American Speech

SOCIOPHONETIC APPLICATIONS
OF SPEECH PERCEPTION
EXPERIMENTS
ERIK R. THOMAS
North Carolina State University

A lthough studies of perception are still largely assigned to the realms

of experimental phonetics or psychology, sociolinguists have been recog-
nizing the importance of perception. Several lines of experimental inquiry
have emerged. Nevertheless, perception has been studied far less by
sociolinguists than has speech production. One reason is that speech
perception is daunting at first. Examining it requires careful attention to
experimental design, a considerable amount of preparation, and, in many
cases, use of a speech synthesizer. Even so, research on perception can be
highly productive. This paper attempts to review the sorts of experiments
that have been conducted in the past and to provide guidelines for
sociolinguists interested in studying perception, with suggestions for future
work.
Although perception has been a neglected stepsister of production in
sociolinguistics, it, like Cinderella, may have its day soon. Two important
factors could—and should—move perception to the forefront of socio-
phonetic research. One is simply the huge potential for sociolinguistic
perception studies because the area has been neglected for so long. The
other reason is a more practical one: although perception experiments
require extreme attention to detail in the preparation phase, data analysis
is generally less time-consuming than in production studies, and this
difference may make it more attractive to researchers.
The aversion of much of sociolinguistics to perception has been, to
some extent, more apparent than real. Many sociolinguistic studies over
the past generation, especially instrumental studies, have succeeded in
divorcing speech production from speech perception. However, percep-
tion issues may play a hidden role in studies that ostensibly address produc-
tion. The reason is that variationists have not always carefully distinguished
production from perception. This tendency is an artifact of the reliance of
sociolinguistics on impressionistic transcription. The impressionistic tradi-
tion, based on the development of the International Phonetic Alphabet
and of the Cardinal Vowel system of Daniel Jones, dominated dialect

American Speech, Vol. 77, No. 2, Summer 2002

Copyright © 2002 by the American Dialect Society

115

Published by Duke University Press

 American Speech

116 american speech 77.2 (2002)

geography, which was largely conceived before modern acoustic equip-

ment was developed. This reliance has continued, for the most part, in
sociolinguistics. The shortcoming of this approach is that, although linguis-
tic atlas transcribers and sociolinguists believe that they are recording
subjects’ production, what they actually record is filtered through their
own perceptual labeling abilities, practices, and strategies (Kerswill and
Wright 1990).
An example of a hazard inherent in such a mixture involves one of the
most frequently studied phenomena of English: the deletion of final stops
in consonant clusters, as in lift pronounced [lIf] or desk pronounced [dEs].
Studies of consonant cluster simplification have customarily involved im-
pressionistic transcription. However, Browman and Goldstein (1990) showed
experimentally that, because of the overlapping nature of articulatory
gestures, listeners may be unable to hear a stop in a consonant cluster, such
as the [t] in perfect memory, even when an articulatory gesture is demonstra-
bly present. Surprenant and Goldstein (1998) furthermore found predict-
able patterns in the perceptual masking of stops in clusters. These patterns
matched many of the results reported in sociolinguistic studies of deletion
of consonants in clusters (e.g., Wolfram 1969; Labov 1972; Guy 1980). The
implication is that many of the stops tabulated in sociolinguistic studies as
“deleted” were, in fact, articulated but were inaudible to the transcribers.
Kerswill and Wright (1990) found that the opposite can happen as well:
listeners may record an alveolar stop when none exists. Studies of conso-
nant cluster simplification primarily describe the perception of transcrib-
ers and only secondarily the production of speakers. The binary interpreta-
tion of final stops as either present or deleted may describe perception
satisfactorily, but it cannot truly describe production (see Janson 1983, 22–
23, for another example).
Marriages of production and perception are certainly not invalid per
se; as Lindblom (1980) notes, acoustic measurements of speech are mean-
ingless unless they can be related to perceivable factors. However, variationists
need to acknowledge more consistently the fact that production and per-
ception do not always correspond neatly. Directing more studies at percep-
tion would help them to do so.

PREVIOUS SOCIOPERCEPTUAL APPROACHES

Sociolinguists and other investigators have utilized a variety of types of

perception experiments to examine several sorts of sociolinguistic ques-
tions. Most work on perception in language variation has used spoken

Published by Duke University Press

 American Speech

Sociophonetic Applications of Perception Experiments 117

words or phrases that were played to listeners for the purpose of testing one
of five issues: (1) the ability of listeners to identify the regional dialect,
ethnicity, or socioeconomic level of speakers; (2) how stereotypes can
influence the perception of sounds; (3) the presence of vowel mergers or
splits in perception; (4) how dialectal differences affect the categorization
of phones; and (5) stereotypical attitudes, which are investigated by having
subjects assess the personality of a speaker, the speaker’s suitability for
particular jobs, or other personal traits of the speaker. These issues are
reviewed in the following sections, with the first split into three sections and
the other four issues each represented by one section. This list of issues is
somewhat limited, however, and many other objects of perceptual investi-
gation are certainly possible. If the topic of socioperceptual studies is
broadened to include not only synchronic language variation but also
inquiry into the perceptual causes of diachronic change, especially by
comparisons with misperceptions in a laboratory setting, additional studies
could be added: see in particular John J. Ohala’s work on the role of
misperception in sound change (e.g., Hombert, Ohala, and Ewan 1979;
Ohala 1981, 1985, 1989, 1993), but see also Browman and Goldstein
(1991), who place more emphasis on the phasing of articulatory gestures,
Jonasson (1971), and Foulkes (1997). Another important related topic to
which perception experiments have been applied is the degree of
accentedness of second-language or ethnic speakers (see, e.g., Brennan,
Ryan, and Dawson 1975; Brennan and Brennan 1981; Sebastian and Ryan
1985). Further broadening of the topic could encompass studies correlat-
ing acoustic parameters of speech with emotional states (see, e.g., Goldbeck,
Standke, and Scherer 1988).

identification of the regional dialect of speakers. Within this cat-

egory, I include studies in which recordings of speakers of different dialects
are played to subjects. Not included is “perceptual dialectology” or “folk
dialectology,” in which subjects demarcate dialect boundaries (usually on a
map) or label dialects according to how “correct,” “pleasant,” and the like
the subject considers them. Most studies of that type do not involve record-
ings. Examples of papers on perceptual dialectology include much of
Dennis R. Preston’s recent work (e.g., 1986, 1993a, 1993b, 1996) and the
collection of articles in Preston (1999).
One of the first experiments involving dialect identification of record-
ings was Bush’s (1967) study in which listeners were asked to identify the
dialect of voices speaking American, British, or Indian English. The stimuli,
which included nonsense words, real words, and sentences, were played
either in original form, low-pass filtered, high-pass filtered, or center-

Published by Duke University Press

 American Speech

118 american speech 77.2 (2002)

clipped. The filtering provided some indication about whether listeners

could base their identifications on prosodic factors. As it turned out,
listeners were generally able to identify the dialect of the filtered stimuli
75% or more of the time.
A more recent experiment is described in Preston (1993a, 359–67;
1993b, 41–46; 1996, 320–28). In this study, tapes of nine speakers who
lived along a transect from Michigan to Alabama were played to two groups
of subjects, one from southeastern Michigan and the other from southern
Indiana. The subjects were asked to match the voices they heard with points
along the transect, as shown on a map. They were fairly accurate in their
assessments. However, the Michigan and Indiana subjects differed in which
speakers they were best able to discriminate. Similar experiments have
been conducted elsewhere. Stephan (1997) played recordings of 12 speak-
ers of English dialects from around the world to German university stu-
dents and found that American English was identified correctly most often,
while South African and Welsh English were identified correctly least often.
Williams, Garrett, and Coupland (1999) played excerpts of recordings of
speakers from six parts of Wales, as well as standard British speakers, to
students and schoolteachers from various parts of Wales. Subjects were
asked to identify the dialect of each speaker and to rate the speaker’s
“Welshness” and likability. The teachers were more accurate at identifying
the dialect than the students, but the two groups were fairly consistent in
their ratings of speakers’ Welshness.
Another experiment involving identification of the dialect of speakers
is described in Wolfram, Hazen, and Schilling-Estes (1999, 129–31). They
played to listeners four variants of /O/, as in caught, uttered by speakers of
different dialects. They asked the listeners to rate the variants on scales of
most-to-least Northern-sounding and most-to-least Southern-sounding and
then to guess where each speaker came from. Their aim was to test
reactions to the raised, monophthongal variant of /O/ that occurs around
the Pamlico Sound, and the results showed that it ranked high on the
Northernness scale and low on the Southernness scale. A rather similar
protocol was used by Munro, Derwing, and Flege (1999), who investigated
how much the dialect of adults is modified after they move to a new region.
Recordings were made of Canadians living in Canada, Canadians living in
Alabama, and Alabama natives. Canadian listeners were then asked to rate
the voices on a scale from “very Canadian” to “very American,” while
Alabama subjects were asked to rate the voices on a scale from “definitely
from Alabama” to “definitely not from Alabama.” Both groups of listeners
rated the Canadians living in Alabama as intermediate between the Canadi-
ans living in Canada and the Alabama natives, indicating that the Canadi-

Published by Duke University Press

 American Speech

Sociophonetic Applications of Perception Experiments 119

ans living in Alabama had undergone some dialectal shifting. The authors
determined that the overall rate of speech could not have influenced the
ratings, but that diphthong quality may have.
Clopper and Pisoni (2001) investigated how well subjects could distin-
guish speakers from different parts of the United States and what features
they relied on. Sixty-six young, white males from six regions of the United
States (11 from each region) read two sentences. These sentences were
played to subjects from Indiana, who were asked to match each utterance
with a region. Although overall identification accuracy was only 25%, it was
greater than chance (17%), and the difference was statistically significant.
Comparisons with acoustic measurements of the utterance revealed that
the identifications were correlated with certain dialectal variants, including
r -lessness, /s/ or /z/ in greasy, and several aspects of vowel quality.
Reneé van Bezooijen and her colleagues have conducted several dia-
lect identification experiments in the Netherlands and England. Gooskens
(1997) and Bezooijen and Gooskens (1999) investigated the relative im-
portance of prosodic and segmental information for listeners reacting to
recordings of various dialects. They conducted separate experiments for
dialects of Dutch and of British English. Signals to be used for stimuli
underwent three treatments. One treatment was low-pass filtering at 400
Hz, which eliminated most of the segmental information in the signals and
thus forced listeners to focus on prosody. Another treatment was
monotonization, in which the pitch contour of the signals was flattened so
that F0 was always the same, eliminating intonation.1 The other treatment
was simply resynthesis of the original signals to produce control stimuli,
which ensured that these stimuli had the same segmental quality as the
other two versions. Two different tasks were given to subjects. One was to
rate the degree of dialectal divergence of each stimulus from either stan-
dard Netherlands Dutch or standard British English. The other was to try
to identify what region and locality the speaker of each stimulus was from.
The results showed that English listeners relied more on prosody in their
judgments of British dialects than Dutch listeners did for Dutch dialects.
Bezooijen and Berg (1999) investigated the intelligibility of four regional
dialects of Dutch to speakers of near-standard Netherlands Dutch. Speech
fragments from the four dialects consisting only of a noun and function
words were played to listeners who spoke standard or near-standard Dutch.
The listeners were asked to translate the fragments into standard Dutch.
Their rates of mistranslations of the nouns were compared with an objec-
tive rating of how each noun differed from its equivalent in standard
Dutch, that is, the number of different sounds, a different lexical item, or a
semantic difference. As expected, the more similar the dialect noun was to

