292 Thomas C. Purnell
folk psychological essentialism and general perceptual categorization. Basic
processes involved in categorization and concept acquisition (Wisniewski
2002: 475) are no doubt active in dialect perception.^3 The existence of these
perceptual processes gives rise to a folk principle of transparency, which
super¿ cially contradicts acoustic opacity. This belief in transparency can be
stated as: absent dissonant evidence, a speaker of a variety af¿ liated with a
certain social or geographical group X is an authentic or credible X-ian. This
is considered to be a case of psychological essentialism, i.e., “... people’s
concepts contain the belief that things have essences or underlying natures
that make them what they are” (Wisniewski 2002: 501).^4 This is also assumed
in the matched guise perception paradigm where listeners display the ability
to believe that a speaker is of a certain ethnicity just because they produce
an utterance with characteristics peculiar to a speci¿ c speech community.^5
For example, in a forced-choice matched guise experiment, a subject might
respond that a speaker is African American on the basis of hearing the speaker
say [ ks] for ask. What is important within the context of Figure 13.1 is that
the operating procedures behind the belief of transparency often make use
of a reduced set of acoustic triggers. That is, matched guise studies show
that one need not have an “authentic” speaker, just one who can produce the
features well enough to trigger the identi¿ cation by listeners. Such perceived
ethnicity via perceptual cues is emphasized in this chapter by the reanalysis
of previously reported ethnically af¿ liated data.
Regarding the role of phonetic aspects as an additional internal process,
we have known for some time that phonetics fosters variation in general by
providing multiple pathways to perception. To begin with, phonetic speech
events are often highly complex and redundant with respect to both articula-
tion and acoustics affording listeners a “family” of cues to attend to. Lisker
(1986) observes this stabilizing effect of a many-to-one relation holding
between perceptual cues and the voicing contrast in English coda obstruents
(the difference between the words bid and bit). Voicing, Lisker and others
have found, is reÀ ected in pitch, formants, duration of vowel or consonant,
amount of glottal pulsing, etc. The upshot of this multi-layering of gesture,
acoustic characteristic and perceptual cue ensures correct perception of pho-
nemes (MacNeilage 1970).
We might expect, then, that every acoustic cue has some role to play in
the perception of speech, even if it is just a supporting one. Yet, not all acous-
tic characteristics are perceptual cues. By way of example, Abramson (1986)
considered the perception of word initial geminate stops in Pattani Malay to
come from four signi¿ cant acoustic characteristics. However, he discovered
later that amplitude differences on the following vowel are the perceptually