THE INFLUENCE OF BLACK
NON-STANDARD ENGLISH ON PERSON PERCEPTION
JOHN ALBERT CRAFT
A DISSERTATION PRESENTED TO THE GRADUATE COUNCIL
OF THE UNIVERSITY OF FLORIDA IN
PARTIAL FULFILLMENT OF THE REQUIREMENTS
FOR THE DEGREE OF DOCTOR OF PHILOSOPHY
UNIVERSITY OF FLORIDA
TABLE OF CONTENTS
ABSTRACT .................... ... ................ ... ii
Black Non-standard Dialect Defined.........
General Person Perception Research.........
Other Factors Influencing Perception.......
General Studies of Dialect.................
Studies of Black Non-standard Dialect......
Dialect and Scholastic Success.............
Purpose of the Study.......................
........... ........ .
THREE RESULTS. ...................., ............... 48
1 SPEAKER SCRIPTS......
2 RATING INSTRUMENT....
.................... .... 67
Abstract of Dissertation Presented to the Graduate Council
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Doctor of Philosophy
THE INFLUENCE OF BLACK
NON-STANDARD ENGLISH ON PERSON PERCEPTION
JOHN ALBERT CRAFT
Chairman: Dr. Norman N. Markel
Major Department: Speech
The need for and merits of bi-dialectalism are currently
an area of some controversy. Previous investigation suggests
that speakers of sitgmatized dialects of English in the U.S.A.
culture are at an evaluative disadvantage when compared with
speakers of General American dialect. These same studies
have consistently found that Black Non-standard dialect
is associated with very negative stereotypes. The studies
which have arrived at these conclusions are less than
compelling, however, due to a number of common methodological
This study was designed to test, under more stringent
conditions, the validity of the findings that BNS speakers
are at a disadvantage when compared to General American
speakers. In addition, an attempt was made to separate the
grammatical and phonological channels of speech in order to
identify the basis for speech linked stereotypes.
Spontaneous speech samples were collected from white
and black males of similar education and age describing a
non-salient landscape. These speech samples were then
transcribed. The scripts were then exchanged among groups
and re-recorded in an attempt to separate the two major
dimensions of speech: grammatical and phonological. The
16 speech samples were divided into 32, 15 second segments
(2 from each condition) and randomly ordered on a master
tape. This tape was then independently judged for dialect
representativeness by three dialect experts. Twenty-four
white female and twenty-four white male graduate business
students then rated the speech segments in terms of the
evaluative dimension of the semantic differential.
The results of a 3-factor ANOVA supported the hypothesis
that BNS speakers are negatively perceived as compared with
GA speakers. No difference was noted between male and female
raters. BNS speakers were not more negatively perceived
when using GA grammar. The major finding of this research
is a significant interaction between the phonological and
grammatical aspects of dialect.
Does it make a difference which dialect of American
English a person employs? Is there a preferred dialect
style? Are there depreciated or negatively perceived
dialects of American English? If so in which situations?
These and related questions have been the focus of
communication research since the early days of radio. On
a common sense level we are all aware that skin color, hair
texture, and physiological structure serve to distinguish
large groups of individuals. In a general way we are also
aware that these physical features often serve as clues to a
broad range of (physical traits and personality charac-
teristics) qualities, some factual and some imagined.
Research into listener reactions to various speech
styles takes notice of another distinguishing characteristic
that can label a speaker as readily as the traits mentioned
above. Recently articles and studies have delt with dialect
in general and the use of Black Non-standard Dialect
James Sledd, in a 1969 article, felt that the imposition
of "white" dialect on Black children is a cause of Black
"self hate." He rejected the idea that the use of Black Non-
standard English (hereinafter BNS) limits the upward mobility
of those who employ it. He concluded by suggesting that
it is impossible to effectively teach the Black student
"bi-dialectalism" (fluency in BNS as well as standard English)
because of the hostility and rejecting white dialect evokes
in the Black student's peer groupss. Sledd's views while
lacking in objectivity, still serve to indicate the strength
of emotion that is attached to a person's perception of a
style of speech.
Other studies (Belcher, 1977; Marino, 1977; Holt, 1978;
Quay and Mathews, 1977; Fetcher, 1978) have concluded that
the use of BNS to some degree inhibits the learning process of
the Black child. It is not the design or intention of this
study to promote or sponsor an argument in favor of any
position with regard to the use of BNS. This study is
designed to determine the affective consequences (if any) of
the use of BNS in the southeastern U.S.A. culture. This will
be accomplished by determine how people (Black, white, male,
female) react to the use of this dialectal style in comparison
with standard English, General American Dialect). Research
reviewed in this study suggests that Black Non-standard Dialect
speakers are at a perceptual disadvantage when compared with
General American Dialect speakers (hereinafter GA). This
cannot be stated more definitely, however, due to several
methodological flaws (present in most research in the field)
which may dilute or confound the generalizability of the
extant experimental results.
BNS: A DESCRIPTION.
Before these previous studies are reviewed it would be
advisable to describe what BNS is. Proponents of the
unstigmatized use of BNS contend that it should be
recognized as a legitimate speech style. Smith, in his
book (Trans Racial Communication, 1973), states that Black
English has its roots in Black Africa. According to
Smith, Black English borrowed English vocabulary and
retained African grammatical elements. He contends that
Black English is not incorrect English for it has its own
regularized grammar and predictable grammatical pattern.
Rich (1974) also feels that BNS is a legitimate speech
style which represents the Black cultural experience in
the United States.
Even so, Rich notes that dialects and accents pose
two problems, First, they pose a barrier to simple under-
standing insofar as they disrupt information transfer.
Second, she notes that negative reactions to certain dialects
can cause listeners to "turn off" a speaker, and to
disregard the content of his message.
Carter (1977) notes that Black English employs
alternate verb forms, deviate phonology, and alternate
communication styles such as "shucking" and "rapping."
Carter reports that the function of these alternate styles
is to allow users an acceptable level of self respect in
"no-win" situations, involving the power group. These
alternate styles developed as a reflection of Black social
and cultural experiences. In terms of operationalizing
(BNS),Hoover (1978) characterized Black English as differing
from Anglo Saxon English (GA) in the following ways:
(1) weak final consonants, unless grammar is affected
(not dropping the ed to signify past tense); (2) simplified
consonant clusters; and (3) general "1" and "r" lessness.
A general set of features which distinguishes black
from white dialects is as follows (Bailey, 1965); (Stewart,
1. Clear /1/ in prevocalic position
2. Bilabial (sic) aspirants, /0/ and /e/, replaced
/b/ and /f/, respectively.
3. /b/, /d/, and /g/ as fully voiced imploded stops
(positions not specified by authors)
4. Neutralization of final /m/ and /n/, with a
resulting nasalization of the preceding vowel
5. /f/ and /v/ as bilabial spirants
6. Absence of the copula
7. Absence of the possessive morpheme
8. Was used for the completed past, and been
for the past, up to and including the present
9. Be for durative aspect or future, and gonna
for the conditional future
10. They and you for their and your, respectively.
Before speaking directly to the question of the evaluational
consequences of BNS, a brief review of general research in
the person perception area is appropriate.
General Person Perception Research.
The earliest vocal characteristic research was radio
related. Research by Cantril and Allport (1943), Fay and
Middleton (1939) and Taylor (1934) found listeners to be
only moderately accurate in judging the sex and age of a
speaker and somewhat less accurate in inferring occupation,
height, weight and appearance. Is it interesting to note
that no characteristic was judged correctly all of the time,
and furthermore, judges tended to agree more with each other
than with actual speaker characteristics. This general
agreement among perceptions, combined with little better than
chance accuracy, suggests that the judges were responding to
stereotyped perceptions of certain vocal traits, rather than
genuine speech based clues of personality and physical
characteristics. Later studies (Licklider and Miller, 1951;
Kramer, 1963) also found that evaluations of personality,
based solely on voice, have little or no accuracy.
Research with psychiatric patients conducted by Markel
et al. (Markel, Meisels and Houck, 1964; Markel, 1969) has
shown some success in identifying speech traits of
schizophrenic individuals that accurately predicted aspects
of personality. A later study by Markel (1972) with non-
psychiatric patients found that trained judges could
isolate some general personality traits with accuracy
from brief, spontaneous, taped passages. In general
though, research suggests that in most situations
very little can be accurately inferred about an individual
based solely on his vocal style. This is not to say, however,
that nothing is suggested about an individual by his voice.
On the contrary, many studies have identified strong
stereotypical associations with certain vocal characteristics.
Addington, in a (1968) study, examined the premise that
certain voices are stereotypes; they definitely impress
listeners as being the voice of persons who might be
classified (according to one or another personality types).
Addington contends that our voices elicit stereotypes of
personality which may or may not be consistent with more
valid assessment techniques. He investigated the relation-
ships among nine vocal characteristics and forty
personality characteristics. Using a methodology typical
of most research in the field, Addington recorded male and
female speakers imitating seven voice qualities (breathy,
tense, thin, flat, throaty, nasal, orotund, rate and pitch
variety) while reading a standardized (non-spontaneous)
passage. The recorded samples were first judged by experts
for validity and then evaluated by college freshmen.
The experimental subjects were found to ascribe specific
personality traits from certain vocal qualities. Further,
these noted ascriptions tended to be uniform, supporting
the premise that the evaluators shared stereotypes of
certain speech cues. In short, it was found that vocal
manipulations do affect the nature of personality
perception. Each of Addington's vocal characteristics
(with one exception) was effective in altering the listener's
image of the speaker.
