“■TBBB
8

m

Developmental Linguistic Realization in Black & White
Preschool Children: Local Normative Data

by
Lisa Petras

Submitted in Partial Fulfillment of the Requirements
for the Master of Arts Degree

Appp

bv:
Date

?son, Speech & Communication Studies, Edinboro University of PA

Chai

JqJjZdvisor

fomHiittee Member

/&//*/?£,
17

Date

LMJAZ

O-//?! 9,^
Date

AZ/M/9J

%

Date

Committee Member

Thesis Sp Com 1992 P493d
c.2
Petras, Lisa.
Developmental linguistic
realization
1992.

/'•

ACKNOWLEDGEMENTS
The author wishes to express her gratitude to several persons
without whose help and guidance, this research would not have been
possible:

to my mentor, Charlotte Molrine, for her invaluable editing

and statistical input, and also for her consistent support throughout my
graduate career; to John Boylan, for his knowledge and expertise in
field of phonology and interpretation of the PEPPER program; to Ruth
Nash-Thompson, for her personal insights and suggestions about BEV;
to the faculty of Edinboro University’s program in Speech-Language
Pathology, for enhancing my knowledge and encouraging me to be the
student I am today; to Rachelle L Laubach, for her consistent
encouragement and support throughout this study; to Keith M. Petras,
my brother, who has challenged me and motivated me to succeed; and
to my parents, Albert E. Petras and Constance R. Petras, whose belief in
my ability has supported and encouraged me to succeed in life.

••
li

TABLE OF CONTENTS
ACKNOWLEDGEMENTS

ii

TABLE OF CONTENTS

iii

LIST OF TABLES

v

LIST OF FIGURES

vi

I.

INTRODUCTION

1

II.

REVIEW OF THE LITERATURE

2

Assessment

2

Characteristics

4

Overview of Research

5

Statement of Problem

10

RESEARCH METHODOLOGY

11

Subjects

11

Examiners

12

Materials

12

Procedures

13

Phonetic Inventory

14

Inter- and Intra-Judge Agreement

15

Statistic Analysis

16

RESULTS AND DISCUSSION

17

Inter- and Intra-Judge Agreement

17

Statistical Analysis

17

Canonical Form

18

III.

IV.

ni

V.

Manner Features

21

Place Features

28

Phonological Processes

37

Syllable Structure Processes

38

Subtitution Processes

40

Assimilation Processes

42

CONCLUSION

43

REFERENCES

46

APPENDIX

51

Appendix A : Characteristics of Black English Phonology

IV

LIST OF TABLES
Table 1

Table 2

Table 3

Table 4

Table 5

Table 6

Table 7

Table 8

Table 9

Table 10

Table 11

Three-Way ANOVA for Between-Group Factors
of Race and Age and Within-Subject Factor of
Canonical Form as Computed by GANOVA-4

19

Three-Way ANOVA for Between-Group Factors
of Race and Age and Within-Subject Factor of
Nasals as Computed by GANOVA-4

21

Three-Way ANOVA for Between-Group Factors
of Race and Age and Within-Subject Factor of
Glides as Computed by GANOVA-4

22

Three-Way ANOVA for Between-Group Factors
of Race and Age and Within-Subject Factor of
Stops as Computed by GANOVA-4

24

Three-Way ANOVA for Between-Group Factors
of Race and Age and Within-Subject Factor of
Stops as Computed by GANOVA-4

25

Three-Way ANOVA for Between-Group Factors
of Race and Age and Within-Subject Factor of
Fricatives as Computed by GANOVA-4

28

Three-Way ANOVA for Between-Group Factors
of Race and Age and Within-Subject Factor of
Labiodentals as Computed by GANOVA-4

31

Three-Way ANOVA for Between-Group Factors
of Race and Age and Within-Subject Factor of
Alveolars as Computed by GANOVA-4

33

Three-Way ANOVA for Between-Group Factors
of Race and Age and Within-Subject Factor of
Palatals as Computed by GANOVA-4

35

Three-Way ANOVA for Between-Group Factors
of Race and Age and Within-Subject Factor of
Velars as Computed by GANOVA-4

36

Children Evidencing Unstressed Syllable
Deletion Process (USD)

38

v

Table 12

Children Evidencing Final Consonant
Deletion Process (FV)

39

Children Evidencing Cluster Reduction
Process (CR)

40

Children Evidencing Liquid Simplification
Process(LS)

40

Table 15

Children Evidencing Stopping Process (ST)

41

Table 16

Children Evidencing Velar Fronting Process (VF)

41

Table 17

Children Evidencing Assimilation Process (AS)

42

Table 13
Table 14

LIST OF FIGURES
Figure 1

Interaction of Age across Canonical Form

20

Figure 2

Interaction of Race Across Fricative Production

26

Figure 3

Interaction of Race Across Labiodental Production 29

Figure 4

Interaction of Race Across Age Groups in
Alveolar Production

vi

34

Developmental Linguistic Realization in Black & White
Preschool Children:

Local Normative Data

Chapter 1
Review of the Literature
Phonological differences between the adult model of Black
English (BE) and that of standard English (SE) have been well
documented (Dillard, 1972; Labov, 1972; and Wolfram & Fasold, 1974).
However, very little has been reported regarding phonological
development in Black English speakers. Although speech-language
pathology is sensitive to the importance of distinguishing dialectal
differences from language disorders, there are very few published
guidelines for making a distinction between dialectal difference and
phonological processing disorder in the emerging phonology of young
Black children. A clinician must compare the phonology of children
who use Black English either to the adult model of Black English, or to
norms reported for Standard English. Neither comparison seems
appropriate for developmental phonological assessment purposes
(Haynes & Moran, 1989).
For purposes of this discussion, Black English (BE) will be used
to denote a dialect used by African Americans in the inner cities of
most large urban areas and in the rural South (Owens, 1988). As such,
BE should be viewed as a systematic language of its own, with
characteristic differences in phonology, morphology, syntax, prosody,
1

pragmatics, and discourse, and not as a deviant or an improper form of
Standard English (Cole & Taylor, 1990).

Assessment
Because ASHA recognizes the legitimacy of all dialects of
language (Taylor, 1986), it mandates that speech-language pathologists
implement nondiscriminatory testing procedures (Committee on the
status of Racial Minorities, 1983; 1987). Nonetheless, one of the most
critical problems facing professionals in the field of speech-language
pathology is the absence of adequate assessment tools to provide a valid
and reliable evaluation of language varieties spoken by nonmainstreamed speakers (Taylor, 1986). Unfortunately, at the present
time, no widely accepted standardized techniques exist for assessing the
linguistic abilities of children and adults who speak nonstandard
dialects (Terrell & Terrell, 1983). In fact, many researchers (Peters, 1979;
Reveron, 1984; Seymour & Miller-Jones, 1981; and Vaughn-Cooke,
1980) have demonstrated that most assessment procedures in speechlanguage pathology are culturally- and linguistically-biased towards
middle class White Americans and biased against an overwhelming
number of Black Americans. In order to differentiate speech-language
disorders from speech-language differences, such as might be exhibited
by non-mainstreamed speakers, and to avoid systematic discrimination
practices that often result in a high proportion of African-American

children being diagnosed as speech or language impaired (Cole &
Taylor, 1990), accurate assessment procedures must be developed.
Various scholars and practitioners have proposed alternate
procedures to assess speech and language functions in nonmainstreamed speakers. These include: (1) standardizing existing tests
on non-mainstreamed English speakers; (2) one including a small
percentage of minorities in the standardization sample when
developing a test; (3) one utilizing a language sample when assessing
the language of non-mainstreamed speakers; (4) utilizing criterionreferenced measures when assessing the language of nonmainstreamed speakers; (5) refraining from using all standardized test
that have not been corrected for test bias when assessing the language
of non-mainstream speakers; (6) developing a new test which can
provide a more appropriate assessment of the language of nonmainstreamed English speakers; and (7) modifying or revising existing
tests in ways that will make them appropriate for non-mainstreamed
speakers, (Boyd & Vaughn-Cooke, 1983). As each of these alternatives
have inherent strengths and weakness, the reader is referred to
Vaughn-Cooke (1983) for a more thorough discussion. For the AfricanAmerican population, however, the most common assessment
alternatives proposed by the assessment researchers are modifying or
revising existing tests, utilizing criterion-referenced measures, and
developing new tests.

