REPORT RESUMES
ED 011 612 24
SUBCULTURAL DIFFERENCES IN CHILDREN'S LANGUAGE DEVELOPMENT.

ENTWISLE, DORIS R.
JOHNS HOPKINS UNIV., BALTIMORE, MD.
REPORT NUMBER BR- 6- 1610 -2 PUB DATE MAY 67

GRANT OEG2..7061610...0207
ECRS PRICE MFS0.09 HC -$1.16 29P.

DESCRIPTORS- *VERBAL DEVELOPMENT, *RURAL URBAN DIFFERENCES,
RACIAL DIFFERENCES, *LANGUAGE DEVELOPMENT, *ENVIRONMENTAL
INFLUENCES, LANGUAGE.ABILITY, *SOCIOLINGUISTICS, LINGUISTICS,
INTERVIEWS, SOCIOECONOMIC STATUS, CULTURALLY DISADVANTAGED,
BALTIMORE

PATTERNS OF THE LINGUISTIC DEVELOPMENT OF CHILDREN OF
DIFFERENT SOCIOECONOMIC ENVIRONMENTS WERE DETERMINED BY A
STUDY OF WORD ASSOCIATIONS. THE RELATION OF RESIDENTIAL AREA,
SOCIAL CLASS, OR SUBCULTURAL GROUP MEMBERSHIP TO LINGUISTIC
DEVELOPMENT WAS THE MAIN CONCERN OF THE STUDY. EACH
MEMBERSHIP GROUP WAS FURTHER CATEGORIZED ACCORDING TO IQ
LEVEL, SEX, AND GRADE. GROUPS WERE COMPARED BY HOLDING
CONSTANT AGE AND IQ. THE WORD ASSOCIATIONS WERE OBTAINED IN
RESPONSE TO A LIST OF 96 STIMULUS WORDS. EACH CHILD. WAS
INTERVIEWED ALONE, AND WAS ASKED TO RESPOND WITH THE FIRST
WORD THOUGHT OF AS THE INTERVIEWER SAID A WORD ALOUD. RESULTS
OF THE STUDY SUPPORT THE FOLLOWING CONCLUSIONS - -(1) THERE ARE
NEGLIGIBLE DIFFERENCES. BETWEEN SURBURSAN CHILDREN FROM UPPER
MIDDLE CLASS AND BLUE COLLAR NEIGHBORHOOCS, (2) RURAL
MARYLAND CHILDREN TEND TO DEVELOP MORE SLOWLY THAN THE
SUBURBAN CHILDREN, ESPECIALLY THOSE WHOSE IQ IS AVERAGE OR
BELOW, (3) AMISH CHILDREN DEVELOP EVEN MORE SLOWLY THAN THE
RURAL MARYLAND CHILDREN, AND (4) WHITE SLUM CHILDREN ARE
ADVANCED COMPARED TO SUBURBAN CHILDREN AT FIRST GRADE, BUT
RETARDED AT THIRD GRADE. NEGRO SLUM CHILDREN ARE GENERALLY
BEHIND WHITE SLUM CHILDREN, BUT AT FIRST GRADE THE NEGRO SLUM
CHILDREN ARE ON A PAR WITH WHITE SUBURBAN CHILDREN. RELATED
'REPORTS ARE AA 000 046 AND AA 000 046. (AL)



THE JOHNS HOPKINS UNIVERSITY

Subcultural Differences In Children's

Language Development
1

Doris R. Entwisle
2

The Johns Hopkins University



U. S. DEPARTMENT OF HEALTH, EDUCATION AND WELFARE
Office of Education

This document has been reproduced exactly as received from the
person or organization originating it. Points of view of opinionsstated do not necessarily, represent official Office of Education
Position or policy

Subcultural Differences In Children's

Language Development
1

Doris R. Entwisle
2

The Johns Hopkins University



Subcultural Differences In Children's
Language Development

Project No. 6-1610
Contract No. OEG-2-7-061610-0207

Doris R. Entwisle

May 1967

The research reported herein was performed pursuant to a grant
with the Office of Education, U.S. Department of Health, Education
and Welfare. Contractors undertaking such projects under Govern-
ment sponsorship are encouraged to express freely their professional
judgment in the conduct of the project. Points of view or opinions
stated do not, therefore, necessarily represent official Office of
Education position or policy.