Although the 40 perceptual categories and 9 vocal
characteristics are not exhaustive, Addington's results
suggest that many speech based variables such as word
choice, intensity, disfluencies and dialect have the
potential to color a listener's perception of a speaker.
Other Factors Influencing Perception.
Intuitively we recognize that the actual words an
individual employs can potentially influence the way he or
she is perceived. Language intensity is one area which
research has determined can alter a speaker's perception
by an audience. Bowers (1964) determined that the use of
intense terms (terms that strongly suggest emotions, death,
sex, etc..), multisyllable words that connote strength and
indicate direction of attitude, favorably influence
perceptions of a speaker. McEwen and Greenberg (1970)
in an investigation of intensity and credibility found that
sources employing highly intense terms were perceived as
more dynamic than sources using less intense terms. Another
area of non-dialectally based variation in person perception
is vocalized non-fluencies. Studies of the relationship
between speech fluency and source perception are few and
have generally been concerned with some aspect of credibility.
Miller and Hewgill in a (1964) study examined the
influence of vocalized non-fluency on speaker credibility.
Non-fluencies were operationalized in this study as vocalized
pauses and repetition. The authors recorded several versions
of a persuasive message with various levels of non-fluency
present. These taped passages were evaluated by experimental
subjects with regard to three levels of source credibility;
competence, trustworthiness and dynamism. The findings of
this study suggested that as the frequency of non-fluencies
increases in a speech, audience ratings of perceived source
credibility decrease. Specifically, audience ratings of a
speaker's trustworthiness were most negatively correlated
with the incidence of vocalized non-fluencies. A similar
study by Sereno and Hawkins (1967) investigated a number
of other non-fluencies. In this investigation the authors
manipulated five categories of non-fluency: (1) The "ah"
sound or vocalized pause, (2) Sentence correction- a correction
in the choice of words, (3) Stuttering, i.e., the serial
superfluous repetition of words, and (5) Tongue-slip correction,
a correction of an unintended sound. All five categories
were simulated by a speaker in a persuasive speech about
Black Muslims. Several tape recordings were made of the
speech with various levels of non-fluency present. The
recordings were presented to experimental subjects. The
speakers were evaluated by these subjects over a number
of dimensions designed to measure the persuasiveness of the
speeches and perceptions of source credibility.
It was found that varying amounts of non-fluencies
did not diminish the persuasiveness of the speech. The
perceived credibility of the speaker was, however, adversely
affected by increases in vocalized non-fluency. Specifically,
non-fluencies were found to negatively correlate with
perceptions of a speakers dynamism and competence.
The foregoing reviews represent an overview of studies
that have dealt with the general, non-dialectal aspects
of a person's speech style. Judging from these studies, it
is obvious that the way we speak can and does affect the
way we are perceived.
The studies that follow investigated the variables of
status, background, and dialect with regard to their influence
on a speaker's perception.
An early study by Putnam and O'Hern (1955) investigated
the relationships between speech style and status. They
recorded samples of speech from twelve speakers of different
social and educational backgrounds. University students
were asked to judge the speakers' status and background after
listening to the tapes. Putnam and O'Hern found that subjects
were able to somewhat accurately (mean correlation = .80)
discern the speakers' background and status.
Harms (1961) in a similar study attempted to (1) obtain
subjective listener judgments of speaker status and compare
them with an objective status index; (2) determine the
perceived credibility of speakers from different status groups,
and (3) determine how accurate listeners could be in judging
status from voice. Nine speakers provided the stimulus
material for this experiment. All were, (1) male, (2) 30-50
years of age, and (3) had lived all their life in the American
Midwest. The speakers were classified into two status groups
based on education and occupation. Three were classified as
high status. They held advanced degrees (DDS,PhD) and had
prestige occupations. Those classified as middle class (3)
held middle status occupations and had completed high school
and/or one year of college. The final three speakers were
classified as lower class. This group had eighth grade
educations and held unskilled jobs.
Each speaker made a 40-60 second field tape recording.
The material was elicited by having each speaker respond to
questions and statements on cards, such as "How are you?",
"Ask for the time," etc.. The recorded conversations were
non-socially salient and were similar to the kind of talk
usually associated with introductory situations. One
hundred and eighty non-college adults living in Columbus,
Ohio, served as subjects for this experiment. These listeners
were classified according to status in the same manner as the
speaker. Each speaker was heard by 60 listeners (20 from
each status group); listening was done in settings as diverse
as firehouses, living rooms and church basements.
Most listeners reported making their judgments of both
status and credibility after hearing only 10 or 15 seconds
of speech, even though the recorded samples ran 40-60 seconds.
Harms felt that perhaps a listener notices pronunciation
and other stereotyped features most readily after he responds
to some yet-to-be identified microscopic speech cues. Harms
stated his findings as follows:
-(1) Listeners of all statuses on hearing short
voice recordings assign mean ratings which
group speakers in accordance with their
objectively measured status; listeners
distinguish among speakers according to status;
(2) Listeners of various statuses agree on the
amount of credibility they assign to speakers
of various statuses; listeners find high status
speakers to be most credible and low status
speakers to be the least credible;
(3) The correlation between ratings of status
and credibility is significant statistically,
but cannot be said to be high socially." p 168
It is interesting to note Harm's speculation that some
microscopic speech variable could be responsible for the
varying perceptions of speakers by listeners. It is
possible that intonation or some phonological speech
difference could serve to label a given speaker. Obviously
the listeners were responding to some consensually held speech
based perceptions that transcended their own status group.
The foregoing studies suggest that status (actual or
stereotyped) to some extent can be judged from non-accented
speech. In the Harms study the speakers and listeners were
all from Ohio, an area which is characterized by accentlesss"
General American Dialect. Common sense and experience tell
us that spoken language can be a identifying feature of
members of a national or cultural group. Individuals in
these regional or national groups are easily identifiable by
their accent or dialectal speech patterns alone.
Some General Studies of Dialect and Perception.
Lambert, Hodgson and Gardner (1960), in the first of a
series of studies on evaluation reactions to spoken languages,
attempted to evaluate the effect dialect has on person
perception. Lambert theorized that because the use of the
language is one aspect of behavior common to a variety of
individuals, hearing the language is likely to arouse mainly
generalized or stereotyped characteristics of the group.
This suggests that when a radio broadcast of an international
meeting or passages of a foreign language are encountered, a
person's evaluational reactions to that communication are based,
in part, on a generalized attitudinal reaction to the group that
uses the particular language or dialect.
The purpose of Lambert's study was to determine the
significance spoken language has for listeners by analyzing
their evaluational reactions. The study was conducted in
the Canadian Province of Quebec, an area charged with
rivalry between French and English speaking groups.
Lambert felt that this schism is as socially significant
for residents of Quebec as that of the North and South
for southern residents of the United States.
To test their theory, Lambert and associates, translated
a 2.5 minute passage of French prose (basically non-salient)
into fluent English. This passage was then recorded in
French and English by four male bilinguals. Each subject
recorded in both English and French. Two other males
recorded French and English versions as fillers. The ten
taped passages were then exposed to both French and English
speaking natives. The subjects were not informed that the
speakers were bilingual. The subjects were then asked to
complete a response sheet for each voice which directed them
to rate each of 14 traits on six point scales. These scales
rated physical and personality traits. English speaking
subjects, as expected, showed more favorableness to members
of their linguistic group. Surprisingly, French subjects
also rated English speakers more favorably.
French subjects perceived English speakers as having
more favorable physical and personality traits than speakers
of their own language. Lambert felt that these findings
are similar to other studies (Adelson (1953); Sarnoff (1951);
Steckler (1957)) which have noted the tendency of minority
groups (as the French are in this case) to sometimes adopt
the stereotyped values of the majority groups. The French
subjects may regard themselves as members of an inferior
group. Since the French and English versions were recorded
by the same subject the listeners had to have based their
evaluations on community wide stereotypes of both groups.
In a similarly oriented (1962) study, Anisfeld, Boco
and Lambert investigated evaluational reactions to accented
English speech. This study was intended to continue on
the findings of the aforementioned Quebec Study. Lambert
and associates attempted to extend the implications and
rationale of his previous study to Jewish people who speak
English with a distinctive accent. Specifically, the study
was undertaken to find out whether Jewish and English
subjects will evaluate differently Jewish and English speech
guises when spoken by the same person.
As in the Quebec experiment, "bilingual" speakers recorded
a non-salient passage in both English and Jewish accented
English. English and Jewish accented English speaking
subjects were exposed to the recording and asked to evaluate
the speakers physiological and personality traits.
Similar to the Quebec study English speaking subjects
did not rate Jewish accented speakers favorably on any trait.
Jewish subjects rated English speakers as more physically
attractive and better leaders. Jewish subjects felt Jewish
speakers were more entertaining, kinder, and had a better
sense of humor.
A variable not present in the Quebec study was the
possibility of incorrectly identifying the speaker's guise.
Since both speakers in this study used English it was
possible for listeners incorrectly to identify the
speaker's accent. In fact, listeners did make this
mistake, although not in sufficient numbers to warrant a
statistical analysis. It was found, however, that Jewish
subjects tended to assign Jewish guises to non-Jewish
speakers and the Gentile subjects exhibited a reverse
tendency, in that they identified less accented guises as
Jewish. As with the subjects correctly identifying the
guises, Jewish speakers were downgraded on physical and
personality ratings. In comparing the correctly identified
guises, very little difference between the two groups'
reactions was found.