Characteristics of Black English
In order to modify, revise, or develop new tests, the
characteristics of the dialect in question must be examined. Since the
focus of this research is the development of linguistic realization in
African American children, the discussion of Black English will be
confined to the features of the sound system. Black English exhibits
two types of sound system variations: a variation within itself, from
region to region; and a variation between its sound system and that of
Standard English. Like standard English with its regional variable
rules for the presence versus absence of /r/ in /’-less types words (i.e.,
/bar/ versus /bah/), Black English also has its own regional variables
rules, such as the vocalic lowering /fa/ to £•/ in the South for words
like cheer and steer. Black English also displays a second type of
variation. Depending upon the perceived formality of the context
and/or audience, African-Americans may pronounce forms more like
those of Standard English, or less like those of standard English. Black
English, like other dialects then, is doubly varied, both within itself,
and in its closeness to Standard English (Moulton, 1976).
Extensive sociolinguistic research (Moulton,!976; Owens, 1988;
and Wolfram & Fasold, 1974) into the major phonological structures of
BEV has resulted in an inventory of potentially different phonological
processes in the BE speaker. The reader is referred to Appendix A for
an annotated inventory of the BE phonological processes and the rules
governing their phonetic realization.

Although this type of research illuminates the variable
differences in phoneme production of adult BE speakers from adult
Standard English speakers, it does not adequately describe the
developmental sequence of sound acquisition in children exposed to
Black English. To this end, Haynes and Moran (1989) researched final
consonant production in southern Black children from preschool
through third grade. Using the sound-in-words subtest of the
Goldman Fristoe Test of Articulation Revised (GFTA-R) (Goldman &
Fristoe, 1986) data was collected on the responses of 222 Black children
in rural east central Alabama who used the Black English dialect
common to that region. The children's thirteen phonological
processes were analyzed using the PROPH computer program. Overall,
the data suggested that southern Black children continued to delete
final consonants well beyond the age indicated by norms gathered on
predominantly White subjects. Both Grunwell (1982) and Ingram
(1976) suggested that the final consonant deletion process should be
eliminated by age three, in the largely White, standard English
speaking population that was studied. Given the contrast between the
results of the two studies, Haynes and Moran speculated that Black
children appeared to be on a different developmental timetable for the
acquisition of final consonant phonemes, as final consonant deletion
continued during the preschool and kindergarten years. Indeed, they
hypothesized that deletion of final consonants might persist beyond
the age of five with certain types of consonants, specifically voiced stops

and some fricatives being more susceptible to deletion. Although the
researchers speculated that omission of stops might be related to later
final consonant deletion and final consonant devoicing rules of the
adult model of Black English (i.e., /p/ for /b/f /t/ for /d/, and /k/ for
/g/), the deletion of fricatives /v/ and /•#•/ was more difficult to
interpret, given the lack of developmental data available for such
insights. Because it was of dubious value to compare Black English to
norms generated on White, middle-class children, and because neither
the existing norms, nor the adult model of Black English, described this
development, Haynes and Moran (1989) suggested that clinicians
needed to generate local norms for phonological process occurrence
and to compare children to these data as appropriate (Cole, 1983).
Haelsig and Madison (1986) completed research on the
phonological processes exhibited by three, four, and five-year-oldchildren of unspecified race. The subjects were fifty English speaking
children, 25 males and 25 females, ranging in age from 2-10 to 5-2. All
subjects were obtained from preschools and daycare centers. The
Phonological Process Analysis (PPA) (Weiner, 1979) was administered
to all subjects as specified in the manual. The PPA elicited 144
responses from which 16 phonological processes were examined. The
researchers found that although three-year-old children exhibited
errors involving three of 16 processes, 15 of the 16 processes were
present across this age group although, the distribution was uneven.
Ouster reduction was exhibited by all children; eight of the 10 subjects

demonstrated gliding of liquids, weak syllable deletion, glottal
replacement, labial assimilation, vocalization, and deletion of final
consonants. In contrast, final consonant devoicing, denasalization,
velar assimilation, prevocalic voicing, and gliding of fricatives
occurred in three or fewer subjects; affrication did not occur at all in
three year old subjects. Labial assimilation typically involved the
substitution of /•£/ for /£/ as in /Pnrfl/ for thumb.
The greatest reduction in the use of phonological processes
occurred between the three and four year olds. Deletion of final
consonants, stopping, fronting, and gliding of liquids were processes
whose frequency was reduced by 50% in the four year old subjects,
aiding of liquids was also the most frequently occurring process in
four year olds, but no longer occurred in the five year old group. Weak
syllable deletion, by contrast, was a frequently exhibited process in all
age groups. Glottal replacement of phonemes was infrequently used by
four and five year old children. Labial assimilation and cluster
reduction showed a similar pattern. Use of vocalization, stopping and
alveolar assimilation decreased systematically in the four and five year
old age groups.
From these findings Haelsig and Madison (1986) suggested that
those evaluating phonology should remember that there is a great deal
of individual variation in the three year old age group. Phonological
acquisition, whether defined in terms of individual phonemes or their
realization rules, varies from child to child.

Cole and Taylor (1990) studied the extent to which phonological
performance of a group of African-American children varied as a
function of test-client congruence on three tests of articulation
containing Standard English (SE) assumptions; and the extent to which
the children's test performance was likely to be misdiagnosed as being
pathological when dialect considerations were not taken into account.
Ten African-American children, five males and five females, ranging
in age from 5-11 to 6-11, served as subjects for this study. The subjects
were first graders enrolled in a public school in Lafayette County,
Mississippi. They were from a lower socioeconomic status as
determined by the schools' free lunch standards. The picture test of the
Templin-Darley Tests of Articulation Second Edition (Templin &
Darley 1969), the Arizona Articulation Proficiency Scale: Revised
(Fudula, 1974), and the Photo Articulation Test (Pendergast, Dicky,
Selman, & Sorder, 1969) were administered to each subject
individually. Two scores were assigned to each response. One score
was assigned from live elicitations in accordance with the rules of
Standard English or the stated criteria of each test. The second score
was based on judgments from both the live elicitations and the
audiotapes of the live elicitations for all items on which phonological
rules of BEV could be appropriately employed by the subjects.
The results showed the mean raw score obtained by the subjects
on the Arizona Articulation Proficiency Scale using SE norms was
90.9, while the mean raw score was 96.95 using BEV rules. According

to the norms of standard English, seven subjects met the test's criteria
for an articulation disorder, however, none of these same subjects was
considered as having an articulation disorder when BEV criteria was
employed. Similarly, the mean score obtained by the subjects on the
Templin Dariey Test of Articulation using standard English norms
was 125.2; however, the mean score increased to 132.6 when BEV rules
were employed. Six of the children were diagnosed as having an
articulation disorder when SE norms were used, but this number was
reduced to two when BEV rules were applied. Finally, the mean
number of errors obtained by the subjects on the Photo Articulation
Test was reduced from 5.2 using SE norms to 2.3 using BEV rules, and
the number of children meeting the test's criterion for articulation
disorders was reduced from three to one when BEV scoring criteria was
employed.
In general, these results corroborate Cole's (1986) claim that
African-American children perform differently on standardized tests of
articulation as a function of the linguistic norms used to score items.
Moreover, the results clearly showed that a failure to take the issue of
dialect variation into account substantially increases the likelihood of
misdiagnosing normally speaking African-American children as
having articulation disorders.