Published by Duke University Press

 American Speech

120 american speech 77.2 (2002)

the standard noun, the fewer mistranslations there were. Bezooijen and
Ytsma (1999) had listeners identify various Dutch dialects, rate their diver-
gence from each other, and give their subjective personality impressions of
them. In general, the more southerly dialects were easiest to identify and
rated as most divergent, while standard Dutch was regarded as sounding
most arrogant.

identification of the ethnicity of speakers. Perception experiments

testing the ability of listeners to name the ethnicity of speakers have
involved playing recordings of African Americans and European Ameri-
cans, generally either field recordings or tapes of speakers reading a story,
to subjects who were asked to name the speakers’ ethnicity. Using this
method, Dickens and Sawyer (1952); Stroud (1956); Hibler 1960; Larson
and Larson (1966); Roberts (1966); Buck (1968); Shuy, Baratz, and Wol-
fram (1969); Tucker and Lambert (1969); Shuy (1970); Abrams (1973);
Irwin (1977); Lass et al. (1979); Bailey and Maynor (1989); Haley (1990);
and Trent (1995) found that listeners could identify the ethnicity of
speakers much of the time or even nearly all the time. Some three-way
ethnic identification studies have been conducted as well. Baugh (1996)
showed that listeners could distinguish African Americans, European Ameri-
cans, and Chicanos; see also Purnell, Idsardi, and Baugh (1999). Wolfram
(2000) conducted ethnic identification experiments using speakers from
two unusual communities in North Carolina: Hyde County, in which Afri-
can Americans show considerable assimilation to the local European Ameri-
can dialect, and Robeson County, which has a triethnic assemblage of
European Americans, African Americans, and Lumbee Native Americans.
Outsiders nearly always misidentified older Hyde County African Ameri-
cans and usually misidentified Lumbees, but Lumbees themselves were
usually able to distinguish other Lumbees. While all of these studies have
demonstrated that listeners can often identify the ethnicity of a speaker,
their major limitation is that they could not determine what features
listeners rely on to make the identifications, though Roberts (1966) found
mispronunciations to be associated with identification as African American.
A number of studies have attempted to address this issue. Bryden
(1968) conducted two experiments testing whether differences in nasality
could serve as a cue. In the first experiment, ten European American and
ten African American speakers read a short passage. The recordings were
presented either unmodified or with bandwidth compression, which re-
duced nasality. The ethnicity of speakers was identified accurately more
often for the unmodified versions. In the second experiment, 91 speakers
of both ethnic groups and a variety of socioeconomic backgrounds read the

Published by Duke University Press

 American Speech

Sociophonetic Applications of Perception Experiments 121

same passage and the words beet and boot in frames. The recordings were
labeled for ethnicity and evaluated for speech proficiency. A direct and
statistically significant correlation was found between the number of read-
ing errors and perception of a speaker as African American, but correla-
tions with nasality and vowel quality were not significant. Koutstaal and
Jackson (1971) determined, by comparing acoustic measurements with
identification accuracy, that intonation and timing differences could not
be the only cues. Lass, Mertz, and Kimmel (1978) found that playing
recordings of African Americans and European Americans backward (which
would make the segments difficult to recognize but preserve some aspects
of prosody and voice quality) or compressing the recordings temporally
reduced the accuracy of ethnic labeling by European American listeners.
Lass et al. (1980) played recordings of ten African American and ten
European American speakers to listeners under three treatments—
unfiltered, low-pass filtered at 225 Hz, and high-pass filtered at 225 Hz—
and found that filtering adversely affected the accuracy of ethnic
identification by listeners.
Two studies have examined aspects of voice quality more specifically.
Hawkins (1993) conducted a two-part study on the role of F0 in ethnic
labeling. In the first phase, equal numbers of African Americans and
European Americans, male and female, produced [æ] and [i] in isolation,
in words, and in sentences. Listeners of both ethnicities and sexes were able
to identify the speaker’s ethnicity with better than random accuracy. Mea-
surements of the stimuli implicated F0 as a cue. In the second phase,
isolated [æ] produced by two African American and two European Ameri-
can speakers was synthesized at nine F0 levels. Listeners equally divided by
ethnicity, sex, and residence labeled the ethnicity of the stimuli. Lower F0
was found to be correlated with labeling as African American, but compari-
son of the results for different listener groups suggested that the difference
was based on stereotype, not physiology. Walton and Orlikoff (1994)
presented listeners with 50 paired recordings of sustained /A/ vowels, one
spoken by a European American speaker and one by an African American.
All 100 of the speakers were adult males. Listeners had to judge which
speaker belonged to which ethnicity and were able to do so with 60%
accuracy. Then the researchers conducted a careful acoustic analysis of the
stimuli, examining jitter (F0 perturbation), shimmer (amplitude perturba-
tion) and harmonics-to-noise ratio, and found ethnic differences for all
three cues. They also found correlations between how much the paired
speakers differed in those cues and how accurate the listeners were in their
ethnic judgments. In opposition to Hawkins, they speculated that physiol-
ogy was the basis for the difference.

Published by Duke University Press

 American Speech

122 american speech 77.2 (2002)

Other studies have focused on vowels or intonation. Graff, Labov, and

Harris (1986) took short excerpts of recordings containing two examples
of either /au/ or /o/ and modified F2 of those vowels in the recordings with
a synthesizer. Fronted nuclei of these vowels typify the production of
European American Philadelphians and backed nuclei that of African
American Philadelphians. African American, European American, and
Puerto Rican subjects from Philadelphia were asked to identify the stimuli
as African American or European American. The results showed that /au/
or /o/ tokens with higher F2 s were more likely to be identified as European
American speech than those with lower F2 s, although the listening task
forced subjects to focus on vowel quality, which could easily have biased the
results. Purnell, Idsardi, and Baugh (1999) found that European American
listeners could distinguish productions of hello by African Americans, Euro-
pean Americans, and Chicanos and determined that the quality of /E/ was
the most important factor, but that the duration of /E/, the location of the
stress on the first or second syllable, and the signal-to-noise ratio also
influenced the results. Thomas and Reaser (2002) used five-second ex-
cerpts of interviews of Hyde County, North Carolina, natives and control
speakers from inland areas. Two excerpts, one featuring diagnostic local
vowel variants and the other not doing so, were taken for each Hyde County
speaker. The excerpts were given three treatments, much like those in
Gooskens (1997): unmodified; monotonized to eliminate intonation; and
low-pass filtered at 330 Hz to eliminate most segmental information. The
three treatments were played to different groups of listeners. Various
prosodic and voice quality factors were measured in the stimuli. For the
unmodified and monotonized stimuli, only presence of diagnostic vowels
reached or approached statistical significance, and Hyde County African
Americans (who exhibit vowel variants more typical of European Ameri-
cans) were identified less accurately than other groups, indicating that
listeners used vowel variation for labeling. For the low-pass filtered stimuli,
certain prosodic and voice quality factors reached or approached
significance, but only for male voices.
Finally, Foreman (2000) investigated whether intonation could serve
as a cue to ethnic identification. African American and European Ameri-
can speakers read scripts that were designed to imitate conversational
speech. Recorded sentences were then low-pass filtered at 900 Hz to mask
some aspects of segmental and voice quality variation, thereby focusing
more attention on intonation. African American and European American
listeners were classified according to how much exposure they had to
African American and “mainstream” dialects. The results showed that
sentences with more diagnostic intonational cues were identified more

Published by Duke University Press

 American Speech

Sociophonetic Applications of Perception Experiments 123

accurately than those with fewer intonational cues. In addition, speakers

with extensive exposure to both dialects were most accurate in their label-
ing.
Taken together, these studies give the impression that, under certain
circumstances, listeners are capable of accessing a wide variety of cues to
determine whether a speaker is African American or European American.
Nevertheless, it is not yet clear which ones listeners use in real-life situations
or which ones are most important. More studies that compare different
cues are needed; single-feature studies have dominated the discussion so
far. Additional factors, such as rhythm, should be investigated. In addition,
labeling of other ethnic groups, such as Chicanos, and differentiating
ethnic groups in other Anglophone countries have received little attention.
A study by Heselwood and McChrystal (2000) investigating what features—
for example, retroflex stops and clear /l/—distinguish Panjabi English in
England represents a start in that direction. More studies of this sort could
be conducted on other languages as well. A great deal of work remains to be
done on dialect labeling.

identification of the socioeconomic level of speakers. Experiments

investigating the ability of listeners to name the socioeconomic class of
speakers have occasionally been conducted, often in conjunction with
ethnic identification. Two early studies by Harms (1961, 1963) demon-
strated that listeners of different social strata could often determine the
ethnicity of a speaker from a short excerpt. Shuy, Baratz, and Wolfram
(1969; also reported in Shuy 1970) had listeners identify the ethnicity and
social class of African American and European American speakers from
Detroit. They found that lower social classes were identified more accu-
rately than higher ones. Sebastian and Ryan (1985) found that, for both
middle-class and lower-class speakers, ratings by listeners for status-related
traits were lower for Spanish-accented speakers than for speakers with a
“standard” accent. They found in a follow-up experiment that medium-
accented speakers were rated higher than low-accented speakers for social
class, though high-accented speakers, as expected, were rated lower for
social class than other speakers.

influence of stereotypes on the perception of sounds. A few studies

have examined how stereotypes can alter subjects’ perception of speech.
Strand (1999) studied how listeners’ perception of a speaker’s sex affects
their identification of sounds. Her study was based in part on the “McGurk
effect” (McGurk and MacDonald 1976), in which subjects hearing one
sound and watching a video of a speaker uttering a different sound tend to