Jewish and Gentile listeners were also asked to
evaluate the voices of their own ethnic groups. These
ratings were grouped into three clusters: "General evaluation"
which included ratings of good looks, self confidence, ambition,
sociability, character and likeability," secondly,
"dependability, and character," and thirdly, "affability."
These three clusters seem to answer the following
questions: "Am I good or bad?" "Can people count on me?"
and "Am I desired as a friend?" Jewish subjects rated
themselves more favorably than did the Gentile subjects
on all clusters and individual traits except religiousness.
As was previously stated, Jewish subjects categorized more
voices as Jewish than did Gentile subjects.
The most prominent finding of the study is the down-
grading of the accented guises on height, good looks, and
leadership. This reaction occurs in both incorrect and
correct identification, by both Jews and Gentiles. This
suggests that it was not the Jewish speaker specifically,
but any person speaking with an accent (Irish, Negro,
Southern, New York, etc). Since the evaluations were
depicted in relation to the accented and non-accented voices
of the same speaker, it seems unlikely that the basic
quality of the voice played any part in eliciting reactions.
As suggested by the previously cited research, it
appears likely that the accented voices aroused certain
perceptual stereotypes which had been acquired through
previous experience. Leadership devaluation can possibly
be based on the tendency of accented speakers to be
immigrants. The authors suggested that an immigrant might
not be expected to occupy positions of leadership and might
logically be considered as possessing few leadership qualities.
The devaluations of height and appearance can possibly
be explained by previous research, Bruner and Goodman (1947),
Carter and Schooler (1949), and Lambert et al. (1969),
which suggests that magnitude is a close associate of value.
By extension (to the realm of person perception) it is possible
that immigrants, who historically have been afforded low
status roles, would be regarded as short and unattractive.
In contrast with the Quebec experiment, where the
French minority consistently devaluated themselves, there
was a tendency for Jewish subjects to maintain Jewish
superiority when comparing Jews with non-Jews. The Jewish
subject allowed for some superiority of their "immigrant"
co-religionists as well. It appears that Jews have not wholly
adopted negative stereotypes concerning Jews in general
(as had the French); on the contrary, Jews appear to have
adopted positive stereotypes concerning Jews, even for
There are several possible explanations for these
findings, as Lambert suggests. Since the Jewish accented
speech was an adoptive guise for the speakers, their
unfamiliarity with this mode could have caused a hesitance
which resulted in the devaluated ratings of self-confidence.
Similarly, Jewish listeners could have been more sensitive
to the Jewish guises due to their greater familiarity with
them. Finally, in contrast with the French, Jews might
consider themselves a superior minority.
In the present study,the difficulty of judging people
accurately from their voices is again evident. The subjects
apparently seized upon whatever information was available
to them. The main sources of information were community-
wide stereotypes about people with accents. This study
thus reinforces the findings that stereotypes are functional
in prompting subjectively or objectively correct judgments
and need not necessarily function to justify prejudiced
attitudes or satisfy personality needs.
In a similar study, Lambert, Anisfeld and Yeni-
Komshian (1965) investigated the evaluational reactions of
Jewish and Arab adolescents to dialect variations. As in
the French Canadian example, bilingual speakers were
recorded in several language modes, i.e., variations of
Hebrew and Arabic. Taped sequences were evaluated by Arab
and Jewish adolescents. Evaluational scores were recorded
on a semantic differential style scale.
As in the previous studies, speech based stereotypes
were found to be operating along predictable ethonocentric
skews. Arab Ss found the recorded individuals in the
Jewish guise less capable, less dependable and less desirable
than speakers in the Arabic mode. Conversely, Jewish Ss
showed a marked preference for speakers of Hebrew. Further,
the Jewish Ss were found to prefer one specific dialect of
Hebrew (Ashkenazic) over another (Yemenite). As in the
Hebrew-Arabic comparison, negative ascriptions were
registered for speakers of Yemenite Hebrew.
Generally speakers who sound "foreign" to a particular
group of listeners tend to be rated lower than speakers
who do not. Mulac, Hanley and Prigge (1974) assume that
the findings of Lambert and associates dialect studies
were the results of the listener's reactions to one or more
of the three elements of the speaker's linguistic
presentation: phonology, semantics and syntax.
It was the purpose of Mulac, Hanley and Prigge's (1974)
study to determine the impact of phonological aspects of
European born persons' speech upon attitudinal judgments
made by several groups of American listeners. The thrust
of this investigation follows the findings of previous studies
by assuming that the way a given communication sounds (as
opposed to what it means) determines how the speaker is
Four independent variables were manipulated in this study:
(1) Listener sex; (2) Listener age, occupation group (middle
aged, middle class townspeople and college students); (3)
Speaker sex; (4) Speaker country of origin (Norway, Italy,
Eastern Europe and United States). Twenty-six foreign born
graduate students were recorded during impromptu monologues
in English. An important methodological divergence of this
study was the feeling that spontaneous conversations sound
more natural than the prepared passages of Lambert and others.
Content was controlled by asking the speakers to
describe two large landscape pictures, while refraining
from mentioning his or her native country, language, age or
field of study. The speakers were originally from the
previously indicated areas in Europe and the United States.
Male and female speakers were employed. The 16 samples
determined to be most representative were selected for use
in the experiment. Speaker sex and origin were randomly
assigned to a 45 minute master tape. For purposes of
measurement the experimenters devised a speech dialect
attitudinal scale. This instrument consisted of a scale
of 50 bi-polar adjectives designed to test the orthogonal
dimensions along which speech foreignness is judged. Three
factors were established for the final factor solution:
socio-intellectual status, aesthetic quality, and dynamism.
Middle aged, middle class townspeople (26 males- 26 females)
and university students (25 males- 15 females) served as
experimental listeners. The subjects were all exposed to
the same 45 minute tape and asked to complete the rating
instrument. The results of this study corroborated the
earlier findings of Lambert. Greater degrees of phonological
speech foreignness exhibited by the twelve European born
speakers elicited lower judgments on all three attitude
dimensions than similar American born speakers. The initial
findings of Lambert were confirmed in this experiment, even
though many other variables were manipulated and a different
experimental method was utilized. In addition,this
preference for non-accented English was operative across
sex and status lines.
A (1967) study by Markel, Eisler and Reese investigated
the influence a person's dialect exercises on perceptions
of his personality. Taped samples of speakers with
Buffalo, New York,and New York, New York,"dialects" were
recorded reading a standard passage. These speech samples
were then evaluated by a group of undergraduates enrolled
in the State University of New York at Buffalo. These
experimental subjects evaluated the taped speakers via a
semantic differential scale. Markel found that linguistically
naive decoders were sensitive to a dialect variation of
their language and this dialect stimulated a stereotypic
response concerning the personality characteristics of the
speakers of that dialect.
Studies involving Chicanos and American subjects have
also revealed devaluation and ethnic reactions similar to
Lambert's studies. In a series of "wrong number" experiments
(whereby the speaker would "accidentally" call the experimental
subject), Harris and Baudin (1973) found that Chicano subjects
helped a Spanish surnamed confederate who spoke Spanish more
than one who spoke English (this suggests that the language
may have cued an ethnic sympathy reaction similar to that
of the Jewish subjects in the Lambert experiment).
Building on the findings of her first study, Harris
further predicted that a Spanish accent for Spanish surnamed
subjects would have the same effect. Harris theorized that
Anglo subjects would be more aggressive and less helpful
to a caller with a Spanish accent and Spanish surnamed
subjects would be more aggressive and less helpful to
callers with an English (Anglo) accent.
Harris used a methodology similar to her pilot
experiment. Forty-eight Spanish surnamed and forty-eight
Anglo surnamed telephone numbers were randomly selected from
the telephone book. Bilingual female subjects called these
numbers in either the guise of Spanish accented or non-accented
"wrong numbers." The responses of the "wrong number" answerers
were recorded and evaluated.
The results of this study supported the idea that
prejudice against people with Spanish accents does exist;
it provided no support, however, for the idea that Spanish
surnamed subjects would be less aggressive and more helpful
to someone with a Spanish accent. Harris found that all
subjects (both Spanish and Anglo), who became more aggressive
as the call went on, were in the Spanish accented condition.
These findings partially contradict Harris's expectations,
but lend added support to Lambert's findings of prejudice
toward all accented speech. On the basis of Lambert's
Quebec experiment, it could be theorized that the Mexican
subjects responded unfavorably to Spanish accented callers
due to a stereotype of ethnic inferiority. This would
correlate closely with Lambert's rationale for similar
responses by the French ethnic group.
In the foregoing studies listeners were able to make
credibility and personality judgments based on stereotypes
associated with certain speech characteristics. It is
significant to note, however, that listeners were not able
to identify which language features acted as cues for
Studies of Black Non-standard Dialect.
The final area of dialect research to be reviewed in
this work deals with the perceptual consequences of employing
Black Non-standard English in the U.S.A. culture. It is
this body of research which instigated the present study.
In general, these studies have employed methodologies
identical to the more general studies in the voice/perception
Lambert (and Tucker) employed techniques similar to
his earlier investigations in a (1969) study testing reactions
to various American English dialects. In this study Lambert
attempted to answer the following questions: (1) Are both
Negro and white subjects sensitive enough to dialect
variations to make reliable differentiations? (2) If so,
will there emerge a meaningful pattern of dialect preferences,
i.e., some particularly favored and others disfavored? A
rating instrument was devised for this study by asking
Southern Black college students to indicate those traits
they felt were most important for friendship and success.