The Johas Hopkins University
Baltimore, Maryland



CONTENTS

Acknowledgments

I. Introduction 1

II. Method 1

III. Results 6

School Differences 9

Racial Differences 10

Urbanization 12

IV. Discussion 16

V. Summary 18

References Cited 19

Footnotes 22

Map (Pennsylvania - Maryland) "Geographical 3

Locations of Sample Clusters"

Tables

Table 1. Design of Clusters of Children Living

In and Around Baltimore, Maryland

Table 2. Percentage of Paradigmatic Responses

4,5

7,8



ACKNOWLEDGMENTS

The preparation of this paper was supported by The Center for
the Study of Social Organization of Schools, The Johns Hopkins
University. Most of the Research upon which it is based was
supported by the National Institute of Child Health and Human
Development, grant HD-00921.

I am grateful to James Coleman, Robert Gordon, Edward McDill,
and Arthur Stinchcombe for critical reading of this paper.



I. INTRODUCTION

Word associations, although long a topic for psychological

inquiry, have only within the last decade assumed prominence among

the tools of developmental psycholinguistics. Two pioneering studies,

one by Brown and Berko (1960) and another by Ervin (1961), show cor-

relation between the prevalence of paradigmatic associations and the

grammatical competence of the child. Independent evidence of the

validity of word associations as an index of language maturity is that

children's free associations are highly related to word comprehension
and ability to place words in appropriate contexts (Riegel, Riegel,

Smith and Quarterman, 1964). Also a high correlation is observedbe-
tween semantic meanings and primary associates (Di Vesta, 1966). Other

studies (Entwisle, Forsyth, and Muuss, 1964; Entwisle 1966a, 1966b,

1967a, 1967b) confirm both the sharp rise in paradigmaties over ages

5 to 8 and the replacement of syntagmatic with paradigmatic associates.

(The syntagmatic-paradigmatic shift, although discovered 50 years ago

(Woodrow and Lowell, 1916), has received little attention until lately.)

High-speed data-processing equipment has opened the door to cross-cul-

tural study of linguistic development using word associations of chil-

dren from different social classes and different cultural groups. This

paper reports subcultural variations in language development for chil-

dren from 4 to 10 years of age, and speculates upon the socialization

and cultural factors that spawn the differences. It is not very clear

what specific environmental factors are important in verbal development

and the major concern here is to elucidate those factors.

This research, which began in 1961, was undertaken partly to pro-
vide normative word association data for sizeable groups of young chil-

dren, and partly to shed light on the course of linguistic development.

The present research differs from other studies of associations in that

the intelligence level of various samples of children is specified

rather closely and the socio-cultural environment, including the school

milieu, is described in detail. As expected, intelligence (measured
IQ) is highly related Lo appearance of paradigmatic associates, espe-
cially for the younger children. Intelligence is of secondary interest

here, however. Residential locus, social class, or subcultural group
membership, and their relation to linguistic development is the main
focus of this paper. Insofar as possible, social class is conceived
in fairly specific terms--median income, father's educational level,
quality of school attended, housing conditions, and so forth.

II. METHOD

Several different groups of children at first-, third-, and fifth-
grade have been studied. Rural American children, both Amish-Pennsyl-
vanian and a cross-section of typical Maryland farm children, and some

rural German children, are represented. Urban Baltimore children of
three different socioeconomic levels, upper middle class, blue collar,

and slum, are also represented. Baltimore City has a population of

1



about one-million persons. Entirely surrounding the City is Baltimore
County, and the City and County have separate school systems. The
racial composition of schools is vastly different, with about 3 per
cent Negro pupils in the County compared to 60 per cent Negro in the
City, but in neither place are Negroes homogeneously distributed. For
upper middle and working class, only white children are studied and
they live in the suburbs (Baltimore County). For the lowest socio-
economic group, slum residents of Baltimore City, equal numbers of
Negro and white children were procured.

Each residential or socioeconomic group is further categorized
according to IQ level, sex, and grade, and groups can be compared
holding constant age and IQ. Three IQ strata are defined: Low IQ (85
or below); medium IQ (95-105); and high IQ (122 or higher). IQ data
were mostly taken from school records. Not all IQ strata are repre-
sented in all residential groups. For instance, there are no "high IQ"
slum children. Background data on the different groups are given in
Table 1 which displays the incomplete factorial design of the sampling
plan. Figure 1 shows geographical locations of sample clusters.