Published by Duke University Press

 American Speech

124 american speech 77.2 (2002)

perceive the sound that they “lip-read” from the video, not the sound that
they hear. Strand played to subjects recordings of [sA ~ SA], with the fricative
synthesized to match male or female frequencies of [s] or [S]. The vowel
had the same formant values for all the stimuli. Subjects simultaneously
watched a video of a male or a female uttering such a syllable. The results
showed that subjects altered their perception of the fricative depending on
the sex of the speaker that they saw, shifting the /s/- /S/ boundary to lower
frequencies for male faces and to higher frequencies for female faces.
Strand concluded that speech perception is influenced not just by the
physical attributes of sound but also by gender stereotypes. The fact that
male and female vowel formant values are not scaled uniformly (see, e.g.,
Fant 1966; Yang 1992) suggests that a similar phenomenon may exist for
vowel perception, and, in fact, a follow-up study by Johnson, Strand, and
D’Imperio (1999) showed that it does. The latter study examined the
/U/- /√/ boundary in hood and hud and found that it was affected by whether
subjects saw, or even imagined, a male or female face saying the words.
Niedzielski (1999) studied a different kind of perceptual shifting
caused by stereotyping. It is known that bilinguals shift their perceptual
boundaries between sounds depending on which language they believe
they are listening to (Elman, Diehl, and Buchwald 1977; Janson and
Schulman 1983, 331). Niedzielski found that such shifting may also occur
depending on what dialect listeners believe they are hearing or on their
expectations of their own speech. She played recordings of /au/, as in about,
and other vowels spoken by a Detroit native to a group of subjects from the
Detroit metropolitan area. Half of the subjects were told that the speaker
was from Detroit and the other half that she was from Canada. Subjects
were asked to match the vowels that they heard with resynthesized vowels to
best approximate the quality of the original vowel. Subjects told that the
speaker was from Detroit chose lower /au/ nuclei than those told that the
speaker was from Canada, which indicated that their perception was al-
tered by their stereotypes of American and Canadian speech. For the other
vowels, they chose qualities that matched widespread American forms
more closely than those found in the Detroit dialect. The latter finding
apparently resulted from the fact that most Detroit residents do not recog-
nize the distinctiveness of their own speech and hence harbor precon-
ceived notions that their speech is unmarked for dialect features. Both
Strand (1999) and Niedzielski (1999) show that speech perception does
not depend purely on physical factors, but also on listeners’ expectations
based on sociological factors. This finding demonstrates that variationists
can contribute a great deal to theories of speech perception: it is not just
the other way around.

Published by Duke University Press

 American Speech

Sociophonetic Applications of Perception Experiments 125

vowel mergers and splits in perception. There are a few studies of

vowel mergers in perception. Janson and Schulman (1983) used synthetic
stimuli to investigate how the perception of residents of Lycksele, Sweden,
who distinguish Swedish short /E/ and short /e/ in their production, might
differ from that of residents of Stockholm, who merge those vowels in
production. Subjects from each community were asked to categorize stimuli
that represented a continuum from [a] to [i]. The results showed that most
of the Lycksele subjects were unable to distinguish /E/ and /e/ consistently in
perception, even though they maintained the distinction in production.
The Stockholm subjects, as expected, were unable to distinguish /E/ and /e/.
Janson and Schulman suggested that Lycksele speakers had enough expo-
sure to the Stockholm dialect that they had ceased to use the distinction as
a means of distinguishing words because it was useless for understanding
Stockholm Swedish. Costa and Mattingly (1981) examined /A/ and /Ar/ in
the Boston area, where they are differentiated by length, and found that
listeners could not distinguish them. They reached the same conclusion as
Janson and Schulman, that distinctions persist in production after they
disappear in perception.
Labov, Karen, and Miller (1991) objected to the notion that listeners
may cease using a distinction for perception when they retain it in produc-
tion; they reasoned that the earlier findings were based on experiments
involving the labeling of isolated stimuli, an unnatural situation, not on
discrimination involving semantic distinctions in running speech. They
constructed experiments to test Philadelphians’ discrimination abilities
with pairs such as ferry and furry. These pairs are merged by some Philadel-
phians and in close approximation for many others. The experiments used
unsynthesized speech signals. One experiment involved interpretation of
an ambiguity in a story, which tested Janson and Schulman’s assertion; the
other involved identifications of isolated words from commutation tests, in
which recorded words uttered by a native speaker of the dialect are played
to subjects who are asked to identify them as, for example, ferry or furry.
Their results were not as categorical as those of Janson and Schulman, but
they did indicate that many Philadelphians could distinguish nearly merged
pairs perceptually, even though they were impaired in their ability to do so.
Labov (1994) and Labov and Ash (1997) report results from several other
commutation experiments conducted in various locations. Di Paolo and
Faber (1990, 166–67) describe a similar test given to subjects in Utah to
investigate the near-mergers of /il/ and /Il/ (as in feel and fill, respectively), of
/el/ and /El/ (as in fail and fell), and of /ul/ and /Ul/ (as in fool and full).
Perception experiments investigating vowel splits are scarce. Guenter
(2000) investigated whether California subjects identified vowels preced-

Published by Duke University Press

 American Speech

126 american speech 77.2 (2002)

ing /r/, /l/, and /N/ with vowels in other contexts. He played recordings of a
voice saying pairs such as beer/beet and sing/bid, as well as control pairs such
as grief/beet and bit/beet, and asked the subjects to identify the vowels in the
two words as “the same” or “different.” He also recorded response times.
Subjects showed lower rates of identification as “the same” when one vowel
in a word pair preceded /r/, /l/, or /N/ than when identical vowels preceded
other consonants. Response times were slower for pairs in which one
member preceded /r/, /l/, or /N/ than for pairs in which identical vowels
preceded other consonants or pairs in which the vowels differed. The
difficulty that subjects exhibited in identifying vowels before /r/, /l/, and /N/
with any phoneme suggested that vowels in these contexts were splitting
from their counterparts in other contexts.

dialectal differences in categorization of phones. Relatively few

experimental studies test discrepancies in how speakers of different dia-
lects categorize sounds. Experiments on cross-dialectal differences in the
perceptual boundaries between sounds are relatively easy to design and
could be applied extensively, though few have been carried out. Willis
(1972) asked subjects in Buffalo, New York, and neighboring Fort Erie,
Ontario, to categorize synthetic vowels according to phoneme. He found
that some of the phoneme boundaries differed for residents of the two
communities and that two of the most strongly differing boundaries (/E/-/á/
and /á/-/A/) corresponded directly with production differences between
the dialects of the communities. Janson (1983, 1986) investigated a differ-
ence in generational dialects in Stockholm Swedish, in which /a:/ is under-
going backing. He had subjects identify synthetic stimuli that represented a
continuum between the Swedish phrases ett tag ‘a while’ and ett tåg ‘a train’
as one or the other of the two phrases in order to locate the perceptual
boundary between /a:/ (as in tag) and /o:/ (as in tåg). He found that the
change was reflected in encroachment of /a:/ on the perceptual space of /o:/
across generations in Stockholm, though the perceptual change was not as
fast as the change in production. Furthermore, Janson (1986) found that
the shift in the perceptual boundary was not occurring in Helsinki Swedish,
which shows no shift in production. Malderez (1995) conducted a similar
experiment on the encroachment of /o/ upon /ø/ in French. A potential
pitfall of this type of experiment is that, as Niedzielski (1999) found,
listeners are not always able to match sounds with their own production
norms accurately.
Experiments in cross-dialectal differences in which phonetic cues are
used to distinguish sounds are possible as well. In Thomas (2000), I

Published by Duke University Press

 American Speech

Sociophonetic Applications of Perception Experiments 127

compared the perception of varying heights of the glide of /ai/, as in tide and
tight, by Anglos from central Ohio and Mexican Americans from southern
Texas. In production, both groups produced higher glides before a voice-
less consonant, as in tight, than before a voiced consonant, as in tide, though
the Anglos did so to a greater extent than the Mexican Americans. A
perception experiment using synthetically modified stimuli representing a
continuum from tide to tight revealed that both groups used the glide
difference as a perceptual cue, but in a different way. The Ohio Anglos used
the glide difference mainly as a cue to the identity of following [d] versus
[t], but the Texas Chicanos used it more as a cue to following [d] versus null
(e.g., tide vs. tie). A possible reason was that the stimuli lacked stop releases,
which may have been more important as a cue to the identity of final [t] in
Texas Chicano English than in Ohio Anglo English.
A somewhat similar experiment was Peeters’s (1991) investigation of
cross-linguistic and cross-dialectal differences in steady-state patterns of
diphthongs. He generated synthetic stimuli representing [ai], [au], [ei],
and [ou] diphthongs but with varying proportions of onset steady state,
transition, and offset steady state. Subjects who spoke northern Dutch,
British English, dialect-neutral German, and Austrian German were asked
to rate the stimuli for goodness—that is, how well they matched their own
productions. Each language and dialect investigated differed, indicating
that steady-state patterns, like the height of the glide, are a phonetic cue
that is ordinarily noncontrastive but can vary dialectally. The method that
Peeters employed would be useful for many dialectal comparisons, though
the effects of prestige on particular variants would have to be accounted
for.
Another type of cross-dialectal perception experiment involves testing
how well listeners can identify dialectal pronunciations. William Labov and
his team at the University of Pennsylvania have conducted extensive inves-
tigations of the perception of geographic dialects using unsynthesized
speech samples. Labov, Yaeger, and Steiner (1972, 135–44) describe an
experiment involving frontward-gliding forms of /u/ (as in boot), /o/ (as in
boat), and /au/ (as in out) uttered by natives of the Pamlico Sound region of
North Carolina. Isolated words with these vowels were played to subjects
from other regions of the United States, who had almost no success in
identifying them correctly. Labov (1994) and Labov and Ash (1997) report
results from their Cross-Dialectal Comprehension project. This project
used utterances of a variety of vowels by natives of Chicago, Philadelphia,
and Birmingham, Alabama (in fact, Labov, Karen, and Miller 1991, dis-
cussed above, was a spin-off project). Subjects included listeners from each