Their responses were tabulated and ranked in order of
popularity. They were then asked to assign synonyms and
free associations to the trait names. Dialect samples were
selected by trained dialectologists. Recordings were made
of four representatives of each of the following six dialect groups:
(1) Network English ("middle American dialect")
(2) College educated white southern speakers
(3) College educated Negro southern speakers
(4) College educated Negro speakers from Mississippi
currently attending college in Washington, D.C.
(5) Southern Negro students (peer group) who spoke a
dialect similar to that used by students at the
Black college where testing was conducted
(6) Alumni from a New York City College.
Speakers in groups 1 and 2 were white, while the
remaining speakers were Black. Each speaker recorded an
identical short, non-salient passage. Both male and
female speakers were used, except in category 2 (due to
and oversight), which included only males. The 24 recordings
were placed on two tapes, twelve speakers on each.
The listening subjects were Negro male and female college
under-graduates from a southern Negro college, white male and
female students from a New England college, and white male
and female students attending a southern university, The
students serving as judges in all three cases were asked
to listen to the voices and to evaluate each speaker in
terms of a 15 trait scale devised by Lambert.
Lambert and Tucker found only a few instances where
sex differences in response occurred ( in these the females
tended to rate the speakers slightly more favorably), so
the ratings of the males and females were combined.
Statistical analysis clearly demonstrated that each
group was able to differentiate the various dialects. All
three rating groups (nearly unanimously), perceived the
network speakers as having the most favorable profile of
The dialect group rated next most favorably by both
northern white and southern Negro judges was the educated
southern Negro. Southern white judges on the other hand
rated members of their own peer group next most favorably.
Southern white students rated educated Negro southern
In terms of the least favored group, Negro judges
rated educated white southern speakers least favorably on
every one of the 15 traits. Whereas, white judges, both
northern and southern, rated the Mississippi peer speakers
least favorably and the New York alumni speakers only
The two groups of white judges were also asked to
indicate which race they thought the speakers-were.
Northern whites estimated race with the following accuracy
for the six dialects: Network 95% white, educated white
southern 87% white, New York alumni 49% black, educated
Negro southern 49% black, Howard University 85% black,
Miss. peer 94% black. White southern judges estimates were:
(1) 98% white, (2) 96% white, (3) 47% black, (4) 54% black,
(5) 70% black, (6) 89% black, respectively for the above
six dialects. Southern whites were slightly more accurate
in race estimation than northern whites, but in most cases
the true race of the speakers was judged a majority of the
In general, subjects were clearly able to differentiate
the dialect groups and they clearly favored the network
style of spoken English. Lambert and Tucker felt the
different perspectives of blacks and whites (with regard
to least favorable) reflect basic comparisons in effectively
toned attitudes that representatives of America's major
ethnic groups hold toward one another. The contrasts
also make it evident that speech styles which are pleasing
to one social group will not necessarily be so perceived
Buck, in a similar study (1968),attempted to
measure the effect that a non-standard dialect has on
perceptions of the speaker. In this experimental investi-
gation Buck dealt with non-standard "New York" versions
of black and white speech. Speech samples were recorded
of standard and non-standard speakers reading a passage of
Alice in Wonderland. Buck found that both white and black
speakers employing non-standard dialectal speech styles
were devaluated. Black and white standard speakers were
perceived by the experimental Ss as more competent. In
general, the experimental Ss preferred the "standard" New
Buck's report of her methodology is rather brief and
as a result it is difficult to evaluate its validity. The
sample size was small (N=25); neither the particular scales
employed nor a tabulation of results was provided. Speech
samples were elicited via a script reading situation which
would serve to stifle spontaneity and standardize word
choice. Both of these dimensions could significantly
influence the ratings a speaker received.
Bochner and Bochner in a (1974) study explored the
relationship between American social status and social
dialect. The lack of systematic descriptions of listeners
and speakers, listener attitudes toward speakers and
language cues used in forming listener judgments of speakers,
and dialects in previous studies was avoided.
Ninety-six undergraduate subjects were ranked in
one of three social categories (high, middle and low)
based on their occupation, education and residence.
Linguistic samples representing urban Boston (deemed
upper class) and the Black ghetto, Washington, D.C. (deemed
lower class) were prepared. Two messages identical in
semantic content, but differing in syntactical and lexical
choice were prepared. Both messages were recorded by
native speakers of the respective dialects. Each
experimental subject heard only one message. No information
pertaining to the source was supplied. Factor analysis for
the experimental sample was conducted across 18 scales
of authority and 18 scales of character.
Subjects responded very positively to the standard
English high status syntactical patterns. Conversely,
subjects evaluated negatively non-standard low status
speakers. According to Bochner and Bochner these findings
support the premise forwarded by most American linguists: Syntax
(word arrangement) does differentiate dialect. Further-
more, Bochner and Bochner theorized that listener status
plays a weak role in speaker evaluations. The determinant
may well be the norm of the listeners linguistic community.
This finding is substantially the same as Lambert's, Harris's,
and others' findings of preference for a dialect other than
the listeners own. In this study all subjects, regardless
of their own social status, held as their linguistic norm
white standard American English. The positive responses
to high status dialect confirm this preference.
The negative responses were viewed as rejections of
a racial minority linguistic norm of which the subjects
had no part. Some evidence, although slight, was found
indicating perhaps certain verbal structures may affect
the reaction to the meaning of a given sentence more than
Dialect and status effects were found for isolated
vocabulary pairs along lexicon (diction) factor dimensions.
These results suggest that while subjects did not respond
to vocabulary as an exclusive indicator of dialect, they
may have responded to vocabulary as an indicator of social
status. This may be an effect of the aesthetic or intense
quality of the language used. It is possible that differing
groups hold different views of what is linguistically
As in the previous attitude studies, high status
(standard English) speakers were positive. Conversely, low
status speakers were devaluated. Bochner and Bochner feel this
may also be a result of the linguistic norm of the
listeners. They theorize that the dialect of the low
status speakers served to identify them as non-members
of the community. In addition, middle status (class)
listeners were the most uniform in their preference for
standard English speakers across the scales.
In conclusion, Bochner and Bochner present the
following five conclusions with regard to speaker ethos
based on social dialect:
"(1) A person's response to the dialect speech
of another is not uni-dimensional. Three
dimensions reflecting the factor structures
of this investigation were response to
meaning, response to grammar, and
response to aesthetic quality.
(2) Syntactical cues appear to be primary in
listener-differentiation of dialects.
(3) The social dialect of a speaker may
influence listener judgments of his
character, but not of his authority in
a given situation.
(4) Persons of differing social status who
nevertheless subscribe to a similar set
of linguistic norms are not likely to
make significantly different judgments
about another's dialect.
(5) Dialect in speech is more likely to be a
significant determinant of speaker ethos
for persons of middle social status than
for persons of higher or lower social status."p82
The findings that syntax is the primary cue in listener-
differentiation of dialects does have some support. It is
important to note that the primary determinant of the results
of this study (syntax) was pre-prepared rather than
spontaneous. This fact could possibly diminish the
generalizability of the results.
The conclusions and contradictions of Bochner and
Bochner's study serve to underscore the tentative nature of
similar experiment's findings. Their apparent refutation
of the belief that "intonation" plays a major role in
person perception may, however, be a consequence of a faulty
research design. Like most other researchers in the field,
Bochner and Bochner utilized a carefully worded prepared
statement as their experimental stimulus. It may well be
that in their attempt to guard against content reactions
to their recorded stimulus they over-compensated and
confounded their findings anyway. Although prepared scripts
insure content free dialogue, the mere fact that they are
contrived and not spontaneous may have an influence.
On a common sense level it is apparent that naturally
occurring evaluational reactions would, of course, occur
as the result of spontaneous encounters. Further, in
those situations where a speakers discourse is not
spontaneous (i.e. speeches, presentations, plays, etc.), it
can be assumed that the speaker is familiar with the text,
either through composing or rehearsing it. This element
of naturalness is decidedly lacking in the foregoing
studies utilizing prepared scripts.
A study by Linn (1975) described the attitudes of black
and white adolescent and pre-adolescent youths toward black
English. Specifically, it looked for an impact of the
"Black is beautiful" movement on Black and white Americans.
The methodology developed by Lambert (bi-dialectal speakers)
was employed. Sixth grade and high school students
listened to taped segments of three black speakers in
two "guises," standard and non-standard English. They
then evaluated them on a number of personality traits.
Whites and blacks judged the standard English speaker to be
smarter, nicer, better educated, richer, less prejudiced and
more interpersonally oriented. Blacks rated the Black
English speaker as taller, better looking, and better liked,
while whites judged the standard English speaker to possess
Delia (1975) investigated the effect that a speech
style has on the way the spoken message is processed.
Specifically, Delia predicted that when a listener makes
evaluations based on perceived incongruity (between his
and the heard speech style) he will anticipate certain
attitudes, opinions and values in the speaker. Secondly,
Delia expected to find that speakers employing stereo-
typical impressions will continue to use them until they
are disproven. Via a methodology using the familiar tactic
of pre-recorded script speech samples,Delia confirmed
his hypothesis that listeners assume incongruity (of ideas)
with speakers employing different speech styles than their
own. As a result listeners orient themselves toward
speakers based upon the evoked "stereotypical guidelines"
until further information confirms or contradicts them.