The Amish, a distinct subcultural group who develop in a con-
siderably different cultural stream from rural Maryland children, re-
side on farms north of the Maryland-Pennsylvania line (see Fig. 1).
They are descended from Anabaptist immigrants of 200 years ago, and
are of special interest because religious scruples cause them to iso-
late themselves from TV sets, radios, and other mass media. Amish
child-rearing practices also reduce exposure to spoken language be-
cause conversation between adults and children is limited and educa-
tion is not valued. These children are not economically deprived,
for the farms are well-run and prosperous, but automobiles and other
"luxuries" are fnrbidden. The Amish children were interviewed in
exactly the same way as the Maryland children, with an Amish woman
serving as interviewer.

Briefly, 96 stimulus words, chosen to represent different form
classes and, where possible, different degrees of rarity, are used to
elicit associations. The Thorndike-Lorge J count was used to estimate
frequency ( Thorndike and Lorge, 1944). For nouns, adjectives, and
verbs, stimuli are divided into 3 strata, one with frequency greater
than 1000, a second with frequency 500 to 1000, and a third with
frequency below 500. No frequency division was possible for adverbs
and pronouns. A young woman interviewed children one at a time, saying
aloud each word on the list and then recording the child's response.
The child is instructed that he is going to play a "word game", and
when the interviewer says a word aloud, he is to respond with the first
word he thinks of. Details of administration are given in Entwisle
(1966b).

2



FIGURE I

I Urban Slum
2 Low SES Suburban
3 High SES Suburban
4 Rural Maryland
5 Rural Amish



TABLE 1

Design of Clusters of Children Living In and Around Baltimore, Md.

Suburban White Suburban White
High-SES (Income
est. $9171 p.a.)

Low-SES (Income
est. $6219 p.a.)

Grade High-IQ Med.-IQ Med.-IQ Low-IQ

Kindergarten

N 100 100

Mean Age 5.6 56
Mean IQ 125 100

Mean grade
completed by father

16.6a 12.5
b

Grade 1

N 70 70 70 70

Mean Age 6.4 6.6 6.5 6.7

Mean IQ 129.9 102.6 102.8 75.0
Mean grade
completed by father

15.5 13.8 10.5 9.6

Grade 3

N 70 70 70 70

Mean Age 8.5 8.7 8.7 9.0

Mean IQ 131.5 100.5 100.4 79.8

Mean grade
completed by father

15.6 13.4 9.9 8.7

Grade 5

N 70 70 70 70

Mean Age 10.4 10.7 10.6 10.11
Mean IQ 131.0 100.8 100.1 79.7
Mean grade
completed by father

14.7 13.0 10.1 9.7

a. for ninety-nine subjects
b. for eighty-eight subjects

4

.....wwNtVlnlmMlnAWrWOWprmwmpgwpkimp.....p.w...y



TABLE 1 (cont.)

Baltimore slum
Rural Maryland Amish White Negro

High Med. Low Med. Low Med. Low
Grade IQ IQ IQ IQ IQ IQ IQ IQ

Grade 1

N 20 20 20 20 37 40 29 40
Mean Age 6.6 6.7 6.9 6.9

c
7.0 7.3 6.5 - 6.9

Mean IQ 128.1 99.8 77.4 85.3 997 79.8. 98.1 79.9.
Mean grade 12.9 10.4 7.5 8.0 10.3

d
8.1e 10.5

f
8.3g

completed
by father

Grade 3

N 20 20 19 20 39 40 40 36
Mean Age 8.6 8.8 9.2 8.7 8.9 10.1 8.6 9.7
Mean IQ 131.9 99.6 78.3 92.8 100.0

h
80.7. 99.7. 80.9

Mean grade 12.2 9.0 8.5 8.0 8.2 8.7 9.43 8.2k
completed
by father

Grade 5

N 20 20 20 20 40 40 40 40
Mean Age 10.4 10.7 11.7 10.9 10,11 11.7 10.9 11.8.
Mean IQ 131.9 100.1 77.1 101.1 99.7 80.5 99.5 80.6
Mean grade 12.3 9.7 7.8 8.0 9.9

1
8.gm 7.3n 8.2

(3

completed
by father

c. for two of the twenty children IQ scores were not. available .
d. for seven subjects
e. for fifteen subjects
f. for ten subjects
g. for six subjects
h. for seventeen subjects
i. for six subjects
i. for five subjects
k. for six subjects
1. for ten subjects
m. air eleven subjects
n. for three subjects
o. for nine subjects

5



r.

III. RESULTS

Relative Positions of Subcultural Groups. Word associations were
analyzed in a number of ways (see Entwisle, 1966a, 1966b, 1967a,
1967b), but the percentage of paradigmatic responses in the various
sample strata offers the most convenient way to study relative rates
of development (see Table 2). Analyses of variance of paradigmatic
races support ale following conclusions about relative rates of devel-
opment. (All comparisons assume children of the same IQ level.)