Published by Duke University Press

 American Speech

128 american speech 77.2 (2002)

of the three cities and from different ethnic groups. Stimuli were gated so
that subjects could be asked to identify excised vowels and to identify words
excised from their context, with carrier phrases, or in sentences. As ex-
pected, subjects had difficulty identifying some of the vowels and words
uttered by speakers from cities besides their own, but a rather surprising
result was that the listeners sometimes had trouble identifying vowels from
their own city.
Similar experiments have been conducted elsewhere. Flanigan and
Norris (2000) played words spoken by a southeastern Ohio native that
were excised from their context, in phrases, and in sentences to students at
different southeastern Ohio campuses, who were asked to identify them.
The listeners had a great deal of trouble identifying the words in phrases or
in isolation, especially the latter, even though most of them were from
southeastern Ohio. Traill, Ball, and Müller (1995) played individual words
uttered by speakers of South African English to listeners from the British
Isles and asked them to identify each word from three choices. Several of
the front vowels, which were undergoing a chain shift in South African
English, were prone to misidentifications. A number of experiments have
investigated cross-dialectal comprehensibility of running speech, especially
comprehensibility of African American speech and comprehension by
African Americans of the speech of other groups (e.g., Arahill 1970;
Rundell 1973), with a mixture of results.

assessing personal traits of speakers based on voices. Experiments in

which listeners rate voices for personality traits have been used extensively,
largely by social psychologists but also by sociolinguists (e.g., Underwood
1974). Although studies of listeners’ assessments of the personality traits of
speakers go back as early as Pear (1931), much of the later work has its roots
in Lambert et al.’s (1960) matched guise experiment. In this experiment,
recordings of bilinguals speaking French or English were played to subjects
who rated the speakers on various scales in order to compare their attitudes
toward English speakers and French speakers (see also Lambert 1967). The
listeners did not recognize that the same individuals were speaking English
and French in different stimuli and rated them differently for various traits
depending on which language they were speaking. Subjective reaction
studies, including matched-guise studies, are too numerous to list here.
However, a detailed review of earlier work is found in Giles and Powesland
(1975), and reviews of subsequent studies are found in Ryan (1979);
Brown and Bradshaw (1985); McMillan and Montgomery (1989, 396–
407); Bradac (1990); and Cargile et al. (1994). Ball and Giles (1988)
provide guidelines for conducting matched-guise experiments.

Published by Duke University Press

 American Speech

Sociophonetic Applications of Perception Experiments 129

Most studies of this sort have used as stimuli voices that were unedited
(except for splicing), but a few have employed synthetic modification. One
series of these studies began as an attempt to refine the methods of
Addington (1968), who had studied the subjective reactions of listeners to
utterances by trained speakers varying their voice quality, rate, and pitch.
Refinement was desirable because even trained speakers cannot control
one vocal factor precisely or without affecting other factors, while a synthe-
sizer can. Brown, Strong, and Rencher (1972, 1973, 1974); Smith et al.
(1975); and Apple, Streeter, and Krauss (1979) synthetically modified the
intonation and the speaking rate of voices and had listeners rate the
resulting stimuli subjectively on scales of benevolence and competence.
The relationship between speaking rate and competence was linear, with
faster rates being correlated with higher competence. The relationship
between speaking rate and benevolence was U-shaped; high and low rates
were correlated with low benevolence and moderate rates with high be-
nevolence. There was some controversy about how naturalistic (“ecologi-
cal”) the synthetic stimuli were; see the discussion in Brown, Giles, and
Thakerar (1985). Another study that used synthetically modified stimuli is
Bezooijen’s (1988) investigation of how listeners judge speakers’ person-
alities. She presented listeners in the Netherlands with speech from people
of various socioeconomic statuses from the same city (Nijmegen). The
speech was presented in four ways: as excerpts of the unaltered recordings;
low-pass filtered at 300 Hz, which focused listeners’ attention on prosody;
with the recordings cut digitally and randomly respliced, which focused
listeners’ attention on voice quality; and as written text. The listeners rated
the speech on various personality scales—“much education,” “strong-willed,”
and “fair.” Then the original recordings were rated on voice quality scales
by a different set of listeners, and statistical comparisons between the
personality ratings and the voice quality ratings were run. The correlations
suggested that listeners inferred a strong personality from prosody, but that
they inferred a speaker’s intellectual qualities and socioeconomic status
from segmental aspects of speech.
Fewer experiments have involved evaluations of voices for traits other
than personality features. Examples of such studies are Labov (1966),
Frazer (1987), and Plichta (2001). Labov assembled stimuli consisting of
22 sentences uttered by five women from the Lower East Side of Manhattan
and played them to other residents of the Lower East Side. The stimuli
highlighted several phonological variables that were important in New
York City. Subjects were asked to rate the speakers’ suitability for various
jobs. In general, speakers who themselves showed a high degree of stylistic
conditioning of a local variant tended to be most aware of it, as indicated by

Published by Duke University Press

 American Speech

130 american speech 77.2 (2002)

negative responses upon hearing it. Frazer (1987) played excerpts of the
reading of a story by rural Illinois natives, highlighting several variables.
Listeners, who were college students from Illinois, were asked to rate the
excerpts according to the question, “Would you be proud if a friend or
member of your family spoke this way?” The results showed that stimuli
exhibiting variants associated with Southernness were evaluated negatively.
Plichta (2001) showed European American and African American subjects
videotapes of two African Americans and two European Americans reading
a passage; the audio components of the videos were interchanged. Subjects
were asked to rate the speakers they saw according to standardness, similar-
ity to their own speech, education, and region. The most important result
was that, although European American subjects rated speakers of each
ethnicity equally highly for standardness, African American subjects tended
to rate African American speakers lower for standardness than they rated
European American speakers.

GUIDELINES FOR PERCEPTION EXPERIMENTS

As the studies discussed above indicate, sociolinguists have a wide variety of

experimental issues and designs from which to choose for perception
experiments. Preparation is the most difficult part of a perception study.
The following considerations offer a cookbook approach to conducting
perception experiments.

choice of speakers. For the vast majority of socioperceptual experiments,

real voices, rather than artificially generated ones (text-to-speech), are
desirable as the basis of the stimuli, even if the voice will be synthetically
modified. As a result, it is important to consider carefully what speakers are
chosen and what sort of speech they are asked to produce. Many experi-
ments require read speech, but reading fluency varies among speakers and
may become a source of spurious variation. Excluding speakers who are not
fluent readers may produce a bias for other variables, however. Conversa-
tional interviews may provide the most naturalistic speech, but the content
cannot be controlled and sound quality may be defective, especially for
field interviews. Another issue is how strongly marked a speaker’s dialect is.
Although it may be advantageous to use speakers who show traits of the
dialect under study especially strongly, such speakers may not typify their
speech communities, which could add a bias to the study. Conversely,
speakers who are most easily accessible to researchers, such as university
students, may be more exposed to outside influences and may not be

Published by Duke University Press

 American Speech

Sociophonetic Applications of Perception Experiments 131

especially vernacular because of their socioeconomic background. In, for

example, a dialect identification experiment, these sorts of biases could
cause listeners to rely on different variants than they ordinarily would or to
emphasize certain cues more than usual. If a stereotyped variant occurred
in a stimulus, it would quite likely divert listeners’ attention away from
other cues. While these issues may not affect every experiment, they should
be considered by experimenters.

recording equipment and conditions for signals to be modified. The

sound quality of the recordings to be used as stimuli is crucial, especially if
the recordings are to be modified with a synthesizer, because a low signal-
to-noise ratio in the recordings—caused by background noise, machine
noise, or poor reception of the speaker’s voice—can make synthesis pro-
grams inoperable. For this reason, field recordings are often unusable or
difficult to use with synthesizers, though it may be worthwhile to try anyway.
Even if the stimuli are not synthetically altered, recording problems and
variations in recording quality can affect the responses of subjects. When
possible, recordings to be synthetically modified should be made in a
soundproof or sound-treated room with a directional microphone. If such
a room is unavailable, however, recordings made in a quiet room may be
the best option. Recording conditions should be noted.
Researchers usually report the make and model of recorders, micro-
phones, sound rooms, and such as a matter of course. The make of
equipment can be a factor, but, obviously, factors such as age, wear, and
cleanliness can also affect equipment performance. For that reason some
editors ask that researchers perform a frequency response test on equip-
ment, especially earphones. Frequency response tests determine how well
the equipment handles various frequencies. Unfortunately, the equipment
needed for frequency response tests is often unavailable to sociolinguists.
Nevertheless, equipment deterioration primarily affects high-frequency
sounds such as sibilants, so it is normally a minimal factor for most of the
variables that sociolinguists study. For that reason, frequency response tests
may not be crucial for the majority of sociophonetic experiments. A
comparison of some currently available models of microphones and tape
recorders is given in Plichta and Mendoza-Denton (2001).

types of synthesis and synthesizers. Synthetic stimuli are not necessary

for all perception experiments, and for some they are not even desirable
(e.g., Labov, Karen, and Miller 1991). However, the speech synthesizer is a
valuable tool that sociolinguists have underutilized, and it is indispensable

Published by Duke University Press

 American Speech

132 american speech 77.2 (2002)

for some types of experiments. A brief description of synthesis issues

relevant to sociolinguists is given here.
Most synthesis packages available today can modify signals that are fed
into them. Many synthesizers also have text-to-speech, that is, a dictionary
of individual words that they can generate and the ability to string words
together into phrases and sentences. For sociolinguists, feeding signals into
the synthesizer is normally more useful than text-to-speech because the
input will be naturalistic. Words that are generated by the synthesizer
depend on the comprehensiveness of the synthesis program, and since
there are aspects of speech that are not completely understood, no synthe-
sis program is perfect. The problems are especially acute for strings of
several words. For any sociolinguistic study involving text-to-speech, it may
be necessary for listeners to rate the tokens for naturalness.
The two types of synthesizers seen most often today are LPC synthesiz-
ers and Klatt synthesizers. LPC synthesizers, which are analysis-synthesis
systems, are simpler in their design. Their main limitation is that LPC
(linear predictive coding) is useful only for periodic sounds, so LPC
synthesizers are poorly suited for modifying frication noise or aspiration.
Klatt synthesizers, which fall within a group called terminal analog formant
synthesizers, can produce aperiodic as well as periodic sounds. Production
of vowels and approximants can be set in either cascade mode, in which
various filters are employed sequentially, or in parallel mode, in which the
filters are employed simultaneously. Consonantal sounds are produced in
parallel mode. The cascade mode is used more often for vowels, because it
sets formant amplitudes automatically. Klatt synthesizers were designed for
text-to-speech, but some of the newer packages that contain a Klatt synthe-
sis system also include capabilities for modifying signals. For more detailed
discussions and descriptions of other kinds of synthesizers, see Klatt (1987)
and Carlson and Granström (1997).
One problem that can be encountered when only part of a running
signal is synthetically modified or when different signals are spliced to-
gether is that bursts of noise, or transience, can result at boundaries
between unmodified and modified sections or at boundaries between
spliced sections. Transience may be quite noticeable and distracting to
listeners. It can be avoided by starting and ending modified sections at
zero-crossing points, that is, points where the waveform crosses the zero
amplitude mark. Another remedy for this problem is to attenuate the signal
gradually toward boundaries.
Various synthetic treatments can be performed on stimuli. Low-pass
filtering is often used to focus subjects’ attention on prosody. Monotonization

