating recordings of
speakers with “standard” /r/. Three of ¿ ve subjects judged to have nonstan-
dard /r/ were able to produce standard /r/ while imitating a taped stimulus
with standard /r/. This is especially interesting in light of Rogers’ (1998)
conclusions about the inability of non-native speakers to approximate a more
“native” accent and again highlights the need for research on a person’s abil-
ity to modify a speech idiosyncrasy.
The issue of F0 as a unique feature is widely debated (e.g., Braun 1995;
Gfroerer and Wagner 1995; Boss 1996). Künzel (2000), for example, studied
the effects of and preferences for a different type of voice disguise with regard
to F0. Respondents were invited to read material presented by the investiga-
tor using raised or lowered pitch or in a denasalized (while pinching the nose)
manner. Five recordings were made with six-week intervals. Mean funda-
mental frequency was then determined for undisguised and disguised speech.
Künzel’s results indicate that speakers changed F0 consistently according to
the F0 in their “normal” speech. That is, speakers with higher-than-average
F0 tend to increase their F0 levels and speakers with lower-than-average F0
tend to disguise their voice by lowering F0. Künzel concluded that, in terms
of identi¿ cation of speakers, “a speaker’s natural level of fundamental fre-
quency can be predicted with a reasonable error margin from his/her dis-
guised values” (173).
It seems that a common procedure for studies of imitation involves respon-
dents who have been instructed how to disguise their voice or professional
imitators. What could be learned by exploring the ability of non-professionals
to imitate a dialect different from their own?

Free download pdf