(1) There are negligible differences between suburban children
from upper middle class and blue collar neighborhoods.

(2) Rural Maryland children tend to develop more slowly than
the suburban children, especially those who fall in the lower two IQ
strata.

(3) Amish children develop even more slowly than the rural
Maryland children.

(4) White slum children are advanced compared to suburban chil-
dren at first-grade, but retarded at third-grade. Negro slum children
are generally behind white slum children, but at first-grade the Negro
slum children are on a par with white suburban children.

These observations in American children suggest a relation be-
tween rate of development and degree of urbanization. At first-grade
the children rank as follows: white slum children, Negro slum chil-
dren and suburban middle class and working class children of both SES
levels, rural Maryland children, and Amish children residing in rural
Pennsylvania. Study of rural German children living in Schleswig-
Holstein (Entwisle and Muuss, 1967) and of Parisian children (Rosen-
zweig and Menahem, 1962) suggests a similar rural-urban difference in
non-English-speaking children.

Appearance and prevalence of paradigmatic responses is taken as
an index of linguistic development, and rates of development among
various groups are compared. Differences among groups have been
pointedup, but these are related to timing rather than to qualitative
factors. It should be emphasized that the sequence of changes and the
eventual level of paradigmatic responding are theosame for all groups.
The paradigmatic response rate is a rough measure and only appropriate
over early childhood. Other more refined measures would probably
reveal differences at ages 10-11 when paradigmatic response rates have
attained fairly constant levels. The increase in paradigmatic re-
sponses, the increase in commonality or percentage of primaries, and
the diminution in syntagmatic responses are common features in all
groups. Pronouns and adjectives develop sooner than verbs and adverbs,
and there is only a small increase for nouns. The overall picture
looks much the same for all groups and the process itself seems in-
variant. Some children just take longer than others to pass through
the sequence of changes.

6

,.....a,11RIPM.nown,r



S
u
b
u
r
b
a
n
 
B
a
l
t
i
m
o
r
e
 
C
h
i
l
d
r
e
n
 
1
9
6
1
-
6
3

K
i
n
d
e
r
g
a
r
t
e
n

F
i
r
s
t
-
g
r
a
d
e

T
h
i
r
d
-
g
r
a
d
e

H
i
g
h
-
 
L
o
w
-

H
i
g
h
-
S
E
S

L
o
w
-
S
E
S

H
i
g
h
-
S
E
S

L
o
w
-
S
E
S

F
i
f
t
h
-
g
r
a
d
e

H
i
g
h
-
S
E
S

L
o
w
-
S
E
S

S
E
S

S
E
S

H
i
g
h

I
Q

M
e
d
.

I
Q

M
e
d
.

I
Q

L
o
w
I
Q

H
i
g
h

I
Q

M
e
d
.

I
Q

M
e
d
.

I
Q

L
o
w

I
Q

H
i
g
h

I
Q

M
e
d
.

I
Q

M
e
d
.

I
Q

L
o
w

I
Q
'

N
o
u
n
s

5
6
.
3

6
6
.
1

6
6
.
8

6
5
.
1

5
9
.
0

5
9
.
7

7
4
.
8

7
1
.
8

7
3
.
1

7
0
.
7

7
9
.
3

7
8
.
2

7
7
.
1

7
6
.
3

A
d
j
.