Published by Duke University Press

 American Speech

Sociophonetic Applications of Perception Experiments 133

is used to eliminate intonation. The fundamental frequency can be modified

in less drastic ways as well, such as for modifying intonational contours or
for changing voice quality. Various mutations of temporal factors can be
used to change the rhythm or to affect the identity of segments, particularly
for voiced versus voiceless consonants. Formant values can be changed to
constant values, such as those for schwa, to eliminate vowel quality informa-
tion. Formant values can also be changed to modify particular segments—
for example, to make them sound like variants found in a different dialect
or like a different phoneme. Splicing and random reordering of portions
of the signal are occasionally used to focus subjects’ attention on voice
quality. The stimuli resulting from these operations should be checked
auditorily, because one of the hazards of such modifications is that the
synthesized tokens may not always represent accurately what they are
supposed to represent. In addition, as Bezooijen and Boves (1986) note,
some standard synthetic operations do not accomplish what they are com-
monly assumed to do. They found that low-pass filtering preserves voice
quality factors as well as prosody and that splicing and random reordering
do not eliminate all aspects of segmental quality or prosody.
There are also some other hazards with the use of synthesized speech
that have to do with experimental design, since synthetic stimuli are usually
played to subjects under controlled conditions. The first is that the content
of the stimuli can sometimes force listeners to pay more attention to a
particular perceptual cue than they ordinarily would, especially if other
cues that they usually rely on are absent. The second hazard, described by
Labov, Karen, and Miller (1991, 52–53), is that social norms biased against
variants represented in the stimuli can skew reactions from subjects. The
third hazard, mentioned in Labov (1994, 402), has to do with the fact that
subjects in experiments involving synthetic stimuli are usually asked to
label the stimuli as, for example, one word or another. At issue is the fact
that labeling, a mostly conscious act, is not the same as perception in
conversations, which is largely a subconscious act.
In spite of these pitfalls, perception experiments involving synthetic
stimuli—when used judiciously—allow researchers to impose tighter ex-
perimental controls than experiments using recordings of real speech. For
example, with a synthesizer, an experimenter can be certain that a phonetic
factor is being modified at steady increments and independently of other
factors, something that is impossible with even an adept impersonator.
Synthesizers can also provide data on some issues that no other technique
can. They are most useful for modifying vowel formants, the fundamental
frequency, and the duration of parts of the signal, and can be used for some

Published by Duke University Press

 American Speech

134 american speech 77.2 (2002)

consonantal attributes, such as voice onset time. Though sociolinguists

have used synthesizers sparingly, they represent perhaps the biggest single
aid in perception research.

listening equipment and environment. The factors noted above for

recording equipment also apply to listening equipment; the main differ-
ence is that the use of earphones is an issue. Phoneticians usually prefer to
play the stimuli to subjects through earphones. Earphones are useful for
some types of experiments, especially when the experiment involves a small
number of listeners. The use of earphones, of course, creates a sociolinguis-
tically unnatural environment, so many sociolinguists may prefer not to
employ them. In addition, earphones may be impractical for large num-
bers of subjects or when subjects are recruited in the field. It may instead be
more expedient to play stimuli to subjects in a quiet location, such as a
classroom. This approach is useful when the experiment can be linked to
class discussions of phonetics or sociolinguistics.
For tests conducted in classrooms, any loudspeaker system in good
working order is acceptable, but researchers should remember that high-
frequency sounds travel better forward than sideways, while low-frequency
sounds travel about equally well in all directions. This problem mainly
affects sibilants, but it is still important to note the placement of the
loudspeakers relative to subjects and perhaps how the subjects are seated in
the room. For any perception test conducted without earphones, whether
in a classroom or in the field, it is crucial to eliminate as much background
noise as possible. When earphones are not used, it may be important to
take note of the acoustic reflectivity, that is, the amount of echo, in the
listening environment. The volume should be set to a level comfortable for
subjects, and either the test administrator or the subjects may adjust the
volume.

screening for hearing impairment. Subjects in perception tests should

be asked whether they have any hearing impairment. Those who report
hearing impairment should ordinarily be excluded from the analysis,
though for special groups of listeners, such as elderly subjects, it may be
necessary to include subjects with some hearing loss. Speech pathologists
and phoneticians often conduct their own hearing screenings before con-
ducting a perception test, but doing so is likely to be impractical for most
sociolinguists. Inquiring whether any subjects have sinus or ear infections,
which could affect their hearing, may be advisable.

Published by Duke University Press

 American Speech

Sociophonetic Applications of Perception Experiments 135

experimental task. It almost goes without saying that the task that sub-
jects are asked to perform should be tailored to the issue being investi-
gated. Common tasks for subjects in perception experiments include (1)
judging whether two stimuli sound the same or different; (2) judging
which two stimuli out of three or more are alike; (3) selecting which
phoneme, word, ethnic group, or such out of two or more choices a
stimulus matches most closely; (4) identifying a stimulus open-endedly as a
particular word, or such, and writing it down; and (5) gauging how realis-
tic, natural, or typical of a particular group a stimulus sounds. Of course,
researchers can dream up any number of other possible tasks.
For some types of experiments, it may be important to take the
sociolinguistic environment into consideration. Lee (1971) discusses this
problem in detail, noting that experiments with carefully controlled stimuli
may not resemble natural listening conditions, especially when the dis-
course context is an important factor. He also notes various other prob-
lems, such as instances in which the variables being tested are transparent
to subjects or other uncontrolled variables influence responses. The last
problem may be remedied by synthetic manipulation. Nevertheless, Lee’s
criticisms point out the importance of careful experimental design and of
recognizing the limitations of any experiment.

presentation of stimuli. The way that stimuli are presented to subjects

should be noted. For the most part, this means describing the length of
pauses between stimuli, the number of stimuli per set, how subjects were
cued that a set of stimuli had ended, whether the subjects simply listened to
a recording or interacted with the person administering the test, and the
total length of each trial. For most tasks, the order of stimuli should be
randomized. Usually, trials should last no longer than 10–15 minutes
because fatigue by subjects sets in at that point. In addition, it often takes
subjects a little while to become adjusted to a task, so—if a long series of
stimuli is presented—results from the first few stimuli that subjects hear
should be discarded. These stimuli can be repeated later during the trial.
The number of stimuli from the beginning that are discarded should be
reported.

response form. The general manner in which subjects give responses

should be described. The traditional method is to use an answer sheet, but
computers may be more convenient if subjects are being tested one at a
time, and laptop computers may be particularly convenient in the field. For
certain experiments, subjects may even say their responses aloud while the

Published by Duke University Press

 American Speech

136 american speech 77.2 (2002)

test administrator writes them down or tape-records them. Some experi-

ments record not only the responses of subjects but their response time as
well. When response time is recorded, it is important to note whether the
time includes the time it took to play the stimulus.

FURTHER DIRECTIONS

The studies reviewed earlier illustrate the variety of research questions in

language variation to which perception experiments can be applied. Never-
theless, there is considerable room for expansion. Several of the topics
discussed above, such as the effects of gender stereotypes on perception,
have scarcely been touched. Furthermore, numerous other issues remain
for which perception experiments could be useful.
One such issue is the accuracy of impressionistic phonetic transcrip-
tion, of which some aspects have been examined (e.g., Ladefoged 1960;
Kerswill and Wright 1990; Nairn and Hurford 1995; Heuvel and Cucchiarini
2001). Both dialectologists and (as noted earlier) sociolinguists have tradi-
tionally relied heavily on impressionistic transcription in spite of its inher-
ent subjectivity. Other problems are associated with it as well. The transcrip-
tions may be difficult to interpret (see, e.g., McDavid 1981), especially
when different transcribers are involved (see the discussion of “field-
worker isoglosses” in Trudgill 1983, 38–41; see also Allen 1976, 23). A
related problem is that, as Trudgill (1983, 35–38) and Labov (1994, 74)
note, transcribers’ preconceptions affect their transcriptions: old, well-
known variants are differentiated extensively in the transcriptions, while
newer, lesser-known variants tend to be undertranscribed or ignored alto-
gether. Perception experiments could examine the extent to which differ-
ent transcribers are affected by preconceptions and by the influence of
their own native dialects. A different sort of problem has to do with the fact
that listeners compensate perceptually for processes that affect sounds in
running speech, including the coarticulatory effects of neighboring seg-
ments (Lindblom and Studdert-Kennedy 1967; Ohala 1981, 179–87; Ohala
and Feder 1994; Nábělek and Ovchinnikov 1997; Holt, Lotto, and Kluender
2000), the effects of durational variation (Lindblom and Studdert-Kennedy
1967; Janson 1979), and, probably, the effects of variation in stress. In all
likelihood, transcribers are affected by this perceptual compensation and,
as a result, may be unable to detect many effects of phonetic context. Of
course, transcribers can hear some contextual differences. Perception
experiments could shed more light on the degree to which transcribers are
able to escape their own perceptual compensation.

Published by Duke University Press

 American Speech

Sociophonetic Applications of Perception Experiments 137

This perceptual compensation, which Ohala (1981, 183) calls “recon-

structive rules,” could potentially open up another new line of perception
experiments for variationists. The studies noted above that have demon-
strated that compensation occurs avoided the issue of dialectal variation
(except insofar as Ohala 1981 relates it to diachronic variation). One may
assume that listeners’ reconstructive rules are oriented toward varieties of
language most similar to the listeners’ own speech. What happens when
listeners are subjected to unfamiliar dialects, however? Do the listeners’
reconstructive rules lead to errors in perception? Moreover, are listeners
who have extensive exposure to a different dialect able to shift to a
perceptual mode appropriate to that dialect? Another question is whether
listeners exhibit different compensatory patterns for different styles of
speech, given that speakers may adjust how carefully they articulate de-
pending on speech style (Lindblom 1990). Variationists could address
these questions experimentally.
Another perceptual issue to which variationists could contribute is the
controversy between what Strange (1989) calls the “elaborated target
model” and “dynamic specification” theories of vowel recognition; this
issue has attracted a great deal of attention from experimental phoneti-
cians recently. The elaborated target model is based on the notion that
listeners recognize vowels from steady states, while the dynamic specification
model states that listeners recognize vowels from the formant movement
found in consonant transitions. Yet another model states that diphthongal
formant movement is an important cue; see the discussion of all three
models in Pitermann (2000). Target models in various forms are the oldest
theory, and there is evidence that at least some listeners rely on steady states
(Gottfried 1984; Pitermann 2000; Sussman 2001). However, there is also
considerable evidence, in large part from experiments in which the center
of vowels is gated out of signals, that listeners can recognize vowels from
consonantal transitions (Verbrugge and Rakerd 1986; Strange 1989; Strange
and Bohn 1998) or diphthongal formant movement (Assman, Nearey, and
Hogan 1982; Andruski and Nearey 1992; Hillenbrand and Nearey 1999).
The major questions are which type of recognition is primary and which
one listeners generally rely upon in natural dialogue. Nearey (1989) sug-
gested that listeners may use all of these strategies, depending on the
situation. Gottfried (1984) reported that French speakers do not rely on
dynamic properties of a vowel when listening to French, unlike English
speakers listening to English, suggesting that a target model strategy is
more usual for French. He speculated that the reason was that French
vowels are monophthongal, whereas the gliding that typifies most varieties
of English promotes strategies that focus on dynamic properties of vowels.