Statement of Problem
As the preceding research has shown, little normative data has
been published about the developmental phonology of Black English
speaking children, and the normative data that has been reported has
come from southern communities. Most descriptive studies of Black
English have focused on adults, adolescents and children who have
acquired the adult system. The few empirical studies of young Black
children's speech have drawn upon limited phonological samples, and
have isolated only a few features of Black English for investigation
(Seymour & Seymour, 1981).
The purpose of the present study is to fill the gap in the
normative literature on the development of sound acquisition and
phonological processing patterns in White and African-American five
and six-year old children matched for socioeconomic status.
Independent variables for the study are race, age, and phonetic feature,
phonological process, or canonical form.
Given the present trend in the research reviewed to date, two
primary hypotheses were formulated: 1) that five- and six-year old
Black children would perform significantly different than their White
peers with respect to speech production parameters; and 2) that fiveand six-year old Black children would evidence a significantly different
timetable of development with respect to speech production
parameters.

Chapter II
Method
Subjects
From a pool of 67 preschool children attending the Crawford
County Head Start Program in Meadville, Pennsylvania, 56 Black and
White children, ranging in age from five to six years served as subjects
for this study. Four 4-year olds were eliminated from the pool, as their
number was not felt to be sufficient to make constructive inferences
about their phonological development. One five year old was
eliminated also from the subject pool due to the fact that he could not
verbalize his receptive knowledge of the pictures used to elicit
phonemes for the study. Additionally 6 children, ranging in age from 5
to 6 years, who were receiving speech and/or language services at the
Head Start facility, were were eliminated from the subject pool.
All children serving as subjects ranged in age from 60 to 83
months, and were from low socioeconomic backgrounds, as classified
by the associate director of the Head Start Program. All children
resided in the industrial community of Meadville, Pennsylvania,
located in the Great Lakes' region of northwestern Pennsylvania. The
Meadville community represents a mixture of ethnic groups,
predominantly of Irish, Italian, and Polish decent, and more recently of
Hispanic and African-American heritage.

Examiners
In addition to the researcher, five graduate student speechlanguage clinicians from Edinboro University of Pennsylvania served
as examiners for this study. Examiners were White, Standard English
speakers, with experience in articulation test administration. All
examiners were supervised at the Head Start site by a licensed and
certified speech-language pathologist employed by Edinboro University
of Pennsylvania.

Materials
The sound-in-words subtest of the Goldman Fristoe Test of
Articulation Revised{GFTA-R) (Goldman & Fristoe, 1986) was used in
the study to elicit word-level production of 44 consonantal phonemes.
The GFTA-R was chosen for its ease of administration, its attention to
individual sound production, and its compatibility with phonological
process analysis protocols. The GFTA-R was normed on thirty-seven
articulatoiy defective children between the ages of four and eight years.
No information on the race of the children was given in the manual.
The sounds-in-words subtest consists of 35 pictures depicting objects
and activities that are familiar to young children. The subject is
required to name the pictures and to reply to questions about nine of
them, for a total of 44 responses. The 44 responses elicit consonantal
phonemes in each of three word positions: initial, medial, and final.

The Programs to Examine Phonetic and Phonological
Evaluation Records (PEPPER) (Shriberg, 1986) was used to analyze all
responses. The PEPPER program examines distinctive features and
phonological process patterns characteristic of speech responses, be they
syllables, words, or strings of words, for the age groups of concern in
this study.

Procedures
Each child came to one of four examination rooms with a "postit" sticker identifying his/her date of birth. If the child was participating
in facility-sponsored speech and/or language therapy, the addition of a
star accompanied the birthdate. Each clinician recorded the child's date
of birth, and the star designation, if appropriate. Each child was given
the sound-in-words subtest of the GFTA-R. The responses elicited
from each subject were broadly phonetically transcribed at the time of
administration by each of the five examiners, with audiotape back-up
of each test administration. The tape recordings were to be used later to
verify broad transcription, add narrow transcription, and to compute
inter-judge and intra-judge reliability estimates.
After verifying broad transcription, individual test sheets were
narrowly transcribed by the researcher using the narrow transcription
protocol of Shriberg and Kent (1982). Questionable narrow
transcription analysis was resolved through consultation with Mr.

John Boylan, CCC-SLP, a thesis committee member with expertise in
phonology.

Phonetic Inventory
Once narrow transcription of each subjects responses were
complete, the following subject groupings were formed: Group A1 White five-year -olds (N = 25); Group B1 - Black five-year olds (N = 12);
Group A2 - White six-year olds (N =14); and Group B2 - Black six-year
olds (N = 5) These data was entered into the PEPPER computer
program. Due to the PEPPER program's processing limitations of 12
subject entries per analysis, the groups were subdivided as follows. The
Black and White five-year-olds were analyzed in four month
sequences; such that Group A1 consisted of the following White fiveyear-old subgroups: Ala - ages 5-0 to 5-3 (N = 7); Alb - ages 5-4 to 5-8 (N
= 10); and Ale - ages 5-9 to 5-12 (N = 8); similarly, Group B1 consisted of
Black five-year-old subgroups: Bla - ages 5-0 to 5-3 (N = 6); Bib - ages
5-4 to 5-8 (N = 4); and Blc - ages 5-9 to 5-12 (N = 4). Six-year-old White
and six-year-old Black children were analyzed as separate individual
groups individual groups since the number of subjects in each group
did not exceed the 12 entry limitations of PEPPER Specifically, there
were 12 children in the white six-year-old group and 5 children in the
Black six-year-old group.
For each of the 56 subject responses entered, PEPPER performed
the following analyses: maintenance of canonical forms; phonetic
analysis by manner and place feature; and classification of the

percentage of deletions, substitutions, and distortions of phonemes in
each of the three positions; and usage of phonological process forms.
The percentage of accuracy the individual achieved with each of the
forms and features analyzed was generated by PEPPER

Inter- and Intra-Judge Agreement
Following the initial data collection period, the researcher
listened to the test administration recordings of four randomly selected
subjects from her test administration group and broadly retranscribed
the GFTA-R's sound-in words responses to establish intra-judge
agreement. Point-by-point percentage of agreement for English
phonetic transcription of the 44 elicited word responses were
computed, using the transcription reliability formula recommended by
Brenneise, Sarshad, Nicholas, and Brookshire (1991).

(Agreed-upon words/total words transcribed by examiner) X100.
For all 202 phonetic elements, point-by-point percent agreement
reliability was calculated for each phoneme using the following
formula:

[Total agreements/(Total agreements + Total disagreements)] X100.

Inter-judge reliability between the researcher and each of the five
other speech-language clinicians was established from four randomly
selected audiotaped test administrations utilizing the aforementioned
formulas.

Statistical Analyses
A series of 3-way analyses of variance (ANOVA) with fixed
effects were computed by the GANOVA-4 (Brecht, Woodward &
Bonett, 1986) computer program using a two-between-group onewithin-subject mixed design with alpha set at .05. The between-group
factors were always race (Factor A), with two levels, white subjects and
Black subjects, and age (Factor B), with four levels; children 5-0 to 5-3
years of age, 5-4 to 5-8 years of age, 5-5 to 5-12 years of age, and 6-0 to 6-12
years of age. The within-subject factor (Factor C) varied with manner
features (nasals, 3 levels; glides, 2 levels; stops, 6 levels; fricatives, 6
levels; affricatives, 2 levels; and liquids, 2 levels), place features
(bilabials, labiodentals, dentals, alveolars, palatals, and velars, all
having 3 levels for each sound position [initial, medial, and final], and
canonical form (6 levels). Reflecting the population sample under
study, the design was unbalanced with more total White subjects
(N=37) than Black subjects (N=19).