1
6
.
7

1
6
.
9

4
5
.
7

2
9
.
3

3
0
.
1

2
1
.
7

7
4
.
2

7
1
.
4

7
3
.
3

6
2
.
8

8
1
.
0

7
8
.
2

7
8
.
1

7
7
.
1

V
e
r
b
s

1
6
.
7

1
6
.
6

2
2
.
4

2
1
.
1

1
8
.
7

1
9
.
2

5
7
.
1

4
7
.
4

4
9
.
9

3
5
.
5

6
6
.
6

6
0
.
2

5
7
.
0

5
3
.
6

A
d
v
e
r
b
s

9
.
8

7
.
3

3
0
.
9

1
5
.
5

1
6
.
4

8
.
8

6
3
.
0

5
0
.
2

4
9
.
1

4
2
.
7

7
0
.
0

6
5
.
2

5
9
.
8

5
3
.
6

P
r
o
n
o
u
n
s

1
6
.
6

1
6
.
4

4
3
.
0

3
0
.
2

2
7
.
0

1
9
.
1

6
4
.
6

6
7
.
3

6
4
.
8

6
4
.
6

7
2
.
0

6
3
.
6

5
7
.
5

6
7
.
7

F
i
r
s
t
-
g
r
a
d
e

T
h
i
r
d
-
g
r
a
d
e

F
i
f
t
h
-
.
g
r
a
d
e

R
u
r
a
l
 
M
a
r
y
l
a
n
d

A
m
i
s
h

R
u
r
a
l
 
M
a
r
y
l
a
n
d

A
m
i
s
h

R
u
r
a
l
 
M
a
r
y
l
a
n
d

A
m
i
s
h

H
i
g
h

I
Q

M
e
d
.

I
q
_
_

L
o
w
I
Q

I
Q

H
i
g
h

I
Q

M
e
d
.

I
Q

L
o
w
I
Q

I
Q

H
i
g
h

I
g

M
e
d
,

I
Q

L
o
w
I
Q

I
Q

N
o
u
n
s

7
1
.
7

6
2
.
3

5
7
.
7

4
7
.
7

8
0
.
0

6
9
.
4

6
8
.
5

6
6
.
3

7
8
.
9

7
8
.
9

7
9
.
6

7
3
.
3

A
d
j
e
c
t
i
v
e
s

4
0
.
4

2
1
.
3
.

1
3
.
9

1
5
.
4

6
9
.
4

6
7
.
5

6
1
.
5

4
8
.
8

8
2
.
5

7
9
.
6

6
8
.
9

7
2
.
1

V
e
r
b
s

2
5
.
0

1
4
.
6

2
0
.
0

1
6
.
7

5
6
.
9

4
4
.
6

3
5
.
4

3
5
.
6

6
4
.
2

6
0
.
0

4
6
.
0

5
6
.
0

A
d
v
e
r
b
s

3
1
.
9

1
8
.
7

7
.
5

1
3
.
1

6
1
.
3

4
0
.
6

3
4
.
4

2
0
.
6

6
1
.
3

7
0
.
0

4
2
.
5

4
0
.
6
.

P
r
o
n
o
u
n
s

4
1
.
9

2
8
.
1

1
8
.
7

2
3
.
7

7
1
.
9

6
5
.
6

4
5
.
6

5
5
.
6

6
8
.
7

7
1
.
9

5
6
.
3

7
0
.
6



B
a
l
t
i
m
o
r
e
 
S
l
u
m
 
C
h
i
l
d
r
e
n
 
1
9
6
5
-
6
6

F
i
r
s
t
-
g
r
a
d
e

T
h
i
r
d
-
g
r
a
d
e

F
i
f
t
h
-
g
r
a
d
e

W
h
i
t
e

N
e
g
r
o

W
h
i
t
e

N
e
g
r
o

W
h
i
t
e

N
e
g
r
o

M
e
d
.

I
Q

l
o
w

I
Q

M
e
d
.

I
Q

L
o
w
I
Q

M
e
d
.

I
Q

L
o
w
I
Q

M
e
d
.

I
Q

L
o
w
I
Q

M
e
d
.

I
Q

L
o
w
I
Q

M
e
d
.

I
Q

L
o
w
I
Q

N
o
u
n
s

5
7
.
4

6
2
.
8
:

5
7
.
2

5
1
.
7

7
7
.
2

7
0
.
4

7
2
.
5

6
3
.
6

8
1
.
5

7
6
.
5

8
2
.
4

7
8
.
3

,

A
d
j
e
c
t
i
v
e
s

4
5
.
9

3
3
.
9

3
8
.
9

2
4
.
3

6
9
.
1

6
4
.
2

6
3
.
6

5
5
.
0

8
3
.
9

7
8
.
6

6
1
.
0

7
9
.
8

V
e
r
b
s

2
5
.
9

2
3
.
3

2
0
.
2

2
2
.
3

3
8
.
1

3
4
.
7

3
8
.
6

3
1
.
3

5
8
.
1

5
0
.
9

5
8
.
1

5
0
.
5

A
d
v
e
r
b
s

2
6
.
9

1
5
.
0

1
5
.
9

1
2
.
2

4
0
.
6

3
9
.
7

3
5
.
0

3
0
.
3

5
9
.
7

5
1
.
9

6
0
.
3

4
9
.
4

P
r
o
n
o
u
n
s

4
5
.
6

3
2
.
2

3
7
.
5

3
0
.
9

6
9
0
7

6
4
.
1

7
0
.
0

6
0
.
9

7
9
.
1

7
2
.
2

7
0
.
0

8
0
.
3



a.Z.Ares:132weasa_a-

Results are very surprising in two respects: (1) Rural Maryland
children develop more slowly than suburban Maryland children even
'though the children are of the same age and the same tested intelli-
gence, and are closely matched in terms of schools attended, father's
educational level and general economic status; (2) First-grade white
children living in the slums of Baltimore City are accelerated com-
pared to first-grade white children living in the suburbs, and first-
grade Negro children are not behind white suburban children. It is
not so surprising, but it is informative, that rural Amish children
lag behind rural Maryland children. Slum children being behind
suburban children at third- and fifth- grades is not very surprising
either, because reports of verbal or intellectual deficit associated
with economic deprivation are legion (see especially reports by .
Deutsch and his co-workers (1964 to 1966) and by Coleman (1966). In
fact, study of slum children was undertaken after study of upper
middle class and working class children had shown minimal differences
(Entwisle, 1966a, 1966b). It was thought that more extreme economic
and cultural deprivation might be associated with language deficits
even though mild relative deprivation was not. Results directly
counter to expectation are both exciting and challenging.

School Differences. A few words are in order first about sclools
the children attended.

Many of the suburban children had attended kindergarten but
rural children had not. This could account, at least in part, for
rural-urban differences. It could not account for rural Maryland vs.
Amish differences, because neither group attended kindergarten.

A very large proportion of inner city children (80-90 per cent)
attend free kindergartens.3 We estimate that 75 per cent of the middle
class and 30 per cent of the working class children had attended kinder-
garten. Kindergarten attendance cannot explain the superiority of slum
children over suburban upper middle class children, then, because a
very large proportion of both groups attended kindergarten. Another
fact negating the importance of kindergarten is the equivalence at
first-grade of working class and upper middle class suburban children
even though many fewer working class children attend. Nevertheless,
at first-grade there is some correspondence between the per cent of
children in a given residential group attending kindergarten and level
of paradigmatic response rates because there are no reversals and some
consistencies between paradigmatic level and proportion of kindergarten
attendance. A favorable impact of kindergarten is reported by Lee
(1951) and Deutsch and Brown (1964).

A strong reason for starting research in Baltimore County was
that one large school system enrolls children residing in a 603 -
square -mile area, and this permits "school factors" to be held con-.
stant. (Such a large school district is rare and not typical of the
U.S.A.) Rural Maryland children, upper middle class suburban children,
and working class suburban children all attend schools under the same

9



jurisdiction. There is bound to be variation favoring schools nearer
the'city over rural areas,, but this variation is small compared to
variations between systems. Salary scales, curricula, textbooks,
testing programs, and many other conditions are the same throughout
one system. It seems unlikely, therefore, that school quality is a
major factor in rural-urban differences. The Amish, of course, at-
tend schools in Pennsylvania (some parochial) and so differences in
school quality are completely confounded with rural Maryland-Amish
differences. Big differences at first-grade, however, suggest that
more than just school factors are involved, because at first-grade
the school has had little time to exert an effect.

It is hard to make an overall comparison between schools in
Baltimore City and Baltimore County, but generally teacher qualifica-
tions in the City are equivalent to those in the County. The physical
condition of schools in the inner city is mostly inferior to schools
in the County (or to schools in other parts of the city) but this
relates more to lack of ancillary facilities like cafeteria, audito-
rium, sinks in classrooms, and so on, than to lack of educational
materials or the basic classroom essentials. The relative advance-
ment of suburban third- and fifth-graders compared to inner city chil-
dren could be owing, at least in part, to differences in school quality.
First-grade slum children being superior is just the opposite from
what could be predicted from school quality, but as was mentioned in
the case of the rural Maryland-Amish differences, the school's effect
has not had much time to be felt at first-grade.

One might questidn the assignment of children to IQ groups as a
possible cause of differences. The same IQ test (PMA) is used for
first-graders by the two school systems, but rural children and inner
city children may both be in "culturally unfair" positions vis-a-vis
standard tests. Thus these children's "true intelligence level"
might be higher than the test score indicates. For rural children
such a bias would lead one to predict an elevation in verbal develop-
ment rather than the lag actually observed, because to score at an
average level on an unfair verbal test implies that their endowment
is underestimated. For slum children it is the pattern of results
that vitiates the bias argument. First-graders of average IQ are
responding like high IQ (130) suburban children, and third-graders of
average IQ are responding like low IQ (80) suburban children. There
is no reason to believe that a strong bias at one age would be re-
versed in direction two years later.