The results of Delia's study underscore the negative
effects of certain dialectal styles. Not only do they
color the way we perceive a speaker,but also they serve
as tentative indicators of his or her attitudes and
allegiance. Delia's results cannot be taken as the last
word on the subject, however. The subjects were only asked to
estimate the degree of similarity/dissimilarity in the
abstract. The introduction of specific topics and attitude
areas may have some influence on his findings.
Gardner and Taylor's (1968) study somewhat supports
these findings. Their study of ethnic stereotypes in
social pressure situations found subjects employed stereo-
typical responses when confronted with ethnic speech styles.
Their findings suggest that in the absence of other
information evaluational responses based on speech tend to
conform more closely with ethnic stereotypes.
In an interesting (1973) study, Green investigated
the evaluations of Black Non-standard dialect by both
black and white speech therapists. Green asked 24 white
speech therapists to identify articulation errors in the
speech of standard speaking and non-standard speaking
black children. A team of Black speech pathologists
sophisticated in Black English also rated the taped speech
samples. The ratings were compared for agreement. The
findings indicate that disagreements between therapists
and judges were a function of Black English rather than
degree of deficiency. Specifically, the white and Black
judges agreed about the errors present in the standard
English sample but disagreed when evaluating Black dialect
speech samples. Whites perceived more errors than Blacks.
It is interesting that the negative perceptions of Black
Non-standard English color even objective evaluations
of articulation by professionally trained judges.
The foregoing research findings suggest that the use
of Black Non-standard English negatively colors the way an
individual is perceived over a number of personality and
physical characteristic dimensions. It also appears
that a person's vocal style, in this case Black Non-
standard English, provides (stereotypical) clues to the
users' attitudes and interpersonal orientation. Specifically,
the research findings suggest that if a listener perceives
a difference between his speech style and that of a given
speaker he anticipates a corresponding difference between
his beliefs and values and those of the "different"
Dialect and Scholastic Success.
In the last few years many studies of Black Non-standard
English have found that as a speech style it also exerts
an influence on perceptions of a person's performance in
a number of areas. It is potentially possible that
remediation of negatively perceived speech styles could
potentially produce results similar to the transformation in
Shaw's Pygmalion. To this end recent studies have investigated
the effect of ethnic stereotyping on teachers' expectations
and evaluations of student performance, as teachers along
with all people represent their background and fields of
experience. Recent research in the area of teachers'
subjective evaluations of students based on speech has been
conducted. Basing their research on Bernstein's (1968)
theory of elaborate and restricted codes and their socio-
economic determinate, Lambert, Frender and Brown (1970)
investigated the role of speech characteristics in scholastic
In their study, French Canadian third grade boys were
compared with regard to verbal intelligence and speech
characteristics. Both groups came from lower social class
homes. The independent variable was school performance.
Speech samples were elicited from both groups of Ss. It
was found that the groups differed reliably in verbal
intelligence and speech characteristics. Even with
verbal intelligence statistically controlled distinctive
speech patterns differentiated the groups. Speakers'
speech styles were found to differ significantly along a
set of dimensions that were demonstrated to affect their
interpersonal perceptions. Specifically, those third
graders receiving the poorest marks in class also received
the lowest speech style evaluations.
Conversely, "proper" speech styles were an accurate
predictor of good grades. Even when verbal intelligence
was similar it was found that a lower class youngster's
style of speech may mark or caricature him and thus
adversely affect his opportunities to improve, including
his opportunity in school performance. These findings
conflict with Bernstein's "restricted code" theory in that
presentational qualities (intonation, pitch, etc.) not
terms employed, played the key role.
Williams, Whitehead and Miller (1971) set out to
analyze the "pygmalion effect," the attitudes which
language characteristics may elicit in listeners. This
study focused upon teachers' stereotyping of pupils.
Specifically, they tested the degree to which visual
cues of a child's ethnicity will influence judgments of
a standard English speech sample. Prior research showed
that some teachers incorrectly rated the speech and
language of particular children in video taped speech
samples. They matched some standard English samples with
video images of Black, white, and Mexican children.
It was found that the video tape image showing the
child's ethnicity affects ratings of his language in the
direction of racial stereotyping. For Black children the
bias was in the direction of expecting them to sound more
ethnic and non-standard than their white peers.
Although this study does not deal with the actual
consequences of using non-standard English it is an
interesting documentation of the expectations ethnicity
elicits with regards to speech stereotypes.
Williams (1970), in another study investigating the
consequences of Black English, examined the influence a
student's speech style has on teacher evaluations.
Taped samples of fifth and sixth grade children
(from Detroit schools) were played to 33 primary school
teachers. Students were ranked beforehand via their
socio-economic status as high or low status. Teachers
were asked to evaluate the "correctness" of the child's
speech style via a semantic differential type scale.
The teachers were able to make statistically reliable
differentiations of the Black children's social status
but not of the white children. White teacher's evaluations
reflected a direct association between race and status.
White teachers more strongly associated speech standardness
with social status than did black teachers.
Piche, Michlin, Rubin and Sullivan (1977) undertook
a study to determine the relative effect of dialect
ethnicity, social class and quality of written composition
on teachers'subjective evaluations of fourth grade children.
As in most previous studies audio recordings of subjected
speech samples were evaluated by a group of potential
speakers. These samples were elicited by asking the
children spontaneously to tell a story based on a series
of cartoons. In addition, the subjects were video taped
for comparison of visual and vocal cues.
Finally, 75 fourth grade children submitted papers
for evaluation. The four highest and the four lowest
were converted to typed scripts for presentation to the
experimental subjects. The results of their study did
not disconfirm the presence of social stereotyping process
in teachers subjective student evaluations. It did, however,
accentuate the complexity of such processes. Attributions
of socio-economic status were found to have a main effect.
Interactions of socio-economic status, dialect, ethnicity
and quality of written composition were found to influence
teachers' subjective evaluations.
Further, and disquietingly, the study revealed that
the quality of the children's written compositions exerted
the least direct effect on the teachers' judgments.
In line with other studies, Piche et al. conclude
that ethnic stereotypes act in a relative, not absolute
manner. Their first effects are modified over time by
reinforcing or counteracting information such as written
DeMeis and Turner (1978) addressed the effects of
race, dialect and physical attractiveness on teachers'
evaluations of performance. Students were Black and
white and were pre-rated as high, middle or low in
attractiveness and were ranked as either standard or
non-standard English speakers. Sixty-eight white
elementary school teachers listened to responses from
these children and then rated them in terms of personality,
quality of response and current and future academic
potential. Results showed that generally Black and
Black English speaking children and low attractive
students were rated lower. These results provide some
support for attributing these children's academic failures
to their race, dialect or appearance rather than their
Granger, Mathews, Quay and Verner (1977) investigated
teacher evaluations of functionally equivalent speech samples
obtained from twelve middle and lower class (socio-economic)
Black and white third graders. Fifty-six female preschool
and elementary school teachers rated and ranked taped
picture descriptions. Results indicated that middle
socio-economic status children were rated more positively
than the lower status group and that white speakers were
perceived more favorably than Black speakers.
A similar study by Sanford (1978) examined the actual
number of phonological deviations, syntatic deviations,
and reading deviations (Black vs. white) in relation to
teacher evaluations. Sixty-one classroom teachers from
the Midwest listened to tape recordings of three Black
children and three white children reading perfectly.
Subjects were instructed to record all deviations they
considered errors. Teachers recorded a significantly
greater number of deviations for the Black readers.
The subjects recorded significantly more phonological
deviations for the Black readers. The subjects recorded
significantly more syntactical errors for the white readers.
There was no interaction between ratings and the attitude
Purpose of the Study.
The compelling consensus of research centered around
Black Non-standard English is that speakers employing this
speech style are negatively perceived in terms of personality,
physical characteristics, power, and ability. These
devaluations, while stereotypical and inaccurate, are nonethe-
less a potentially limiting consideration for BNS speakers in
the actual practice. Unanswered in these studies, however, is
the question of how listeners are able to make these consensual
ascriptions. It must be remembered that the preponderance of
research in this field involves taped samples of speech with
no demographic data about the speaker. Yet in study after
study, evaluative listeners, ranging from college freshmen to
speech pathologists, have responded in remarkably similar
ways to Black Non-standard English. BNS has been identified
by various studies as being negatively perceived. These
negative perceptions have been noted cross culturally and
have been registered even by those who employ BNS as their
sole or primary speech style. These same studies have
identified standard English General American dialect as the
evaluative opposite of BNS in that it is the most favorably
perceived speech style yet tested. Studies of General
American dialect indicate that, regardless of their own
habitual speech style listeners respond favorably to
speakers employing GA.
The research reviewed for this study suggests that
discrimination made based on the use of BNS are made over
aesthetic and stylistic considerations rather than
comprehensional ones. While research findings do suggest
that the use of BNS does inhibit the user's ability to decode
standard English symbol arrangement and comprehension, no
studies suggest that cross-racial comprehension barriers are
the basis for negative evaluations of BNS. These negative
stereotypical perceptions, therefore, must be based on
stylistic considerations. Simply stated, something about the
way the BNS speaker "sounds" negatively influences the way he
Unfortunately it is unclear what specific aspects of
speech style are responsible for these ascriptions. It is
of immediate importance then to determine the basis for the
speech borne stereotypical ascriptions noted in the
literature. No attempt heretofore has been made to separate
the two elements of a speaker's style which may be
responsible for speech discrimination, (1) grammatical
and (2) phonological differences (accents). Further compli-
cating the situations is the use of "matched guise, or non-salient
prepared script, methodologies in the preponderance of speech
style investigations. Matched guise studies employ only one
speaker who is "fluent" in both of the dialects or styles
under investigation. These "bi-dialectal" speakers record
a prepared script in both modes of speech for evaluation by
listeners. A similar strategy commonly employed in person
perception research involves the use of several dialectally
representative speakers reading content controlled scripts.