Published by Duke University Press

 American Speech

138 american speech 77.2 (2002)

His finding suggests that variationists could test the models of vowel recog-
nition by comparing the behavior of speakers of mainstream dialects of
English with those showing little or no gliding, for example, varieties with
considerable substratum effects from other languages. Other experiments
could test how the different strategies operate in running speech, perhaps
by gating out parts of a vowel spoken in a context in order to create the
potential for ambiguity.
Even though some of the topics mentioned in this section and earlier
are usually studied by nonsociolinguists, especially phoneticians and psy-
chologists, variationists should not shy away from them. Cognizance of
intraspeaker linguistic differences (i.e., stylistic and register variation) and
of various interpersonal differences (e.g., those related to geography,
ethnicity, gender, and socioeconomic status) is fundamental to sociolin-
guistics, but these differences are often regarded lightly by the more
experimental branches of linguistics. Lee (1971) notes that one of the
chief shortcomings of many perceptual studies of dialectal variation is the
failure to take the sociolinguistic setting into account, yet it is this short-
coming that sociolinguists are best equipped to address. An approach to
speech perception that not only takes variation into account, but also
recognizes its usefulness for experimental design, can advance the under-
standing of numerous aspects of speech perception greatly. Furthermore,
if sociolinguists devote more attention to perception, their understanding
of language variation will deepen significantly.

NOTES

I wish to thank Paul Foulkes, John J. Ohala, and two anonymous reviewers for their
comments and suggestions regarding this project and Walt Wolfram for his sup-
port. Research for this paper was supported in part by the National Science
Foundation grant BCS 99-10224.

1. The fundamental frequency (F 0) represents the rate at which the vocal folds
vibrate. Formants (F 1, F 2 , etc.) are resonances produced by the vocal tract.
Both are normally measured in cycles per second, or Hertz (Hz).

REFERENCES

Abrams, Albert S. 1973. “Minimal Auditory Cues for Distinguishing Black from
White Talkers.” Ph.D. diss., City Univ. of New York.
Addington, David W. 1968. “The Relationship of Selected Vocal Characteristics to
Personality Perception.” Speech Monographs 35: 492–503.

Published by Duke University Press

 American Speech

Sociophonetic Applications of Perception Experiments 139

Allen, Harold B. 1976. The Linguistic Atlas of the Upper Midwest. Vol. 3, The Pronuncia-
tion. Minneapolis: Univ. of Minnesota Press.
Andruski, Jean E., and Terrance M. Nearey. 1992. “On the Sufficiency of Com-
pound Target Specification of Isolated Vowels and Vowels in /bVb/ Syllables.”
Journal of the Acoustical Society of America 91: 390–410.
Apple, William, Lynn A. Streeter, and Robert M. Krauss. 1979. “Effects of Pitch and
Speech Rate on Personal Attributions.” Journal of Personality and Social Psychol-
ogy 37: 715–27.
Arahill, Edward Joseph. 1970. “The Effect of Differing Dialects upon the Compre-
hension and Attitude of Eighth Grade Children.” Ph.D. diss., Univ. of Florida.
Assman, Peter F., Terrance M. Nearey, and John T. Hogan. 1982. “Vowel Identifica-
tion: Orthographic, Perceptual, and Acoustic Aspects.” Journal of the Acoustical
Society of America 71: 975–89.
Bailey, Guy, and Natalie Maynor. 1989. “The Divergence Controversy.” American
Speech 64: 12–39.
Ball, Peter, and Howard Giles. 1988. “Speech Style and Employment Selection: The
Matched Guise Technique.” In Doing Social Psychology: Laboratory and Field
Exercises, ed. Glynis M. Breakwell, Hugh Foot, and Robin Gilmour, 121–49.
Cambridge: Cambridge Univ. Press.
Baugh, John. 1996. “Perceptions within a Variable Paradigm: Black and White
Racial Detection and Identification Based on Speech.” In Focus on the USA, ed.
Edgar W. Schneider, 169–82. Amsterdam: Benjamins.
Bezooijen, Renée van. 1988. “The Relative Importance of Pronunciation, Prosody
and Voice Quality for the Attribution of Social Status and Personality Charac-
teristics.” In Language Attitudes in the Dutch Language Area, ed. Roeland van
Hout and Uus Knops, 85–103. Dordrecht: Foris.
Bezooijen, Renée van, and Rob van den Berg. 1999. “Word Intelligibility of
Language Varieties in the Netherlands and Flanders under Minimal Condi-
tions.” In Linguistics in the Netherlands 1999, ed. René Kager and Renée van
Bezooijen, 1–12. Amsterdam: Benjamins.
Bezooijen, Renée van, and Louis Boves. 1986. “The Effects of Low-Pass Filtering
and Random Splicing on the Perception of Speech.” Journal of Psycholinguistic
Research 15: 403–17.
Bezooijen, Renée van, and Charlotte Gooskens. 1999. “Identification of Language
Varieties: The Contribution of Different Linguistic Levels.” Journal of Language
and Social Psychology 18: 31–48.
Bezooijen, Renée van, and Jehannes Ytsma. 1999. “Accents of Dutch: Personality
Impression, Divergence, and Identifiability.” Belgian Journal of Linguistics 13:
105–29.
Bradac, James J. 1990. “Language Attitudes and Impression Formation.” In Hand-
book of Language and Social Psychology, ed. Howard Giles and W. Peter Robinson,
387–412. New York: Wiley.
Brennan, Eileen M., and John S. Brennan. 1981. “Measurements of Accent and
Attitude toward Mexican-American Speech.” Journal of Psycholinguistic Research
10: 487–501.

Published by Duke University Press

 American Speech

140 american speech 77.2 (2002)

Brennan, Eileen M., Ellen Bouchard Ryan, and William E. Dawson. 1975. “Scaling
of Apparent Accentedness by Magnitude Estimation and Sensory Modality
Matching.” Journal of Psycholinguistic Research 4: 27–36.
Browman, Catherine P., and Louis Goldstein. 1990. “Tiers in Articulatory Phonol-
ogy, with Some Implications for Casual Speech.” In Papers in Laboratory Phonol-
ogy I: Between the Grammar and Physics of Speech, ed. John Kingston and Mary
Beckman, 341–76. Cambridge: Cambridge Univ. Press.
———. 1991. “Gestural Structures: Distinctiveness, Phonological Processes, and
Historical Change.” In Modularity and the Motor Theory of Speech Perception:
Proceedings of a Conference to Honor Alvin M. Liberman, ed. Ignatius G. Mattingly
and Michael Studdert-Kennedy, 313–38. Hillsdale, N.J.: Erlbaum.
Brown, Bruce L., and Jeffrey M. Bradshaw. 1985. “Towards a Social Psychology of
Voice Variations.” In Recent Advances in Language, Communication, and Social
Psychology, ed. Howard Giles and Robert N. St. Clair, 144–81. London: Erlbaum.
Brown, Bruce L., Howard Giles, and Jitendra N. Thakerar. 1985. “Speaker Evalua-
tions as a Function of Speech Rate, Accent, and Context.” Language and
Communication 5: 207–20.
Brown, Bruce L., William J. Strong, and Alvin C. Rencher. 1972. “Acoustic Determi-
nants of Perceptions of Personality from Speech.” International Journal of the
Sociology of Language 6: 11–32.
———. 1973. “Perceptions of Personality from Speech: Effects of Manipulations of
Acoustical Parameters.” Journal of the Acoustical Society of America 54: 29–35.
———. 1974. “Fifty-four Voices from Two: The Effects of Simultaneous Manipula-
tions of Rate, Mean Fundamental Frequency and Variance of Fundamental
Frequency on Ratings of Personality from Speech.” Journal of the Acoustical
Society of America 55: 313–18.
Bryden, James D. 1968. An Acoustic and Social Dialect Analysis of Perceptual Variables in
Listener Identification and Rating of Negro Speakers. ED 022 186. Washington,
D.C.: U.S. Dept. of Health, Education, and Welfare, Office of Education,
Bureau of Research.
Buck, Joyce F. 1968. “The Effects of Negro and White Dialectal Variations upon
Attitudes of College Students.” Speech Monographs 35: 181–86.
Bush, Clara N. 1967. “Some Acoustic Parameters of Speech and Their Relation-
ships to the Perception of Dialect Differences.” TESOL Quarterly 1.3: 20–30.
Cargile, Aaron C., Howard Giles, Ellen B. Ryan, and James J. Bradac. 1994.
“Language Attitudes as a Social Process: A Conceptual Model and New Direc-
tions.” Language and Communication 14: 211–36.
Carlson, Rolf, and Björn Granström. 1997. “Speech Synthesis.” In The Handbook of
Phonetic Sciences, ed. William J. Hardcastle and John Laver, 768–88. Oxford:
Blackwell.
Clopper, Cynthia G., and David B. Pisoni. 2001. “Some Acoustic Cues for Categoriz-
ing American English Regional Dialects.” Paper presented at the 30th confer-
ence on New Ways of Analyzing Variation (NWAV 30), Raleigh, N.C., 11–14
Oct.