Chapter III
Results and Discussion
Inter- and Intra-Judge Agreement
Point-by-point percentage of agreement for English phonetic
transcription of the 44 elicited word responses by the researcher was
96.6%, and for all 202 phonetic elements, 99.25%. For each of the
student examiners, point-by-point percentage of agreement for the 44
elicited word responses ranged from 81.3% agreement to 99.5%
agreement, with a mean percentage of agreement of 91.94%. The one
examiner with the lowest percentage of transcription reliability (81.3%)
consistently used a incorrect vowel phoneme in her transcription;
specifically she transcribed / i / for /X / and j&J for /£ /. Similarly,
point-by-point percent of agreement for the 202 phonetic elements
transcribed by the student examiners ranged from 95.9% agreement to
98.6% agreement, with a mean percentage of agreement of 97.1%.

Statistical Analysis
Because compound symmetry was found to be significant (p
<.001) with all 3-way ANOVAs, the Wilks' Lamda (R) value, an
alternative multivariate test computed by GANOVA-4, was utilized for
all within-subject factors to determine the probability of significance.
Only those main effects and interactions found to be significant at the
.05 level of significance were further analyzed.
17

The Scheffe' Test formula for unbalanced designs was used for
post-hoc testing within-subject factors and between-within subject
interactions because of the lack of compound symmetry. The
Neuman-Keuls Test formula for unbalanced designs was used for posthoc testing the between portion and between group interactions of the
design (Maxwell & Bray, 1986). An estimate of

was computed for for

the three-way interactions to determine the percentage of variance
accounted for by the interaction All three-way interactions were found
to have a^f of 5% or less, and were not pursued with further analysis.
The results of significant main effects and first order interactions
follow.

Canonical Form
Table 1 presents the results of the three-way mixed AN OVA for
inbetween-groups factors of race and age, and within-subject factor of
canonical form, as computed by GANOVA-4. A main effect for Factor
Q canonical form, was found at the .006 level of significance (Wilks'
Lamda (R) = 3.844; df = 5, 44). Post-hoc analysis of the differences
between the canonical forms revealed that all children, independent of
race or age, evidenced significantly better performance with CVC forms
(99%) than with CnV forms (87.4%). Similarly, a significant interaction
of age and canonical form was found at the .007 level of significance (F
= 2.300; df = 15,121). Post-hoc analysis of the age factor across canonical

forms indicated that children between the ages of of 5-9 and 5-12,
independent of race, evidenced significant difficulty acquiring CnV
(78.1%) and CnVC (76.8%) forms, whereas children between the ages of
6-0 to 6-12, independent of race, were having significant difficulty
acquiring CnV (83.8%) and 3+ syllable (87.9%) forms (see Figure 1).

Table 1
Three-Way ANOVA for Retween-Croup Factors of Race and Age and
Within-Subject Factor of Canonical Form as Computed by GANOVA-4

Univariate Tests

F

OF

£

Main effect Factor A (race)

0.062

1,48

>.500

Main Effect Factor B (age)

0.447

3,48

>.500

Interaction of A and B

0.732

3,48

>.500

Alternate Multivariate Tests

Wilks'
DF
Lambda (R)

P

Main Effect Factor C (canonical form) 3.844

5,44

.006**

Interaction of A and C

1.130

5,44

.359

Interaction of B and C

2.300

15,121

.007**

Interaction of ABC

2,778

15,121

.001**

*p< .05. **p< .01
h
Note. For interaction ABC= -04

Figure 1. Interaction of Age x Canonical Form

80%

75%
la

lb

lc

2a

Note, o = CVC; x = CnV; U = CnVQ * = c(n)VCn; + = 2-syllable; and o = 3+ syllable
la = ages 5-0 to 5-3; ages lb = 5-4 to 5-8; ages lc = 5-9 lo 5-12; 2a age 6-2 to 6-12
The developmental literature suggests that the processes of
cluster reduction and final consonant deletion contribute to a high
proportion of CVC syllable shapes (Stoel-Gammon & Dunn, 1985).
Indeed, as the phonological process analysis reveals the children
studied in this sample persist in their usage of Final Consonant
Deletion (especially the White children) and Ouster Reduction. The
sample data also indicates that the difficulty with cluster reduction is
most prevalent in the initial position, with consonant sequence
reduction occurring in utterances such as /tov/ for /stov/, and /dum/
for /drum/ (Hodson & Paden, 1991)

Manner Features
Significant main effects for the following manner features were
found: nasals, glides, stops, and fricatives. No significant main effects
or interactions were found for liquids or affricatives, indicating that
neither races nor age groups differed significantly in their productions
of liquids or affricative.

Table 2
Three-way mixed A NOVA for Retween-Group Factors of Race and Age
and Within Suhject Factor of Nasals as Computed by HA NOVA-4
Univariate Tests

F

DF

P

Main Effect Factor A (race)

0.903

1,48

.349

Main Effect Factor B (age)

0.794

3,48

>.500

Interaction of A and B

2.643

3,48

.059

Alternate Mutivariate Tests

Wilks'
Lamda (R)

DF

P

Main Effect Factor C (nasals)

39.054

2,47

<.001**

Interaction of A and C

0.536

2,47

>.500

Interaction of B and C

0.871

6,94

>.500

Interaction of ABC

1.125

6,94

.354

*p = <.05; ** p = <.01
For the significant main effect of Factor C, nasals (/m/, /n/, and
)Y\ /), all the children sampled, independent of race or age, had a

3

significantly poorer performance with the phoneme /^ / (66.7%) than
with either /m/ (97.9%) or /n/ (99.1%). These data are consistent with
the developmental literature that reports that front nasals /m/ and /n/
are usually among the first sounds to be learned by Standard English
speaking children. The velar nasal /fj / is learned somewhat later. It is
frequently replaced by { n ), as in [swin] for swing or [win] for ring
(Edwards & Shriberg, 1983; Moulton, 1986).

Table 3
Three-way ANOVA for Between-Group Factors of Race and Age, and
Within Subject Factor of Glides as computed by CANOVA-4
(Univariate Tests)

F

DF

P

Main Effect Factor A (race)

0.219

1,48

>.500

Main Effect Factor B (age)

0.318

3,48

>.500

Interaction of A and B

2.615

3,48

.061

(Alternate Mutivariate Tests)

Wilks' Lam da (R)

DF

P

Main Effect Factor C (glides)

16.520

1,48

<.001**

Interaction of A and C

0.000

1,48

>.500

Interaction of B and C

0.189

3,48

>.500

Interaction of ABC

1.163

3,48

.333

*p = <.05; ** p = <.01

For the significant main effect of Factor C, glides (/w/, /j/) all the
children sampled, independent of race or age, had a significantly poorer
performance with the phoneme /j/ (76.8%) then the phoneme /w/
(95%.6). Edwards & Shriberg (1983) found that, in general, glides tend
to be replaced by liquids, as in the example

/ realized as

/1 &|o/, or front glides /w/ tends to replace the palatalized /j/ , as in the
example /jelo / realized as /u>elo/. They have found also that /r/ and
/!/ have stable positions during production, but that there tends to be
movement (i.e. gliding) from one articulatory position to another
during the production of /w/ and /j/. Edwards & Shriberg noted that,
in general, sounds produced in the front of the mouth are acquired
earlier than sounds produced in the back of the mouth. Before back
sounds are learned, they may be replaced by similar front sounds. This
was noted by the researcher in a frequent substitution of /w/ for /j/ in
transcriptions of children's responses.
A significant main effect for Factor C, stops, was found at p< .001
level of significance (Wilks' Lamda = 4.681; df = 5, 240). Post-hoc
analysis of the differences between stops cognates (/p-b/, /t-d/, and /kg/) revealed that all children sampled, independent of race or age, had
no significant difficulty producing stops, irregardless of voicing or place
of production features. Accuracy with all stops analyzed was found to
be greater than 90%. In order of production proficiency the following
trend was found: /b/ (99.6%); /d/ (97.9%); /p/ (96.3%); /g/ (94.5%); ft/
(91.9%); and /k/ (91.4%).