The goal of both of these approaches is to control the
salience of the recorded passages and so isolate the para-
linguistic variables for investigation. The pitfall of these
methodologies is their inherent lack of naturalness. It is
possible that this lack of "spontaneity" could have influenced
the results of the studies in which they were employed.
Lee (1970) voices this same criticism of the unnaturalness
of most of the prepared passages employed in person perception
research. He contends that most of these samples are
characteristic of reading style and as such are not related
to spontaneous (oral style) communication in general. In
addition, Lee feels that methodologies which employ repeated
content controlled messages free listeners (due to repetition)
to be more discriminating in their evaluations of speakers
after the initial recitation. Lee contends that most of a
listener's attention is absorbed by simple message decoding,
unless the message is repeated.
If a valid methodology could be developed which provided
a separation between the phonological and grammatical aspects
of oral communication, it might allow the identification of the
specific variables which are associated with the negative
stereotypes engendered by Black Non-standard English. The
ability to identify and distinguish between the major elements
of spoken language would be of great assistance in the
remediation or alteration of the specific variables that
arouse negative perceptions of the person in these cases.
Traditional education in the United States places emphasis on
standard grammar and syntax. Studies of speech style conducted
in secondary schools (reported previously) suggest that even
when using standard English syntax and grammar Black children
can still be negatively evaluated on speech related variables.
Harms (1961) in his conclusions hypothesized the existence of
"micro-scopic" speech cues which act as the basis for
discrimination of status from the voice. It may well be
that the elements of speech (BNS or any style) that arouse
stereotypical responses in the listener operate below the
level of grammar and structure. A methodology sensitive to
the minute levels of speech style could facilitate the
remediation of any undesirable speech characteristics by
isolating them for specific attention.
It was the purpose of this investigation to develop a
valid methodology with the ability to simply discriminate
between the grammatical and phonological channels of spoken
language and to investigate the evaluative consequences of
using BNS (as compared with standard English General American
dialect) via that means.
Based on the implications of previous research, the
following hypotheses were forwarded for testing in this study:
(1) A. Speakers who employ Black Non-standard
English dialect will be less favorably
perceived over three evaluative measures
than individuals employing standard General
B. Due to the pervasive nature of the stereo-
types associated with the two dialects the
preference for SEGA and BNS will be
consistent across racial and sexual lines.
(2) Evaluations of these two speech styles (as
well as others) are primarily based in the
phonological aspects of speech rather than
the grammatical aspects. Specifically, BNS
speakers will be negatively perceived even
when employing the word choice of General
The experimental stimuli of this study were audio
taped samples of 8 male speakers of two American (U.S.)
English dialects: Black Non-standard (N=4) and General
American (N=4). To control the salience of the content
of these speech samples the speakers employed in this study
were recorded describing a landscape by Van Gogh. These
spontaneous descriptions were 60 seconds in duration. By
spontaneous it is meant that the speakers described a common
stimulus, each employing their own unprompted, idiosyncratic
word choice and order. A major goal of this study was to
separate the grammatical and the phonological aspects of the
two speech styles. In essence, to test the veracity of the
cliche "it's not what you say but how you say it." To achieve
this end, once the spontaneous descriptions were gathered, they
were reduced to script form complete with mispronunciations
and vocalized non-fluencies. These scripts were then returned
to the two groups of speakers. Each speaker then recorded the
script of a randomly assigned dialectal counterpart.
The exchange and recording of scripts by dialectal
counterparts is designed to separate the pnonological and
grammatical aspects of the speakers' speech styles, and in
this way, allow the separate evaluation of their impact on
person perception. The initial tape, then, consisted of
16, 60 second speech samples collected from 8 speakers in
each of two conditions: spontaneous description and script
reading. The complete scripts are listed in appendix 1.
To control ordering effects, the experimental tape was
constructed by randomly ordering two (15-20 second) segments
of each speaker's speech in each of the two speaking styles.
In this way each speaker was evaluated 4 different times,
twice from segments of spontaneous description and twice from
segments of script reading. This procedure yielded an
experimental tape that consisted of 32 randomly ordered
segments of speech.
Research by Harms (1961) has determined that listeners
form speech based evaluations of speakers in the first 10-15
seconds they are exposed to their voices. As a result of
these findings the brief duration of the speech segments used
in this study is not considered to pose a threat to validity
or a limitation for the overall purposes of this research.
The brief duration of the segments also prevented rater
fatigue from confounding results. The relatively large number
of samples (32) rated was designed to prevent raters from
focusing (negatively or positively) on one particular speaker
and in that way skewing the data.
The speakers employed in this study were selected by
a three step process. Potential speakers were first selected
by the way they "sounded" to the untrained ear of the
experimenter. Secondly, in an attempt to control all
variables besides the ones in experimental consideration,
only speakers with similar backgrounds were employed. All
speakers were males between the ages of 20 and 40.
Further, all speakers were currently enrolled in college or
had already completed a college degree. General American
speakers were also required to have originated in and spent
the majority of their lives in the midwestern United States.
Finally, to test the "untrained ear" of the experimenter
and determine positively the dialectal representativeness of
the potential speech samples the tape of the 32 randomly
ordered speech samples was independently judged by 3 dialect
experts. In sum, the speakers can be considered valid
representatives of their dialects.
The aforementioned research generally concluded that
accurate specific determinations of personality or physical
characteristics cannot be made from the voice. However, very
reliable and valid judgments can be made on a general
evaluative dimension (Osgood, Suci and Tannenbaum 1957).
The perceptions of the speakers in this study were made via
three highly factor loaded bi-polar adjective pairs which
tapped the evaluative dimension (good-bad, wise-foolish,
rich-poor). The individual scores from these scales were
combined to compute a mean evaluative factor score. The
evaluative dimension was considered highly relevant for this
research since it deals with the factors most potent in real
world speech judgments. The rating instrument consisted of
3 sheets (one for each of the above adjective pairs) on
which each rater rated all 32 segments on one six point bi-
polar adjective scale. The scales were constructed with the
"positive" scores on the left and the "negative" scores on
the right. Each scale position was given a value from one
to six, reading from left to right. As a result low mean
scores indicated positive ratings, while high mean scores
indicated negative ratings. Each rater rated the speakers
on only one of the three scales. The three scales were
randomly distributed to the raters. The sex, race, and age
of the rater were also collected on each sheet.
Raters in this study were 48 graduate business students
(24 male, 24 female) enrolled in the MBA program of the
Business school at the University of Central Florida, during
the spring quarter of 1981. Graduate business students were
specifically used to give some actual practice validity to any
experimental results. It was felt that as these individuals
were (or soon would be) in positions of responsibility in the
realm of business (possibly in hiring and personnel functions)
their evaluations would graphically illustrate the actual
practice consequences of employing the two speech styles tested.
All raters were white. A statistically acceptable number of
Black raters were not present in the rater population; there-
fore, it was not possible to use Blacks as raters.
Administration of Rating Scales.
The design of this study permitted all experimental data
to be collected in one session. This is viewed as an asset
as it controlled any influence that different class settings
and administration times may have had on the results. The
experimental raters were told that they would be participating
in a study judging personality from voice. No information
about the dialectal nature of the study was provided.
Further, no demographic information about the speakers was
On the experimental tape the speakers were identified
only by number (1-32), a 15 second rating period of blank
tape was provided after each speech segment. The three
scales employed to tap the evaluative dimension were
randomly distributed to the raters. Each rater received only
one scale over which he or she evaluated all speech segments.
Once the experimental tape was started, it ran continuously
until its conclusion. The tape itself was played on the
same Wollensak Cassette Recorder Player that was used to
compile the 32 segment tape in order to minimize static.
A 2 x 2 x 2 factorial analysis of variance (Style x
Dialect x Rater's sex, Bruning and Kintz, 1968) was computed
for the evaluative ratings obtained by the procedure
described above. Tables 1 and 2 indicate the results of
the Analysis of Variance. These results indicate that the
speaking style and the dialect employed affect an audience's
evaluations of a speaker. Figure 1 graphically displays
these same results.
The Duncan Multi-range Test was used to make multiple
comparisons of the mean scores. For an N of 96, mean scores
of 4.0 script style and 3.49 spontaneous style were
significantly different (P = .05). Mean ratings of the
GA and BNS dialects also differed significantly (X = 3.27,
GA; X = 4.21, BNS; p = .05). No difference was noted between
means obtained from male and female raters.
Results of Factorial Analysis of Variance
Style x Dialect
Style x Sex
Dialect x Sex
Style x Dialect x Sex
Mean Scores for Each Dialect Style Group
* indicates p = .0001
Graph of Mean Scores for Each Dialect Style Group
The results of this research suggest that speakers
who employ Black Non-standard dialect as their primary speech
style are negatively evaluated because of the grammatical and
phonological aspects of that style. These findings refute
the author's second hypothesis, which predicted that the
phonological aspects of BNS are primarily responsible for the
negative perceptions of that dialect.