Published by Duke University Press

 American Speech

Sociophonetic Applications of Perception Experiments 141

Costa, Paul, and Ignatius G. Mattingly. 1981. “Production and Perception of

Phonetic Contrast during Phonetic Change.” Journal of the Acoustical Society of
America 69: S67.
Di Paolo, Marianna, and Alice Faber. 1990. “Phonation Differences and the
Phonetic Content of the Tense-Lax Contrast in Utah English.” Language
Variation and Change 2: 155–204.
Dickens, Milton, and Granville M. Sawyer. 1952. “An Experimental Comparison of
Vocal Quality among Mixed Groups of Whites and Negroes.” Southern Speech
Journal 17: 178–85.
Elman, Jeffrey L., Randy L. Diehl, and Susan E. Buchwald. 1977. “Perceptual
Switching in Bilinguals.” Journal of the Acoustical Society of America 62: 971–74.
Fant, Gunnar. 1966. “A Note on Vocal Tract Size Factors and Non-uniform F-
Pattern Scalings.” Speech Transmission Laboratory—Quarterly Progress and Status
Report (STL-QPSR) 4: 22–30.
Flanigan, Beverly Olson, and Franklin Paul Norris. 2000. “Cross-Dialectal Compre-
hension as Evidence for Boundary Mapping: Perceptions of the Speech of
Southeastern Ohio.” Language Variation and Change 12: 175–201.
Foreman, Christina. 2000. “Identification of African-American English from
Prosodic Cues.” Texas Linguistic Forum 43: 57–66.
Foulkes, Paul. 1997. “Historical Laboratory Phonology—Investigating /p/ > /f/ > /h/
Changes.” Language and Speech 40: 249–76.
Frazer, Timothy C. 1987. “Attitudes toward Regional Pronunciation.” Journal of
English Linguistics 20: 89–100.
Giles, Howard, and Peter F. Powesland. 1975. Speech Style and Social Evaluation. New
York: Academic.
Goldbeck, T. P., R. Standke, and K. R. Scherer. 1988. “Digital Techniques for Signal
Processing in Vocal Communication Research.” Psychologische Rundschau 39:
191–200.
Gooskens, Charlotte. 1997. “On the Role of Prosodic and Verbal Information in
the Perception of Dutch and English Language Varieties.” Ph.D. diss., Katholieke
Universiteit Nijmegen.
Gottfried, Terry L. 1984. “Effect of Consonant Context on the Perception of
French Vowels.” Journal of Phonetics 12: 91–114.
Graff, David, William Labov, and Wendell A. Harris. 1986. “Testing Listeners’
Reactions to Phonological Markers of Ethnic Identity: A New Method for
Sociolinguistic Research.” In Diversity and Diachrony, ed. David Sankoff, 45–58.
Amsterdam Studies in the Theory and History of Linguistic Science, ser. 4:
Current Issues in Linguistic Theory 53. Amsterdam: Benjamins.
Guenter, Joshua. 2000. “Vowels of California English before /r/, /l/, and /N/.” Ph.D.
diss., Univ. of California at Berkeley.
Guy, Gregory R. 1980. “Variation in the Group and the Individual: The Case of
Final Stop Deletion.” In Locating Language in Time and Space, ed. William
Labov, 1–36. Orlando: Academic.

Published by Duke University Press

 American Speech

142 american speech 77.2 (2002)

Haley, Kenneth. 1990. “Some Complexities of Speech Identification.” SECOL

Review 14: 101–13.
Harms, L. S. 1961. “Listener Judgments of Status Cues in Speech.” Quarterly Journal
of Speech 47: 164–68.
———. 1963. “Status Cues in Speech.” Lingua 12: 300–306.
Hawkins, Francine Dove. 1993. “Speaker Ethnic Identification: The Roles of
Speech Sample, Fundamental Frequency, Speaker and Listener Variations.”
Ph.D. diss., Univ. of Maryland at College Park.
Heselwood, Barry, and Louise McChrystal. 2000. “Gender, Accent Features and
Voicing in Panjabi-English Bilingual Children.” Leeds Working Papers in Linguis-
tics 8: 45–70.
Heuvel, Henk van den, and Catia Cucchiarini. 2001. “/r/-Deletion in Dutch: Rumours
or Reality?” In ‘r-atics: Sociolinguistic, Phonetic and Phonological Characteristics of
/r/, ed. Hans Van de Velde and Roeland van Hout, 185–98. Etudes et Travaux.
Brussels: Institut des Langues Vivantes et de Phonétique, Université Libre de
Bruxelles.
Hibler, Madge Beatrice. 1960. “A Comparative Study of Speech Patterns of Se-
lected Negro and White Kindergarten Children.” Ph.D. diss., Univ. of Mary-
land at College Park.
Hillenbrand, James M., and Terrance M. Nearey. 1999. “Identification of Resynthe-
sized /hVd/ Utterances: Effects of Formant Contour.” Journal of the Acoustical
Society of America 105: 3509–23.
Holt, Lori L., Andrew J. Lotto, and Keith R. Kluender. 2000. “Neighboring Spectral
Content Influences Vowel Identification.” Journal of the Acoustical Society of
America 108: 710–22.
Hombert, Jean-Marie, John J. Ohala, and William G. Ewan. 1979. “Phonetic
Explanations for the Development of Tones.” Language 55: 37–58.
Irwin, Ruth Beckey. 1977. “Judgments of Vocal Quality, Speech Fluency, and
Confidence of Southern Black and White Speakers.” Language and Speech 20:
261–66.
Janson, Tore. 1979. “Vowel Duration, Vowel Quality, and Perceptual Compensa-
tion.” Journal of Phonetics 7: 93–103.
———. 1983. “Sound Change in Perception and Production.” Language 59: 18–34.
———. 1986. “Sound Change in Perception: An Experiment.” In Experimental
Phonology, ed. John J. Ohala and Jeri J. Jaeger, 253–60. Orlando: Academic.
Janson, Tore, and Richard Schulman. 1983. “Non-distinctive Features and Their
Use.” Journal of Linguistics 19: 321–36.
Johnson, Keith, Elizabeth A. Strand, and Mariapaola D’Imperio. 1999. “Auditory-
Visual Integration of Talker Gender in Vowel Perception.” Journal of Phonetics
27: 359–84.
Jonasson, Jan. 1971. “Perceptual Similarity and Articulatory Reinterpretation as a
Source of Phonological Innovation.” Papers from the Institute of Linguistics,
University of Stockholm 8: 30–42.

Published by Duke University Press

 American Speech

Sociophonetic Applications of Perception Experiments 143

Kerswill, Paul, and Susan Wright. 1990. “The Validity of Phonetic Transcription:
Limitations of a Sociolinguistic Research Tool.” Language Variation and Change
2: 255–75.
Klatt, Dennis H. 1987. “Review of Text-to-Speech Conversion for English.” Journal
of the Acoustical Society of America 82: 737–93.
Koutstaal, Cornelis W., and Faith L. Jackson. 1971. “Race Identification on the
Basis of Biased Speech Samples.” Ohio Journal of Speech and Hearing 6: 48–51.
Labov, William. 1966. The Social Stratification of English in New York City. Washington,
D.C.: Center for Applied Linguistics.
———. 1972. Language in the Inner City: Studies in the Black English Vernacular.
Philadelphia: Univ. of Pennsylvania Press.
———. 1994. Principles of Linguistic Change. Vol. 1, Internal Factors. Language in
Society 20. Oxford: Blackwell.
Labov, William, and Sharon Ash. 1997. “Understanding Birmingham.” In Language
Variety in the South Revisited, ed. Cynthia Bernstein, Thomas Nunnally, and
Robin Sabino, 508–73. Tuscaloosa: Univ. of Alabama Press.
Labov, William, Mark Karen, and Corey Miller. 1991. “Near-Mergers and the
Suspension of Phonemic Contrast.” Language Variation and Change 3: 33–74.
Labov, William, Malcah Yaeger, and Richard Steiner. 1972. A Quantitative Study of
Sound Change in Progress. Philadelphia: U.S. Regional Survey.
Ladefoged, Peter. 1960. “The Value of Phonetic Statements.” Language 36: 387–
96.
Lambert, Wallace E. 1967. “A Social Psychology of Bilingualism.” Journal of Social
Issues 23: 91–108.
Lambert, W[allace] E., R. C. Hodgsen, R. D. Gardner, and S. Fillenbaum. 1960.
“Evaluational Reaction to Spoken Language.” Journal of Abnormal and Social
Psychology 60: 44–51.
Larson, Vernon S., and Carolyn H. Larson. 1966. “Reactions to Pronunciation.” In
Communication Barriers for the Culturally Deprived, ed. Raven I. McDavid, Jr., and
William M. Austin. Cooperative Research Project No. 2107. Washington, D.C.:
U.S. Office of Education.
Lass, Norman J., Celest A. Almerino, Laurie F. Jordan, and Jayne M. Walsh. 1980.
“The Effect of Filtered Speech on Speaker Race and Sex Identifications.”
Journal of Phonetics 8: 101–12.
Lass, Norman J., Pamela J. Mertz, and Karen L. Kimmel. 1978. “The Effect of
Temporal Speech Alterations on Speaker Race and Sex Identifications.” Lan-
guage and Speech 21: 279–90.
Lass, Norman J., John E. Tecca, Robert A. Mancuso, and Wanda I. Black. 1979.
“The Effect of Phonetic Complexity on Speaker Race and Sex Identifications.”
Journal of Phonetics 7: 105–18.
Lee, Richard R. 1971. “Dialect Perception: A Critical Review and Re-evaluation.”
Quarterly Journal of Speech 57: 410–17.
Lindblom, Björn. 1980. “The Goal of Phonetics, Its Unification and Application.”
Phonetica 37: 7–26.