Racial Differences. It is not fully clear what subcultural
factors could lead to the differences in paradigmatic rates between
white and Negro slum children. They attend similar schools and live
close to one another, except that the Negro slum dweller is probably
more deprived. (Number of children per family, mothers working and
fathers absent, illegitimacy rates, or almost any other measure of
social class turns out to be less favorable for Negro slum dwellers
(see Keller, 1963).) For census tracts where these particular slum
children's schools are located, percentage of unemployment, number of

10



occupants per room, age of dwelling, and number of women in the labor
force are all directly related to percentage of Negroes in the tract.
But there seem to be many more similarities than differences between
Negro and white slum dwellers, especially when they are being con-
sidered with respect to suburbanites.

The dialect cleavage between middle class interviewers and lower
class Negro children may be greater than that between these same
interviewers and lower class white children, and this might favor the
white children. The speech of Negro college girls who served as
interviewers resembles that of the white community more than that of
the Negro slum.

The most provocative difference that exists between the white
and Negro slum groups is in terms of in-migration.4 For persons who
resided outside the Baltimore area 5 years previously, most lived to
the south, irrespective of race membership. But proportionately many
more Negroes than whites are recent migrants. In the largely Negro
(88 per cent) tract, about 10 per cent had moved up from the south in
the previous 5 years compared to 2.to 4 per cent in the predominantly
white tracts. Persons coming from the south probably come mostly
from rural areas of Virginia and other southern states, Baltimore
being the first large industrial city going north on the eastern sea-
board. The relative advancement of the white over the Negro slum
children may thus reflect a rural-urban rather than a racial differen-
tial. This will be discussed more fully later.

Racial differences of another kind are related to interviewer
effects. There is considerable evidence that word associations are
sensitive to administration procedures (see Entwisle, 1966a, p. 35-45)
probably because of social pressure that inheres in the role relations
of the interviewer and respondent. For this reason the basic factorial
design incorporating sex, grade, and IQ as variables, was replicated
four times with slum children: white interviewers with white children,
Negro interviewers with' Negro children, white interviewers with Negro
children, and Negro interviewers with white children. Generally higher
rates of paradigmatic responding are noted when the race of the inter -..
viewer differs from that of the respondent, and low IQ students' are
most sensitive. (Labov (1966) illustrates at length the differences
in oral speech that people exhibit under varying conditions of for-
mality such as speaking to an interviewer, compared to scolding their
children.) Both white and Negro children tend to give more paradig-
matic responses (more mature behavior) when the race of the inter-
viewer differs from their own. This suggests that racial differences
could be at least partly owing to a performance-vs.-knowledge dis-
crepance.

Some preliminary evidence exists that semantic systems of white
and Negro children do not completely overlap (Entwisle, 1967b). For
instance the primary associate of Negro slum children to "music" is
"dance", while less than 4 per cent of white slum children and 1 per
cent of suburban white children give this response. Also "chicken"
is a high frequency response to "wing" for Negro children but not for

11



white children. Future work is planned to try to point up differences
in semantic and hierarchical. systems between Negroes and whites, and
aiso between' urban and rural dwellers.

Urbanization. Degree of urbanization, perhaps because of its
relation to opportunities for verbal interaction, seems to be generally
correlated with rate of development. Opportunities for verbal inter-
change may be limited for the rural pre-schooler because of isolation
of dwellings, lack of kindergartens, and lower exposure to television
and radio. Older rural children may be hampered by long bus rides to
and from school, and a large number of home and farm chores. Exposure
to mass media, especially television, may be a crucial factor in the
rural-urban difference because TV reception is less dependable in
rural areas and also the time available to watch TV is less for rural
children. TV became prevalent later in rural areas than in urban
areas, too, so its cumulative impact to date could be smaller. The
Amish, who develop even more slowly than the rural Maryland groups,
lack TV entirely and this is a tempting explanation for the Amish-
rural Maryland difference. (Genetic differences, also a possibility,
will be discussed later.)

There is little doubt that in terms of word associations
(Entwisle, et al., 1964; Palermo and Jenkins, 1963; Entwisle, 1966a,
1966b) or other measures (Templin, 1957) American children of fifty
years ago were not as advanced linguistically as modern children,
Modern children may be ahead as much as five years, if we take either
primary associates or degree of paradigmatic responding as an index.
This is a huge discrepancy.