Table 4
Ihr6e-way ANOVA for Between-Group Factors of Race and Age, and
Within Subject Factor of Stops as
Univariate Tests

F

DF

P

Main Effect Factor A (race)

0.050

1,48

>.500

Main Effect Factor B (age)

0.175

3,48

>.500

Interaction of A and B

1.453

3,48

.238

Alternate Mutivariate Tests

Wilks'
Lamda (R)

DF

P

Main Effect Factor C (stops)

6.783

5,44

<.001**

Interaction of A and C

1.807

5,44

.131

Interaction of B and C

1.284

15,121

>.500

Interaction of ABC

1.222

15,121

.264

*p = <.05; ** p = <.01
As expected developmental data supported these findings in that
stops have been found to be acquired early. Macken (1980) reported
that there was no universal or invariant order of acquisition of stops,
although there were general patterns. In contrast, Ferguson (1975)
proposed an order of stop acquisition, from most to least stable: t > k >
b > d > p > g. In the present study, a different stop acquisition pattern
was revealed, from most to least stable: b>d>p>g>t>k, with the

children in the study having more difficulty with the /t/ and /k/
phonemes.

Table 5
Three-way ANOVA for Between-Group Factors of Race and Age and
Within Subject Factor of Fricatives as
Univariate Tests

F

DF

P

Main Effect Factor A (race)

1.161

1,48

.287

Main Effect Factor B (age)

0.293

3,48

>.500

Interaction of A and B

1.772

3,48

.164

Alternate Mutivariate Tests

Wilks'
Lamda (R)

DF

P

Main Effect Factor C (fricatives)

17.438

5,44

<.001**

Interaction of A and C

2.594

5,44

.038*

Interaction of B and C

1.361

15,121

.177

Interaction of ABC

2.056

15,121

.016*

*p = <.05; ** p = <.01
Note. ~y *= .04 for ABC Interaction
A significant main effect of Factor C, fricatives ( /f/, /v/, /O/,
/ &/, /s/, and /z/) (Wilks' Lamda (R) = 17.438; df = 5, 44; p< .001),
indicated that all children sampled, independent of race or age, had
significantly more difficulty with the /v/ (76%) than the / f/ (97%)

phoneme, but a significantly better performance with the /f-v/ (86.8%)
and /s-z/ (89.1%) cognate pairs than with the /&- %/ (60.1%) cognate
pairs. Similarly, a significant interaction of race and fricatives was
found at the .04 level of significance (Wilks' Lamda = 2.594; df = 5,44).
Post-hoc analysis of the race across fricative production (see Figure 2)
indicated that only the production of the phoneme /0*/ by Blacks
(43.5%) was significantly different from that of Whites (63.5%).
Figure Z Interaction of Race Across Fricative Production
100
95
90
85
80

75
70

65
60
55
50
45
40

lol
/ 0/
=
Black
Children
Note. X = White Children; O
If/

/v/

Acquisition of fricatives and affricatives is more complicated
than the acquisition of stops. Like stops, fricatives are acquired
gradually, and correct production often fluctuates with various types of
substitutes (Edwards, 1979; and Ferguson, 1975; ). There is no universal
order of acquisition for fricatives, but in general, front fricatives are
said to acquired before back fricatives. An exception is /©?, which is
acquired late. Ingram, Christensen, Veach, and Webster (1980) found
the following order of acquisition, based on percentage of correct
production, across 73 subjects: /f/, /s/, /v/, /z/, /&/. Similarly, the
present study found the same acquisition of fricatives for both Black
and White children, with the exception of / $y occurring before /-O'/,
which was not looked at in the Ingram et al. study.

Ferguson (1975)

also noted that before some fricatives are acquired, they may be
replaced by sounds of tighter closure (stops and affricatives). Fricatives
may also be omitted, especially in clusters in the word-final position.
Moulton (1976) stated that in Black English Vernacular the voiced "th"
(/#/) becomes /d/ in the initial position, so that then and den sound
alike. Medially, the voiced "th" (/#7)is realized as /d/ or /v/, so that
other and udder or either and eva sound alike. Similarly the
voiceless "th" (/$•/) in the initial position is realized as /f /or /v/, so
that think becomes fink or vink. Medially the voiceless "th" ( /& /) is
realized as /f/, making toothbrush into toofbrush. In the final
position, the voiceless "th" (/ft/) changes to /f/ or /t/, making fourth to
fort. Similarly, the present studied reveal that both Black and White

children are substituting stops /t/ and /d/ for fricatives /$/ and /$ /, as
Moulton found.
Place Features
Significant main effects for the following place features were
found: labio-dentals, dentals, alveiors, palatals, and velars. No
significant main effect or interactions were found for bilabials in any
word position, indicating that neither race nor age group differed
significantly in their production of bilabial phonemes.
Table 6
Three-way ANOVA for R etween - Crou p Factors of Race and Age and
Within Subject Factor of Iabio-Dentals as computed by GANOVA-4
Univariate Tests

F

DF

P

Main Effect Factor A (race)

1.016

1,48

.320

Main Effect Factor B (age)

0.694

3,48

>.500

Interaction of A and B

2.019

3,48

.123

Alternate Mutivariate Tests

Wilks' .
Lamda (R)

DF

P

Main Effect Factor C (labio-dentals)

9.764

2,47

<.001

Interaction of A and C

4.522

2,47

.016*

Interaction of B and C

2.176

6,94

.051

Interaction of ABC

2.527

6,94

.026*

*p = <.05; ** p = <-01
Note.'ty x= .04 for ABC Interaction

The significant main effect for Factor C, labio-dentals, was found
at p< -001 level of significance (Wilks' Lamda (R) = 97.6; df = 2, 47).
Post-hoc testing revealed that all the children sampled, independent of
race or age, had significantly poorer performance on labiodentals in the
medial position (80.4%) than in either the initial (93.5%) or final
(92.4%) positions. Similarly, a significant interaction of race and
labiodental production was found at the .02 level of significance
(Wilks' Lamda (R) = 4.522; df = 2,47). Post-hoc analysis of the race
across labiodental production (see Figure 3) indicated that Blacks
(72.3%) evidenced a significantly different performance from Whites
(88.6%) on labiodentals in the medial position.
Figure 3 Interaction of Race Across Labiodental Production
100
95
90
85
80
75
70

initial
position
Note.

medial
position

X = White Children; O = Black Children

final
position

Developmental data of sound positions from Edwards and
Shriberg (1983) indicate that, in general, sounds produced in the front
of the mouth are acquired earlier than sounds produced in the back of
the mouth. Before back sounds are learned, they may be replaced by
similar front sounds. It was stated that sounds are acquired first in the
initial position, then medial position, and then final. However, while
in the initial position does seem to favored stops, there is evidence that
fricatives may be acquired first in the postvocalic or final position.

The

results of this study indicated that Black and White child have acquired
acquisition of labiodentals in the initial and final positions, but are still
having difficulty with the medial position. Analysis of the data
collected for this study found a significant substitution of bilabials (/p/
and /b/) for labiodental sounds (/f / and /v/) in the medial position.
For example, children realized [telepone] for [telephone], thus
substituting the /p/ for /f/; and realized [shobel] for [shovel], thus
substituting the /b/ for /v/. Therefore, the present data indicated that
more of the Black children sampled are simplifying and/or substituting
labiodental fricatives in the medial position with bilabial stops.
The significant main effect for Factor C, dentals, was found at the
.004 level of significance (Wilks' Lamda (R) = 6.381; df = 2, 47). Post-hoc
analysis revealed that, independent of race or age, all children sampled
had significantly more difficulty with dentals in the final position
(512%) as compared to the initial position (67.5%). No significant

difference was found with dentals in the medial (53.8%) as compared
with the final position (51.2%).