Published by Duke University Press

 American Speech

144 american speech 77.2 (2002)

———. 1990. “Explaining Phonetic Variation: A Sketch of the H&H Theory.” In

Speech Production and Speech Modelling, ed. W. J. Hardcastle and A. Marchal,
403–39. Dordrecht: Kluwer.
Lindblom, Björn, and Michael Studdert-Kennedy. 1967. “On the Role of Formant
Transitions in Vowel Recognition.” Journal of the Acoustical Society of America 42:
830–43.
Malderez, Isabelle. 1995. “The Use of a Category-Perception Test in the Study of
Ongoing Sound Change.” In Proceedings of the 13th International Congress of
Phonetic Sciences, ICPhS 95: Stockholm, Sweden, 13–19 August 1995, ed. Kjell
Elenius and Peter Branderud, 684–87. Stockholm: Royal Institute of Technol-
ogy and Stockholm Univ.
McDavid, Raven I., Jr. 1981. “Low-Back Vowels in Providence: A Note in Structural
Dialectology.” Journal of English Linguistics 15: 21–29.
McGurk, Harry, and John MacDonald. 1976. “Hearing Lips and Seeing Voices.”
Nature 264: 746–48.
McMillan, James B., and Michael B. Montgomery. 1989. Annotated Bibliography of
Southern American English. Tuscaloosa: Univ. of Alabama Press.
Munro, Murray J., Tracey M. Derwing, and James E. Flege. 1999. “Canadians in
Alabama: A Perceptual Study of Dialect Acquisition in Adults.” Journal of
Phonetics 27: 385–403.
Nábělek, Anna K., and Alexandra Ovchinnikov. 1997. “Perception of Nonlinear
and Linear Formant Trajectories.” Journal of the Acoustical Society of America 101:
488–97.
Nairn, Moray J., and James R. Hurford. 1995. “The Effect of Context on the
Transcription of Vowel Quality.” In Studies in General and English Phonetics:
Essays in Honour of Professor J. D. O’Connor, ed. Jack Windsor Lewis, 96–120.
London: Routledge.
Nearey, Terrance M. 1989. “Static, Dynamic, and Relational Properties in Vowel
Perception.” Journal of the Acoustical Society of America 85: 2088–113.
Niedzielski, Nancy. 1999. “The Effect of Social Information on the Perception of
Sociolinguistic Variables.” Journal of Language and Social Psychology 18: 62–85.
Ohala, John J. 1981. “The Listener as a Source of Sound Change.” In Papers from the
Parasession on Language and Behavior, Chicago Linguistic Society, May 1–2, 1981,
ed. Carrie S. Masek, Roberta A. Hendrick, and Mary Frances Miller, 178–203.
Chicago: Chicago Linguistic Society.
———. 1985. “Linguistics and Automatic Processing of Speech.” In New Systems and
Architectures for Automatic Speech Recognition and Synthesis, ed. Renato De Mori
and Ching Y. Suen, 447–75. Berlin: Springer.
———. 1989. “Sound Change Is Drawn from a Pool of Synchronic Variation.” In
Language Change: Contributions to the Study of Its Causes, ed. Leiv Egil Breivik and
Ernst Håkon Jahr, 173–98. Berlin: de Gruyter.
———. 1993. “The Phonetics of Sound Change.” In Historical Linguistics: Problems
and Perspectives, ed. Charles Jones, 237–78. Longman Linguistics Library.
London: Longman.

Published by Duke University Press

 American Speech

Sociophonetic Applications of Perception Experiments 145

Ohala, John J., and Deborah Feder. 1994. “Listeners’ Normalization of Vowel
Quality Is Influenced by ‘Restored’ Consonantal Context.” Phonetica 51: 111–
18.
Pear, Tom Hatherley. 1931. Voice and Personality as Applied to Radio Broadcasting. New
York: Wiley.
Peeters, Wilhelmus Johannes Maria. 1991. “Diphthong Dynamics: A Cross-Linguis-
tic Perceptual Analysis of Temporal Patterns in Dutch, English, and German.”
Ph.D. diss., Rijksuniversiteit te Utrecht.
Pitermann, Michael. 2000. “Effect of Speaking Rate and Contrastive Stress on
Formant Dynamics and Vowel Perception.” Journal of the Acoustical Society of
America 107: 3425–37.
Plichta, Bartek. 2001. “Hearing Faces: The Effects of Ethnicity on Speech Percep-
tion.” Paper presented at the 30th conference on New Ways of Analyzing
Variation (NWAV 30), Raleigh, N.C., 11–14 Oct. Available from http://
[Link].
Plichta, Bartek, and Norma Mendoza-Denton. 2001. “Acquisition, Processing, and
Analysis of Acoustic Speech Signals.” Paper presented at the 30th conference
on New Ways of Analyzing Variation (NWAV 30), Raleigh, N.C., 11–14 Oct.
Available from [Link]
Preston, Dennis R. 1986. “Five Visions of America.” Language in Society 15: 221–40.
———. 1993a. “Folk Dialectology.” In American Dialect Research, ed. Dennis R.
Preston, 333–77. Amsterdam: Benjamins.
———. 1993b. “Two Heartland Perceptions of Language Variety.” In “Heartland”
English: Variation and Transition in the American Midwest, ed. Timothy C. Frazer,
23–47. Tuscaloosa: Univ. of Alabama Press.
———. 1996. “Where the Worst English Is Spoken.” In Focus on the USA, ed. Edgar
W. Schneider, 297–360. Varieties of English around the World, General Series,
vol. 16. Amsterdam: Benjamins.
———, ed. 1999. Handbook of Perceptual Dialectology. Vol. 1. Amsterdam: Benjamins.
Purnell, Thomas, William Idsardi, and John Baugh. 1999. “Perceptual and Pho-
netic Experiments on American English Dialect Identification.” Journal of
Language and Social Psychology 18: 10–30.
Roberts, Margaret M. 1966. “The Pronunciation of Vowels in Negro Speech.” Ph.D.
diss., Ohio State Univ.
Rundell, Edward Eugene. 1973. “Studies of the Comprehension of Black English.”
Ph.D. diss., Univ. of Texas at Austin.
Ryan, Ellen Bouchard. 1979. “Why Do Low-Prestige Varieties Persist?” In Language
and Social Psychology, ed. Howard Giles and Robert N. St. Clair, 145–57. Oxford:
Blackwell.
Sebastian, Richard J., and Ellen Bouchard Ryan. 1985. “Speech Cues and Social
Evaluation: Markers of Ethnicity, Social Class, and Age.” In Recent Advances in
Language, Communication, and Social Psychology, ed. Howard Giles and Robert N.
St. Clair, 112–43. London: Erlbaum.

Published by Duke University Press

 American Speech

146 american speech 77.2 (2002)

Shuy, Roger W. 1970. “Subjective Judgments in Sociolinguistic Analysis.” In Report

of the 20th Annual Round Table Meeting on Linguistics and Language Studies, ed.
James E. Alatis, 175–88. Washington, D.C.: Georgetown Univ. Press.
Shuy, Roger W., Joan C. Baratz, and Walter A. Wolfram. 1969. Sociolinguistic Factors
in Speech Identification. National Institute of Mental Health Research Project
No. MH 15048-01. Washington, D.C.: Center for Applied Linguistics.
Smith, Bruce L., Bruce L. Brown, William J. Strong, and Alvin C. Rencher. 1975.
“Effects of Speech Rate on Personality Perception.” Language and Speech 18:
145–52.
Stephan, Christoph. 1997. “The Unknown Englishes? Testing German Students’
Ability to Identify Varieties of English.” In Englishes around the World: Studies in
Honour of Manfred Görlach, vol. 1, General Studies, British Isles, North America, ed.
Edgar W. Schneider, 93–108. Varieties of English around the World, General
Series 18. Amsterdam: Benjamins.
Strand, Elizabeth A. 1999. “Uncovering the Role of Gender Stereotypes in Speech
Perception.” Journal of Language and Social Psychology 18: 86–100.
Strange, Winifred. 1989. “Evolving Theories of Speech Perception.” Journal of the
Acoustical Society of America 85: 2081–87.
Strange, Winifred, and Ocke-Schwen Bohn. 1998. “Dynamic Specification of
Coarticulated German Vowels: Perceptual and Acoustical Studies.” Journal of
the Acoustical Society of America 104: 488–504.
Stroud, Robert Vernon. 1956. “A Study of the Relations between Social Distance
and Speech Differences of White and Negro High School Students of Dayton,
Ohio.” Master’s thesis, Bowling Green State Univ.
Surprenant, Aimée, and Louis Goldstein. 1998. “The Perception of Speech Ges-
tures.” Journal of the Acoustical Society of America 104: 518–29.
Sussman, Joan E. 2001. “Vowel Perception by Adults and Children with Normal
Language and Specific Language Impairment: Based on Steady States or
Transitions?” Journal of the Acoustical Society of America 109: 1173–80.
Thomas, Erik R. 2000. “Spectral Differences in /ai/ Offsets Conditioned by Voicing
of the Following Consonant.” Journal of Phonetics 28: 1–25.
Thomas, Erik R., and Jeffrey Reaser. 2002. “Perceptual Cues Used for Ethnic
Labeling of Hyde County, North Carolina, Voices.” Paper presented at the
annual meeting of the American Dialect Society, San Francisco, 3–5 Jan.
Traill, Anthony, Martin J. Ball, and Nicole Müller. 1995. “Perceptual Confusion
between South African and British English Vowels.” In Proceedings of the 13th
International Congress of Phonetic Sciences, ICPhS 95: Stockholm, Sweden, 13–19
August 1995, ed. Kjell Elenius and Peter Branderud, 620–23. Stockholm:
Royal Institute of Technology and Stockholm Univ.
Trent, Sonja A. 1995. “Voice Quality: Listener Identification of African-American
versus Caucasian Speakers.” Journal of the Acoustical Society of America 98: 2936.
Trudgill, Peter. 1983. On Dialect: Social and Geographical Perspectives. New York: New
York Univ. Press.

Published by Duke University Press

 American Speech

Sociophonetic Applications of Perception Experiments 147

Tucker, G. Richard, and Wallace E. Lambert. 1969. “White and Negro Listeners’
Reactions to Various American-English Dialects.” Social Forces 47: 463–68.
Underwood, Gary N. 1974. “How You Sound to an Arkansawyer.” American Speech
49: 208–15.
Verbrugge, Robert R., and Brad Rakerd. 1986. “Evidence of Talker-Independent
Information for Vowels.” Language and Speech 29: 39–57.
Walton, Julie H., and Robert F. Orlikoff. 1994. “Speaker Race Identification from
Acoustic Cues in the Vocal Signal.” Journal of Speech and Hearing Research 37:
738–45.
Williams, Angie, Peter Garrett, and Nikolas Coupland. 1999. “Dialect Recogni-
tion.” In Preston, 345–58.
Willis, Clodius. 1972. “Perception of Vowel Phonemes in Fort Erie, Ontario,
Canada, and Buffalo, New York: An Application of Synthetic Vowel Categoriza-
tion Tests to Dialectology.” Journal of Speech and Hearing Research 15: 246–55.
Wolfram, Walt. 1969. A Sociolinguistic Description of Detroit Negro Speech. Washington,
D.C.: Center for Applied Linguistics.
———. 2000. “On Constructing Vernacular Dialect Norms.” In Chicago Linguistic
Society 36: The Panels, ed. Arika Okrent and John Boyle, 335–58. Chicago:
Chicago Linguistic Society.
Wolfram, Walt, Kirk Hazen, and Natalie Schilling-Estes. 1999. Dialect Change and
Maintenance on the Outer Banks. Publication of the American Dialect Society 81.
Tuscaloosa: Univ. of Alabama Press.
Yang, Byunggon. 1992. “An Acoustical Study of Korean Monophthongs Produced
by Male and Female Speakers.” Journal of the Acoustical Society of America 91:
2280–83.

Published by Duke University Press