In line with the predictions of hypothesis two, BNS
speakers were negatively perceived both, when speaking
spontaneously and also when speaking with the grammar of
the GA dialect. The interaction effect noted between the
speaking condition and speakers' dialect (graphically dis-
played in Figure 1) suggests that the negative evaluations
of BNS dialect stem from both the phonological and
grammatical aspects of that speech style. Mean scores for
BNS speakers were consistently negative in both speaking
styles (Table 2) while positive evaluations of GA speakers
decreased dramatically when those speakers employed the
grammar of BNS dialect. It is possible, of course, that
some degree of this difference in evaluation can be attributed
to the differences in spontaneous speech and script reading.
As a result, the differences in rater evaluation of GA
speakers could potentially be viewed as support for Lee's
(1970) criticism of script-based dialect research. It must
be remembered, however, that the BNS speakers were negatively
evaluated across conditions. The author observed that the
taped segments of the GA speakers employing the grammatical
elements of BNS dialect sparked derisive snickers and
laughter during the collection of data. This suggests that
those grammatical elements more than the different speaking
styles were responsible for the significant difference between
Hypothesis 1-A predicted that the BNS speakers would be
negatively evaluated as compared to GA speakers. Hypothesis
1-B predicted no difference between the rating of the male
and female raters when rating the BNS speakers. Both
components of this hypothesis were confirmed (Table one).
These results support the findings of prior investigations
of BNS dialect. The findings of this study suggest clearly
that even in situations where Black Non-standard speakers
have similar backgrounds and qualifications to General
American speakers, and where the speaking situations and topics
are identical, BNS speakers will be at an evaluative
disadvantage due to the negative stereotype evoked by their
speech style. Even greater social significance can be
attached to these findings when it is recalled that the data
responsible for this conclusion were collected from graduate
students of business, individuals who currently, or in the
near future will occupy positions of responsibility and
leadership in the business community where this study was
The results of this study suggest that dialect is a
major dimension that raters rely on for their opinions of
other individuals. Specifically, that to whatever extent
a dialect can evoke a stereotypic response, the Black Non-
standard speaker will be at a disadvantage. It is not the
intention of this study to predict the consequences of that
disadvantage, but it is intuitive that in the competitive
business situations even small differences between individuals
can be significant.
The findings of this study lend renewed support to the
conclusion that Black Non-standard dialect is a negatively
perceived speech style. Further, it confirms that the
negative perceptions of this speaking style are consistent
between male and female raters. Future research should
determine the pervasiveness of this effect across other
variables (e.g. status and race) while underscoring the
pragmatic desirability of altering this negatively perceived
style (in essence to become bi-dialectal).
This study cannot offer a clear identification of the
specific elements of grammar or phonology in each of the two
aspects of BNS dialect which are responsible for negative
evaluations. Future research should attempt to further
"break down" these two aspects of dialect to identify specific
variables and facilitate their remediation. It may well
be possible to alter or eliminate the negatively perceived
elements of Black Non-standard dialect without adulterating
or diluting it as the carrier of culture and source of ethnic
identification its proponents suggest.
The most important finding of this study indicates
that achieving that end will require attention to both the
grammatical and phonological aspects of speech. Clearly a
system of English instruction which focuses solely on one or
the other of these two dimensions will be less than
successful in eliminating the negative stereotypes associated
with BNS dialect.
Scripts of Speakers 1 Through 8
Speaker 1 General American English
I'm looking at a picture by Van Gogh. The description
of this picture is....I think it's an oil. It looks like
a field, a wheat field, there's a road going through the
wheat field with some clouds in the distance and some, a
lot of black maybe crows. There also looks to be some like
green "foilage" next to the field and it looks like wheat
and it looks like there's a wind blowing across the field,
uhm, other than that I can't...it looks like, ah....it's
not very definite and its not very specific it's just an
oil and it's not...it's pretty abstract actually. To give
you more of an idea...the birds there's quite a bit of them
flying....flying to the east, and they're all heading in
Speaker 2 General American English
I'm looking at a picture that has many different colors
in it and various shades of some of the same colors. It
appears to be a wheat field. Within the wheat fie.d are
various yellows and shades of green and red, there is a
blue sky overhead. There seems to be at least two or
three white clouds. There are a number of birds pictured
in the scene flying at various levels. The wheat field
takes up approximately two thirds of the picture and the
sky the remaining portion. The predominant colors in the
picture are yellow and blue. It's a fairly interesting
picture I suppose, if one were to si...sit and....dwell
on it for a long length of time one might be able to place
various meanings in terms of what the author might....ah
have been intending.
Speaker 3 General American English
My interpretation of this picture is...it looks like
a wheat field with a winding road down the middle of it
with two roads at each end. Crows are flying overhead....
the ah sky looks like it's about to rain....a few clouds in
the sky dark blue. It looks like the ah...wind could be
blowin the wheat around, very tranquil looking....looks
like about twenty birds altogether but who's counting.
Speaker 4 General American English
I'm looking at a painting by Van Gogh. It's one of the
last paintings that he did, it's of a cornfield with crows
flying over it. It's done in an oil technique kind of a
"pasteche" that has dimension, it has bright colors which
are typical of the artist, very bright colors which are
typical of the artist. There is a road leading into the
cornfield. This was one of the final paintings that he
did before committing suicide and if I remember correctly
he blew his brains out in the corn field he is painting here.
The dominant colors are blue, yellow, ah.. sort of a reddish
brown and green.
Speaker 5 Black Non-standard English
Dis picture I'm "describin" is blue, yellow, green.
Da road design is green with brown "stripin" da artist he
use something like an arch brush, oil brush base. He have
an arrangement of birds in the blue like shade "symbolizin"
da sky, da yellow is symbolizin a field of some sort, he
have darker shades or it may be symbolizin a storm or a
cloud or something of this nature with a whirl of white in
da mixture of blue may be symbolizin sunlight. Da picture
is very colorful and it's by Van Ghose I'm told. Some
parts look like water but it's hard to tell because of the
minglin of the colors.
Speaker 6 Black Non-standard English
This is a picture by Van Gogh. Theres very lotta
description in this picture. I'm gonn start with the colors
that's in this picture. There's a very dark blue sky in
the picture where you can see about 20 to 25 birds which
is flying through the picture. On each side of the picture
there is ah....real yellow grass which looks more like
wheat and in between the two pictures there is a dark red
clay road which look like it passes through the gre....
the yellow wheat that's on each side of the road. Out in
the field looks like there hasn't been a rain in a while
everything looks dr...dry. The birds look as if they were
flying east in this picture, flying towards the storm or
something else that's on the picture.
Speaker 7 Black Non-standard English
This picture it's in this class I got uhh it's something
bout ahh bout Van Gogh right, and uhh this picture's called
ahh Crows or whatever, and it's like got a bunch of like
black spots but I don't really be into it, and uhh we just
be sitting around talking and everybody be half asleep and
ahh an it too much to it ahh and that's about it.
Speaker 8 Black Non-standard English
I got a, gotta painting here I'm looking at. It's
got a lotta different strokes of the brushes and stuff,
it's got a lotta different colors, sky blue. It's gotta
lot of black marks supposed to be birds, look like they
flyin over....don't know if it's land or what. Just a
bunch a, guess it's supposed to be land cause it's sorta
brownish. Ya got some birds flying high, some flying low,
ya got a few white spots look like ma....maybe they could
be clouds, well it could be running through a running river,
flyin through a valley or something. Land on both sides
birds flyin away uhm looks pret...looks pretty neat. Sorta
a abstract like painting. You probably see different things
in it if you really tried.
Your sex_ Your race Your age
SPEAKER RICH POOR
1 2 3 4 5 6
Addington, D. The relationship of selected vocal
characteristics to personality perception. Speech
Monographs, XXXV, 1968, 492.
Adelson, J. A study of minority group authoritarianism,
Journal of Abnormal and Social Psychology, XLVII, 1953,
Anisfeld, J., Boco, N., and Lambert, W. Evaluational
reactions to accented English speech. Journal of
Abnormal and Social Psychology, LXIV, 1962, 223-231.
Baily, B. Toward a new perspective in Negro English
dialectology. American Speech, XL, 1965, 171-177.
in Eds. Wolfrom, W. and Clarke, N. Black-White Speech
Relationships. Washington, D.C., Center for Applied
Belcher, M. A study of the relationships between certain
selected Black English dialect differences and the
reading achievement of 3rd grade disadvantaged Black
pupils. University of Houston, Dissertation Abstracts
International, V. 38 (5-A), 1977, 2541.
Bernstein, B. Language and social class. Social Class
Language and Education, ed. Lawton, D. New York:
Schocker, 1968, 77-102.
Bochner, B., and Bochner, A. The effects of social status
and social dialect on listener response. Central States
Speech Association, XXIV, 1974, 75-81.
Bowers, J. Some correlates of language intensity. Quarterly
Journal of Speech, L, 1964, 415.
Bruner, J., and Goodman, C. Value and need as organizing
factors in perception. Journal of Abnormal and Social
Psychology, XLII, 1947, 33-44.
Bruning, J.L., and Kintz, B.L. Computational Handbook of
Statistics. Glenview, Illinois: Scott, Foresman and Company,
Buck, J. The effects of negro and white dialectal
variations upon attitudes of college students.
Speech Monographs, XXXV, 1968, 181-186.
Cantril, H., and Allport, G. Judging personality from voice.
Journal of Social Psychology, V., 1943, 37-55.
Carter, C. Prospectus on black communication. School
Psychology Digest, 6(3), 1977, 23-30.
Carter, L., and Schooler, K. Value need and other factors
in perceptions. Psychology Review, LVI, 1949, 200-207.