Table 7
Three-way ANOVA for Between-Croup Factors of Race and Ager and
Within Subject Factor of Dentals as computed hv CIA NOVA-4
(Univariate Tests)

F

DF

P

Main Effect Factor A (race)

3.902

1,48

.056

Main Effect Factor B (age)

1.082

3,48

.366

Interaction of A and B

1.302

3,48

.284

(Alternate Mutivariate Tests)

Wilks'
Lamda (R)

DF

P

Main Effect Factor C (dentals)

6.381

2,47

<.004**

Interaction of A and C

1.614

2,47

.208

Interaction of B and C

1.666

6,94

.137

Interaction of ABC

1.567

6,94

.165

*p = <.05; ** p = <.01
Mouton (1976) found that the dental fricatives / $/ and /©*/ are
frequently replaced by Black children with the corresponding dental
stops /d/ and /t/. For example [thick] is pronounced as [tick! and [this]
is pronounced as [dis], or a dental fricative is replaced with the

corresponding voiced labial fricative /v/, so that [brother] is
pronounced as [bruwer]. However, is is noted that the replacement of
/e/ by /t/ or /f/, and the replacement of /#/ by /d/ or /v/, are by no
means unique to Black English. These substitutions are common
among nonstandard white speakers. The present study found the same
substitutions, with a higher percentage of /*0/ substitutions of /t/ or
/ f/, as in [bath] and [bathtub] for both Black and White children,
possibly accounting for the apparent difficulty evidenced with final and
medial sound positions.
The significant main effect of Factor C, alveolars (Wilks' Lamda
(R) = 6.858, df = 2, 47; /7= .003), found all children sampled, independent
of age or race, to have a significantly more accurate performance with
alveolars in the final position (95.9%) than in either the initial (90.9%)
or medial (91%) positions, as revealed by the post-hoc analysis. A race
by age interaction was found to be significant at the .04 level of
significance (F = 3.073; df = 3, 48). Post-hoc analysis indicated that
Blacks performed differently than Whites as a function of age (see
Figure 4).
Specifically, Blacks evidenced significantly better performance
with alveolar productions (95.9%) than whites (85.6%) at 5-0 to 5-3 years
of age and again at 5-9 to 5-12 years of age (Blacks = 99.5%; Whites =
86.3%). The performance trend, however, was dramatically reversed in
the 6-0 to 6-12 age group, with Whites (96.8%) evidencing significantly
better performance than Black (85.6%) with alveolar productions. An

examination of error patterns exhibited by Black Children in the 6-9 to
6-12 age group revealed that the majority of their errors were one of
substitution involving a velar phoneme (44%).

Table 8
Three-wav ANQVA for Between-Croup Factors of Race and Age and
Within Subject Factor of Alveolars as computed hv GANOVA-4
Univariate Tests

F

DF

P

Main Effect Factor A (race)

0.864

1,48

.360

Main Effect Factor B (age)

0.147

3,48

>.500

Interaction of A and B

3.073

3,48

.036*

Alternate Mutivariate Tests

Wilks'
Lamda (R)

DF

P

Main Effect Factor C (alveolars)

6.858

2,47

<.003**

Interaction of A and C

0.525

2,47

>.500

Interaction of B and C

1.542

6,94

.172

Interaction of ABC

3.298

6,94

.006**

*p = <.05; ** p = <01

Note.

= .05 for ABC Interaction

Figure 4 Interaction of Race across Age Groups
100
96
92
88

84
80

76 72

la

lb

lc

2a

Note. X = White children; O = Black children; la = ages 5-0 to 5-3; lb = ages 5-4 to 5-8;
lc ages 5-9 to 5-12; 2a = ages 6.2 to 6-12.

The main effect for Factor C, palatals, was found at p< .001 level
of significance (Wilks' Lamda = 8.566; df = 2, 47). Post-hoc analysis
revealed that all children sampled, independent of age or race, had
significantly better performance within palatals in the final position
(94.97e) than in either the initial (86.47.0 or medial (84.6%) positions.

Table 9
Three-way ANOVA for Between-Group Factors of Race and Age and
-Within Subject Factor of Palatals as computed hv GANOVA-4
Univariate Tests

F

DF

P

Main Effect Factor A (race)

1.849

1,48

.177

Main Effect Factor B (age)

1.057

3,48

.377

Interaction of A and B

0.950

3,48

.426

'Alternate Mutivariate Tests

Wilks'
Lamda (R)

DF

P

Main Effect Factor C (palatals)

8.566

2,47

<.001**

Interaction of A and C

1.922

2,47

.156

Interaction of B and C

1,722

6,94

.124

Interaction of ABC

0.729

6,94

>.500

*p = <.05; ** p = <.01
Similarly, the main effect for Factor C, velars, was found at p<
.001 level of significance (Wilks' Lamda (R) = 21.271; df = 2,47). Post hoc analysis revealed that all children sampled, independent of race or
age, had significantly more difficulty producing velars in the final
position (73.7%) as compared to either the medial (89.7%) or initial
(92.2%)positions.

Table 10
Three-wav ANOVA for Between-Group Factors nf Race and Ager and
Within Subject Factor nf Velars as computed by nANOVA-4
Univariate Tests

F

DF

P

Main Effect Factor A (race)

0.043

1,48

>.005

Main Effect Factor B (age)

0.208

3,48

>.500

Interaction of A and B

0.916

3,48

.443

Alternate Mutivariate Tests

Wilks'
Lamda (R)

DF

P

Main Effect Factor C (dentals)

21.271

2,47

<.001**

Interaction of A and C

1.378

2,47

.261

Interaction of B and C

0.466

6,94

>.500

Interaction of ABC

1.363

6,94

.237

*p = <.05; ** p = <.01
For both palatal and velar processes these data are consistent
with developmental data. In terms of the place of articulation, the
most common pattern of acquisition is front consonant before back
one: labial > alveolars > palatal > velar. Consonants acquired often
substitute for those learned later, e.g., alveolars for palatals
(depalatalization) (Stoel-Gammon & Dunn, 1985). The present study
revealed substitution patterns by both Black and White children of /s/
for lj / as in /sue/ for /shoe/, thus showing depalatalization of

alveolar phonemes for palatals. Another error of depalatalization that
occurred for both Black and White children was /t/ for /tf /, as in
/turch/ for /church/. In terms of velars, depalatization occurred with
/n/ for /(j /, thus simplifying a velar sound with an alveolar.

Phonological Processes
Seven naturally occurring processes were subjectively examined
across age and race to identify processes that the sample children were
still evidencing. Phonological processes provide a useful framework
for describing children's systematic simplifications of the adult
phonological systems.
Stoel-Gammon and Dunn (1985) group phonological processes
into three major categories; (1) processes that modify the syllabic
structure of the target word, such as Unstressed Syllable Deletion
(USD), Final Consonant Deletion (FCD), and Cluster Reduction (CR);
(2) processes that substitute on sound for another, such as Liquid
Simplification, Stopping, and Velar Fronting; and (3) processes that
assimilate one sound to another, such in Assimilation. AH of these
major categories of error patterns were found in the children sampled
and will be discussed in the same order as that was suggested by StoelGammon and Dunn.

Syllable Structure Processes
Unstressed syllable deletion, as the name suggests, is a process
that modifies the syllabic structures of the target word through deletion
of an unstressed syllable. As presented in Table 11, White children
were found to have noticeably more difficulty than their Black peers
with the process of unstressed syllable deletion during each of the three
earliest age ranges, with the process disappearing entirely in Black
children by 5-9 to 5-12. USD, however, appears to be a noticeable factor
in the performance of Black children in the later age range (6-9 to 6-12).
White children in this study have totally eliminated this process by 6-2
to 6-12, but for Black children in this sample the process reemerges.

