Classi¿ cation of Regional Language Varieties 209varieties, the extent to which the relevant variety is made available through
linguistic context, and the beliefs that the listener has about the talker can all
affect the mapping between the acoustic signal and the cognitive representa-
tion of the vowel category. Additional sociophonetic research is needed to
determine the limits of naïve listeners’ abilities to adapt to unfamiliar dia-
lects, the role of category mismappings on spoken language processing and
speech recognition, and the relationship between category boundaries in per-
ception and production.
- Perception of social categories
Much of the research in variationist sociolinguistics has focused on docu-
menting the acoustic properties of socially-based variation. Labov et al.
(2006) have undertaken an enormous project to document, describe, and ulti-
mately de¿ ne regional phonological variation in the United States. However,
far less is known about naïve listeners’ perception and representation of this
important source of inter-talker variability. Over the past 15 years, a number
of studies have used methods developed in cognitive psychology and speech
science to explore the explicit categorization and discrimination of regional
varieties of American English. These studies have all used samples of natural-
ly-produced speech, typically sentence-length or longer, which contain mul-
tiple dialect-speci¿ c target words or phonemes. The listener populations have
been manipulated to include listeners from different regional backgrounds
and with different degrees of geographic mobility to explore the role of these
two listener-related factors in perception. The results of these studies provide
insights into the cognitive dialect categories naïve listeners construct through
their interactions with people from their own and other dialect regions, the
kinds of phonological variants that naïve listeners ¿ nd salient, and the role
that these acoustic details play in social category construction.
In one early study, Preston (1993) used a dialect identi¿ cation task to
explore naïve listeners’ ability to identify the regional background of unfamil-
iar talkers. The stimulus materials were short extracts of narratives produced
by nine male talkers from nine different cities on a north-south continuum
between Michigan and Alabama. He asked adult listeners in Michigan and
southern Indiana to listen to each sample and then indicate which city they
thought the talker was from. The results revealed that the listeners had dif¿ -
culty distinguishing between northern and midland talkers, but that they were
more accurate in distinguishing northern from southern talkers. In addition,
the region of origin of the listeners (Michigan vs. Indiana) had a signi¿ cant