Delia, J. Regional dialect message acceptance and perceptions
of the speaker. Central States Speech Journal, XXVI, 1975,
Demeis, D., and Turner, R. Effects of students race, physical
attractiveness and dialect on teacher's evaluations.
Contemporary Educational Psychology, V., 3(1), 1978, 77-86.
Fay, P., and Middleton, W. Judgment of spranger personality
and human feelings. Journal of Communication, XI, 1939,
Fetcher, E. The relation between bi-dialectal oral language
performance and reading achievement of black elementary
school children. University of New York, Dissertation
Abstracts International, V. 38(10A), 1978, 5908.
Gardner, R., and Taylor, D. Ethnic stereotypes, their
effects and person perception. Canadian Journal of
Psychology, XXII, (4), 1968, 267-276.
Granger, R., Mathews, M., Quay, L., and Verner, R. Teacher
judgments of the communication effectiveness of children
using different speech patterns. Journal of Educational
Psychology, 62(6), 1977, 794-797.
Green, L. Judgments of articulation errors co-existing
with Black English. University of Georgia, Dissertation
Abstracts International, V. 35(12B), 1975, 6175.
Harms, L. Listener judgments of status cues in speech.
Quarterly Journal of Speech, LXVII, 1961, 164-168.
Harris, M., and Baudin, H. The language of altruism, The
effects of language, dress and ethnic group. Journal
of Social Psychology, XCI, 1973, 37-41.
Hock, C., and Bowers, J. Dialect and identification in
persuasive messages. Language and Speech, XXII, 1969,
Holt, H. Language cognition and social attitudes of the
black child in public education: A sociolinguistic
and psycholinguistic analysis of kindergarten through
3rd grade pupils. Claremont Graduate School,
Dissertation Abstracts International, 32(6A), 1978,
Hoover, M. Community attitudes toward Black English.
Language in Society, V. 7, 1978, 65-87.
Kramer, E. Judgment of personal characteristics and
emotions from non-verbal properties of speech.
Psychological Bulletin, 60, 1963, 408-420.
Lambert, W. Reinforcement and extinction as factors in
size estimation. Journal of Experimental Psychology,
XXXIX, 1949, 637-641.
Lambert, W., Anisfeld, M., and Yeni-Komshian. Evaluational
reactions of Jewish and Arab adolescents to dialect and
language variations. Journal of Personality and Social
Psychology, II, 1965, 84-90.
Lambert, W., Frender, R., and Brown, B. The role of speech
characteristics in scholastic success. Canadian Journal
of Behavioral Science, II, 1970, 299-306.
Lambert, W., Hodgson, R., Gardner, R., and Fillenbaum, S.
Evaluational reactions to spoken languages. Journal of
Abnormal and Social Psychology, LX, 1960, 44-51.
Lambert, W., and Tucker, G. White and Negro listeners'
reactions to various American English dialects. Social
Forces, XLVII, 1969, 463-468.
Lee, R. Dialect perception: A critical review and
re-evaluation. Quarterly Journal of Speech, 27, 1970,
Licklider, J., and Miller, G. The perception of speech.
Handbook of Experimental Psychology, ed. S. Stevens,
New York: Wiley, 1951.
Linn, M. Black and white adolescent and pre-adolescent
evaluational responses associated with selected features
of Black non-standard dialect. University of Minnesota,
Dissertation Abstracts International, V. 35(8A), 1975, 5197.
Marino, P. An investigation into the inter-relationships
among linguistic variables and the ability to recover
deep structure among 2nd grade Black students. University
of South Florida, Dissertation Abstracts International,
V. 38(4A), 1977, 1876.
Markel, N. Relationship between voice quality profiles
and MMPI profiles in psychiatric patients. Journal of
Abnormal Psychology, Y., 74, 1969, 61-66.
Markel, N. Personality traits associated with voice types.
Journal of Psycholinguistic Research, V. 1(3), 1972.
Markel, N., Eisler, R., and Reese, H. Judging personality
from dialect. Journal of Learning and Verbal Behavior,
6, 1967, 33-35.
Markel, N., Meisels, M., and Houck, J. Judging personality
from voice. Journal of Abnormal and Social Psychology,
62, 1964, 458-463.
McEwen, W., and Greenberg, B. The effects of message intensity
on receiver evaluations of source message and topic.
Journal of Communication, XX, 1970, 340-350.
Miller, C., and Hewgill, M. The effects of variations in
non-fluency and audience ratings of source credibility.
Quarterly Journal of Speech, L., 1964, 36-44.
Mulac, A., Hanley, T., and Prigge, D. Effects of
phonological speech foreignness upon three dimensions
of attitude of selected American listeners. Quarterly
Journal of Speech, LX, 1974, 411-420.
Osgood, C.E., and Suci, G. J., and Tannenbaum, P.H. The
measurement of meaning. Urbana: University. Illinois Press,
Piche, G., Michlin, M. Rubin, D., and Sullivan, A. Effects
of dialect ethnicity, social class and quality of
written compositions on teachers subjective evaluations of
children. Communication Monographs, 44, 1977, 60-72.
Putnam, G., and O'Hern, E. The status significance of an
isolated urban dialect. Language, XXXI, 1955, 1-32.
Quay, L., Mathews, M., and Schwarzmuller, B. Communication
encoding and decoding in children from different socio-
economic and racial groups. Developmental Psychology,
1.(4), 1977, 415-416.
Rich, A. Interracial Communication. New York, New York:
Harper and Row, 1974.
Rubin, A. The role of context in information seeding and
impression formation. Communication Monographs, 44, 1977,
Sanford, R. Teacher attitude and black dialect. Purdue
University, Dissertation Abstracts International, 32(2A)
Sarnoff, I. Identification with the aggressor: Some
personality correlates of anti-semitism among Jews.
Journal of Personality, XX, 1951, 100-218.
Sereno, K., and Hawkins, D. The effects of variations in
speakers non-fluency upon audience ratings of attitude
toward the speech topic and speakers credibility. Speech
Monographs, XXXIV, 1967, 58-64.
Sledd, J. Bi-dialectalism: The linguistics of white
supremacy. English Journal, V. 58, 1969, 1407-1329.
Smith, A. Trans Racial Communication. Englewood, New
Jersey: Prentice-Hall, 1973.
Steckler, G. Authoritarian ideology in Negro college students.
Journal of Abnormal and Social Psychology, LIV, 1957,
Stewart, W. Continuity and change in American Negro dialects.
The Florida FL Reporter; Non-Standard Speech Patterns,
Baltimore Bulletin of Education Nos. 2-4, 1966-67,
52-65. in eds. Wolfram, W. and Clarke, N. Black White
Speech Relationships. Washington, D.C., 1971.
Taguire, R. Person perception. in The Handbook of Social
Psychology, ed. Lindzey, G. and Aronson, E., Reading,
Mass: Addison Wesley, 1969, 395-449.
Taylor, H.C. Social agreement on personality traits as
judged from speech. Journal of Social Psychology, V.
Thibaut, J., and Kelly, H. The Social Psychology of
Groups. New York: Wiley, 1959.
Williams, F. Psychological correlates of speech charac-
teristics: On sounding disadvantaged. Journal of
Speech and Hearing Research, XIII, 1970, 17-27.
Williams, S.F., Whitehead, J., and Miller, L. Ethnic
stereotyping and judgments of children's speech. Speech
Monographs, 38, 1971, 155-170.
John Albert Craft
Born Jacksonville, Florida, December 26, 1953.
Resident of Longwood, Florida, 32750
BA, General Communication, 3/75, Florida Technological
University, Orlando, Florida
MA, General Communication, 8/77, Florida Technological
University, Orlando, Florida
PhD, Speech, 12/81, University of Florida
Instructor of English for Speakers of Other Languages (ESOL)
Seminole Community College, Sanford, Florida
Adjunct Instructor of Speech, University of Central Florida,
Orlando, Florida, and Valencia Community College, Orlando, Florida.
I certify that I have read this study and that in
my opinion it conforms to acceptable standards of scholarly
presentation and is fully adequate, in scope and quality,
as a dissertation for the degree of Doctor of Philosophy.
Norman Markel, Chairman
Professor of Speech
I certify that I have read this study and that in
my opinion it conforms to acceptable standards of scholarly
presentation and is fully adequate, in scope and quality,
as a dissertation for the degree of Doctor of Philosophy.
Associate Professor of
I certify that I have read this study and that in
my opinion it conforms to acceptable standards of scholarly
presentation and is fully adequate, in scope and quality,
as a dissertation for the degree of Doctor of Philosophy.
Professor of Speech
I certify that I have read this study and that
in my opinion it conforms to acceptable standards of
scholarly presentation and is fully adequate, in scope
and quality, as a dissertation for the degree of
Doctor of Philosophy.
Jayne C. Harder
Professor of Linguistics
I certify that I have read this study and that
in my opinion it conforms to acceptable standards of
scholarly presentation and is fully adequate, in scope
and quality, as a dissertation for the degree of
Doctor of Philosophy.
Associate Professor of
This dissertation was submitted to the Graduate Faculty
of the Department of Speech in the College of Liberal
Arts.and Sciences and the Graduate Council, and was
accepted as partial fulfillment of the requirements for
the degree of Doctor of Philosophy.
Dean for Graduate Studies
December, 1981 and Research
UNIVERSITY OF FLORIDA
1 lillll 1 lll l1 IIII II IIII I9i41 1 11 IIl
3 1262 08553 9442