Table 11
Children Evidencing Unstressed Syllable Deletion Process (USD)
White
Age

Black

Children

range

Number of
children
using USD

Percentage
of USD

6/7

86%

6/10

60%

5-9 to 5-12

3/8

6-9 to 6-12

0/12

38%
0%

5-0 to 5-3
5-4 to 5-8

Age

Children
range

5-0 to 5-3
5-4 to 5-8

Number of
children
using USD
1/6
1/4

17%
25%

5-9 to 5-12

0/4

0%

6-2 to 6-12

3/5

60%

Table 12 presents the pattern of final consonant deletion exhibited by
the sampled children.

Percentage
of USD

Table 12
Children exhibiting Final Consonant Deletion fFf.ni
White

Children
Number of
Age range children
using FCD
5-0 to 5-3
0/7
5-4 to 5-8
2/10
5-9 to 5-12
2/8
6-2 to 6-12
1/12

Black

Percentage
of FCD
0%
20%

25%
8%

Children
Number of
Age range children
using DFC
0/6
5-0 to 5-3
0/4
5-4 to 5-8
0/4
5-9 to 5-12
0/5
6-9 to 6-12

Percentage
of FCD

Although the percentages for FCD are not significantly large,
there does appear to be a consistent pattern of difference between the
White and Black children sampled in this study above the age range 5-0
to 5-3. White children consistently showed some percentage of FCD
process usage, while Black children consistently demonstrated no usage
of this process. The Black childrens' lack of FCD usage may be related
to the type of error pattern that Black children are exhibiting in final
consonant position; mainly that Black children constantly demonstrate
substitution errors, whereas White children here characteristically had
more errors of omission (deletion).
The last process of syllable modification found in the children
studied was Cluster Reduction. As is evident in Table 13. Black
children present noticeably less percentage of usage of the process of
cluster reduction during the early period of development, than their
White peers. Percentage scores tend to level out by age 6-2 to 6-12 with
both groups at this stage evidencing this process nearly half of the time.

0%
0%
0%
0%

Table 13
Children Exhibiting Clus.ter_Reduction Process (Cm
White Children
Black Children
Number of
Percentage
Number of
of CR
Age range children
Age range children
using CR
using CR
57%
4/7
5-0 to 5-3
5-0 to 5-3
1/6
50%
5/10
5-4 to 5-8
5-4 to 5-8
1/4
75%
6/8
5-9 to 5-12
5-9 to 5-12
1/4
42%
5/12
6-2 to 6-12
6-9 to 6-12
2/5

Percentage
of CR
17%
25%
25%
40%

Substitutions Processes
Substitution processes are those sound changes that substitute
on e class of sounds for another (Stoel-Gammon & Dunn, 1985). Table
14 presents the error pattern of Liquid Simplification evidenced by the
children sampled.

This process shows a mixed pattern of usage across

races and age groups, White children generally evidencing a higher
percentage of usage at all ages, with the exception of ages 5-0 to 5-3.

Table 14
Children exhibiting liquid simplification Process (LS)
Children
Number of
Age range children
using LS
5-0 to 5-3
1/7
5-4 to 5-8
4/10
5-9 to 5-12
3/8
6-2 to 6-12
5/12

Chi dren
Number of
Age range children
using LS
2/6
5-0 to 5-3
1/4
5-4 to 5-8
0/4
5-9 to 5-12
1/5
6-9 to 6-12
Black

White

Percentage
of LS
14%
40%
38%
42%

Table 15 contains the error pattern of stopping evidenced by the

Percentage
of LS
33%

25%
0%
20%

children sampled.
Table 15
White Children
Age
5-0
5-4
5-9
6-2

range
to
to
to
to

5-3
5-8
5-12
6t12

Number of
children
using ST
1/7
2/10
4/8
1/12

Black
Percentage
of ST
14%
20%

50%
8%

Age
5-0
5-4
5-9
6-9

Chi dren

range
to
to
to
to

5-3
5-8
5-12
6-12

Number of
children
using ST
3/6
3/4
2/4
3/5

Percentage
of ST
50%
75%
50%
60%

A noticeable difference exists between race groups with the usage of
this process, with Black children demonstrating consistently higher
usage of this process across all age ranges. Of further interest is the fact
that for this sample of children this was the only process where Black
children showed consistently higher percentages of usage scores across
all ages, than White children.
Finally, Table 16 depicts the Velar Fronting process pattern exhibited by
the sample of children studied.
Table 16
Children Exhibiting Velar Fronting Process (VF)
White Children
Black Children
Age

range

5-0 to 5-3
5-4 to 5-8

5-9 to 5-12
6-2 to 6-12

Number of
children
using VF
0/7
2/10
1/8

Percentage
of VF
0%
20%
13%

2/12

17%

Age

range

5-0 to 5-3
5-4 to 5-8
5-9 to 5-12

6-9 to 6-12

Number of
children
using VF
0/
1/4
0/4

1/5

Percentage
of VF
0%

25%
0%
20%

These data demonstrated that no characteristic differences were
apparent between race groups across the age range in this study. The
only age range that showed noticeable differences in percentage scores
were the 5-9 to 5-12 range, where White children showed some use of
this process, while Black children showed none.

Assimilation Process
Assimilation processes are sound changes in which one sound
becomes more similar to another. It was evidenced in this sample that
both Black and White children were using progressive assimilation,
which is classified as a sound which causes the assimilation to preceded
the affected sound. Table 16 contains the error patterns of Assimilation
evidenced by the children sampled.
Table 17
Children exhibiting Assimilation Process (AS)
White Children
Age

5-0
5-4
5-9
6-2

range

to
to
to
to

5-3
5-8
5-12
6-12

Number of
children
using
AS

1/7
0/10
0/8
0/12

Black
Percentage
of AS

14%
0%
0%
0%

Age

5-0
5-4
5-9
6-9

Children
range

to
to
to
to

5-3
5-8
5-12
6-12

Number of
children
using AS

2/6
0/4
0/4

1/5

Black children evidenced higher percentage of usage at the
earliest (5-0 to 5-3) and latest (6-9 to 6-12) age ranges. Percentage of
usage among Black children at the earliest age range is nearly twice that

Percentage
of AS

33%
0%
0%
20%

of White children of the same age. In addition, it is interesting to note
that Black children are still evidencing this process at the latest age
range, while the process is totally absent from White children in the
same age range.
Although the developmental data on phonological processing
indicates that "normal" children no longer evidenced unstressed
syllable deletion, final consonant deletion, velar fronting, and
consonant assimilation after age three (Stoel Gammon & Dunn, 1985),
the children sampled in this study were still evidencing these and
other processes well into their fifth and sixth year. The pattern of
phonological process usage often differed by race. Black children did
not evidence final consonant deletion, but in contrast White children
did exhibit this process from ages 5-4 to 5-12. Black children exhibited
the process of assimilation at ages 5-0 to 5-4, and again occurring at age
6-2. White children evidenced this process only at ages 5-0 to 5-3. Black
and White children were relatively equal in pattern distribution
distribution velar fronting and unstressed syllable deletion.

Conclusion
The results of this comparative analysis of developmental
realization of canonical form, phonetic features of manner and place,
and phonological processes, in five and six-year-old White and Black
children revealed that these children produced essentially the same

type of errors, but that Black children produced some error types,
especially at ages 6-9 to 6-12, that characterized Black English
phonology. The apparent influence of Black English on
developmental consonants production accounted for only a slightly
different error pattern among place and manner, and phonological
processing features.
Despite dialectal background, the five and six-year old Black and
White children in this sample produced the same developmental
errors. These findings agreed with those of Seymour & Seymour (1981)
who conducted a similar study and found that consonant features that
contrast between Black English and Standard English were
undifferentiated from developmental errors typical of white children
who range in age from three to eight years of age (Ingram, 1976; and
Templin, 1957). Consequently, articulation testing of Black children
who have not yet mastered the adult Black English system would likely
yield developmental errors consistent with normative data on
standard American English. Indeed, Owens (1988) suggests that BE
phonological features are not likely to emerge until around the age that
Black children enter school when peer interaction becomes an
important source of dialectal input.
Nonetheless, it is hypothesized that some differences in the
speech of young Black children and White children might be
significantly greater than this study revealed, especially if spontaneous
speech samples from naturalistic settings of the two groups were

analyzed in a fashion similar to the one presented in this study. Single
word utterances are not believed to capture the extent to which
canonical form and phonological processing might differ in these two
groups. The single word elicitation technique of many articulation test
formats ignores certain important dynamics of articulation such as co­
articulation, prosody, situational, and linguistic variance.
The use of the formal articulation format in gathering the data,
however, was meaningful in that the researcher was able to assess and
contrast consonantal responses in a typical testing procedure. The
similarity of error types among children, and the comparability of
articulatory performance with existing normative data, suggest that
Black and White childrens' consonantal productions in single word
responses in the 5- to 6-year-old range may not be significantly
different. The speech-language pathologist assessing single word
articulatory proficiency in African-American children should be
mindful that preschool Black children are likely to produce
developmental errors consistent with those of standard American
English. Not until the latter part of the sixth year should features of BE
phonology be expected to emerge.

References
Brenneise-Sarshad, R., Nicholas, L, & Brookshire, R (1991). Effects of
apparent listener knowledge and picture stimuli on aphasic and non­
brain damaged speakers narrative discourse. JSHR 34(1), 168-176.
Boyd, F., & Vaughn-Cooke, A F. (1983). Improving language
assessment in minority children. ASHA, 22, 29-33.
Brecht, M., Woodward, ]., Bonett, D. (1986).
Liner Models for Experimental Designs. Los Angeles, CA
Department of Psychology, UCLA.
Cole, L (1983) Implications of the position on soical dialects. ASHA, 25,
(9), 25-27.
Cole, L, & Taylor, 0., (1990). Performance of working class AfricanAmerican
children on three tests of articulation. ASHA, 21,171-176.
Committee on Language, Speech and Hearing Services in the Schools
(1982). Definitions: Communicative disorders and variations. ASHA
24 (ID, 949-950.
Committee on the Status of Racial Minorities (1983). Social dialects
position paper. ASHA 22 (U, 45.
Committee on Status of Racial Minorities (1987). Social dialects
position Paper. ASHA 22 (1), 45.
Dillard J. L (1972). Blank Fnglish. New York: Random House.
Edwards, M., & Shriberg, L (1983)

in

Communicative Disorders San Diego: College Hill Press.
Goldman, R. & Fristoe, M. (1986). Goldman Fristoe Test of
Articulation -

: American Guidance

Service.
Grunwell, P. (1982) Clinical Phonology. Aspen, Rockville, MD
Haelsig, Pv & Madison, C (1986) A study of phonological processes
exhibited by 3-, 4-, and 5-year-old children. Language, Speech, and
Hearing Services in Schools, 17, 107-114.
Haynes, W., & Moran, M. (1989). A cross sectional development study
of final consonant production in southern Black children from

Services in Schools, 20, 400-406.
Hodson, B. & Paden, E. (1981) Phonological processing patterns which
characterized unintelligible and intelligible speech in early
childhood. Speech and Hearing Disorders, 46, 369-373.
Ingram, D. (1976) Phonological disability in children. New York:
Labov, W. (1972). language in the inner city: Studies in the Black
English Vernacular. Philadelphia: University of Penn. Press.
Maxwell, C., & Bray, S. (1986). Robustness of the quasi-F through
violations of spherecity Psychological BulletioiL 22, 416-421.
Moulton, W. (1976). The sounds of Black English. In D. Harrison & T.
Trabasso (Eds.), Black English a seminar (pp. 149-170). Hillsdale, NJ:
Lawrence Erlbaum Associates.
Owens, R (1988). Language Development An Introduction,

Geneseo, New York Merrill Publishing Company.
Pendergast, K, Dickey, Stalley E., Selman, J. W., & Sorder, A. L (1969).
Photo Articulation Test. Illinois: Interstate.
Seymour, H. N., & Miller-Jones, D. (1981). Language and cognitive
assessment of Black children. Speech and language: Advances in basic
research and practice, 6, 203-263.
Seymour, H., & Seymour, C, (1981). Black English and Standard
American English contrasts in consonantal development of four and
five-year old children. Journal of Speech and Hearing Disorders, 2,
274-280.
Shriberg L (1986) Programs to Examine Phonological Evaluation of

Stockman, I. (1986). Language acquisition in culturally diverse
populations: The Black child as a case study. In O. Taylor (Ed.).
Nature of Communication disorders in culturally & linguistically
diverse populations (pp. 117-156). San Diego, CA College-Hill Press.
Taylor, O. (1986). The challenge of assessing the language of
nonmainstreamed speakers. In O. Taylor (Ed.). Treatment of.
communication Disorders in culturally and linguistically diverse
populations (pp. 23-48). San Diego, CA College-Hill Press.
Templin, M. C., & Darley, F. L. (1969).
Articulation. (2nd ed.). Iowa City: Bureau of Educational Research
and Service, Division of Extension and University Services.
University of Iowa.

Terrell, S. L, & Terrell, F. (1983) Distinguishing linguistic differences
from disorders: The past, present, and future of
nonbiased assessment. Topics in language Disorders, 3,1-7.
Vaughn-Cooke, A F. (1980). Evaluating the language of Black English
speakers: Implications of the Ann Arbor decision. In M. F.
Whiteman (Ed.), Reactions to Ann Arbor decision: Vernacular Black
English and education. Washington, DC Center for Applied
Linguistics.
Wolfram, W. & Fasold, K W. (1977) The study of social dialects In
American English. Englewood Cliffs, NJ: Prentice-Hall

!

APPENDIX A

<

Characteristics of Black English Phonology
I. Plosives:
At the end of a syllable, voiced stops, /b/, /d/, and /g/, are
pronounced as the corresponding voiceless stops /p/, /1/, and /k/.
This does not mean that bud and butt, and cab and cap sound alike in
BEV. They are distinguished by length of the vowel, being held slightly
longer when the following sound is voiced. For example, the "u" in
bud is pronounced longer

while the "u" in butt, could be

pronounced /A/.

IL Nasals:
In Black English Vernacular there is a tendency for nasals not
to be realized in the final position. For example, /com/ becomes /co/.
Also, the "ng" becomes /n/ in the medial and final position so that
sing and sin may sound alike, and in an unstressed syllable, as in
swimmin for swimming or buyin for buying.

III. Fricatives:
In Black English Vernacular, the voiced "th" (/$/) becomes /d/ in the
initial position, so that then and den sound alike. Medially, the voiced
"th" (/$/) is realized as /d/ or /v/, so that other and udder or either
and eva sound alike. Similarly, the voiceless "th" (/"0”/) in the initial
position is realized as /f /or /v/ so that think becomes fink or vink.
Medially the voiceless "th" ( ft /) is realized as /f/, making toothbrush

into toofbrush. In the final position, the voiceless "th" (/ft/) changes to
/f/ or/t/, making fourth to fort.

IV. liquids: /r/
In Black English Vernacular, /r / is not realized between two
vowels, so that Carol is realized as Ca'ol. Also, /r/ is not realized in
the final position or preceding a consonant, thus bar is realized as bah.
The liquid /!/ may not be realized before /t/,/d/,or/p/, therefore,
help is realized as hep, and colt is realized as code. Also, /l/ may not be
realized in the final positions, thus toll and toe may sound alike.

V. Syllable Reduction:
Word final consonant clusters in a stop can be reduced when
both phonemes belong to a base word: tes (test), des (desk), han (hand),
and buil (build). Words ending in a consonant are reduced when
grammatical suffix-ed is added, so that rubbed becomes rub.