The relative acceleration of modern children was one observation
that prompted study of the Amish children, for in many ways conditions
of life for Amish children resemble conditions for Minneapolis chil-
dren studied by Woodrow and Lowell.(1916). The AMishare the slowest
to develop of any American group we have observed (even Negro third-
grade slum children are faster), but they are considerably in advance
of Minneapolis children of 50 years ago. Unfortunately the early
data are based mainly on responses to nouns and adjectives, so many
kinds of syntagmatic responses that might help in estimating linguis-
tic maturity are not available. Nevertheless, thete are more immature
patterns (more sequence-type responses) in old data collected before
the days of mass media, and also there are more syntagmatic responSes
of the adjective-noun variety for modern groups living under "old
fashioned" conditions. It is very unfortunate that there are no data
available for American children around 1945 (just before television
was introduced) so that the effect of TV alone, aside from other
changes such as magazines, radios, improved curricula, and so on,
could be estimated.

Life conditions in the American urban slum may favor rapid
development of basic language skills. Houses are crowded, with many
children sharing one bed or bedroom, and time outside the house is
usually spent surrounded by other children on the street or playground.

12

rq



Although there is little evidence to cite, and certainly our data shed
no light on the point, the pressures upon young slum children to at-
tain some kinds of verbal proficiency may be more powerful than those
impinging upon the suburban child. The slum child usually needs to
find his own way around and it is not at all uncommon to see pre-
school children unattended on sidewalks, near busy streets. Anec-
dotes are often told of 4-5 year old slum children who are very
articulate when stopped on the street and asked for directions by
strangers. Lack of close supervision both inside and outside the
home could force the young slum child to develop skills in verbal com-
munication at an early age.

Several lines of evidence attest to the high exposure of modern
American children to television,, some estimates being as high as 5'
hours per day, roughly equal to the hours spent in school. In the
United States it is estimated that three-year old children already
average about 45 minutes of TV a day, and average daily watching time
increases to a peak at ages 12 to 13. The cumulative statistics are
staggering: during the 13 years of school attendance, the total time
spent watching television ranges from 6,000 and 12,000 hours (Schramm,
1964).

We guess that the sheer quantity of exposure (mostly via tele-
vision) may be the most inportant factor in accounting for the verbal
facility of 6-year-old slum children, with the narrowness of exposure
as a secondary but important cause.6 These two factors will now be
discussed in turn.

The slum children were selected to represent the most severe
degree of economic, and cultural deprivation in the city of Baltimore.
As many as 75 percent of children in some schools are on public assist-
ance and median income in some neighborhoods is as low as $2400 per
year. Notwithstanding, it would be very unusual for a house to have
no television set. (Keller's (1963) survey of a similar group in New
York City showed 100 per cent TV ownership.) Middle class children
also have television, but there is differential exposure (Bailyn,
1959) with lower class children having much higher rates than middle
class children. Pre-school slum children spend a large portion of
their time watching TV. Most'of the verbal models presented on TV
emanate from adult speakers. Even cartoon programs are cast in an
adult idiom, and situation comedies, also popular with children, deal
almost exclusively with adults. Commercials which are both strident
and frequent are also good sources of simple models, especially since
the are repeated verbatim over and over. (It is no accident that a
large number of children respond "living" to "pleasant" because this
is a part of a commercial linked to a local product.) The slum child,
although deprived in many ways, is thus afforded a very large amount
of vicarious exposure to simple discourse. There is little reason
to doubt that high TV exposure would enhance language development in
young children, because children who come from towns with TV have
larger vocabularies than children from non-TV'towns (Schramm et al.,
1961).

13



Secondly, the direct exposure to adult speech (parents, neighbors,
etc.) is in terms af simple verbal models. Some experimental work by
Hess and his co-workers, directly relevant to this point, describes
the constricted verbal environment of the slum child. Following a
hypothesis stated earlier by Bernstein (1962), Hess and Shipman (1965)
arranged for Negro mothers of four social levels (college-educated to
lower-lower class) tointeract verbally with their children. Inter-
action was recorded as the mothers taught their children sorting and
simple classification tasks. Later the mothers were also questioned
about how they would deal with certain hypothetical situations, like
preparing the child to enter school. Differences in language by
social class were marked, with middle class mothers using more speech,
and more elaborate speech expressing alternatives and contingencies.
The lower class mothers preferred simple, short sentences and gestures.
Use of abstract words was directly correlated with social class level.
Lower status mothers often conceptualized the child's behavior in
terms of role prescriptions rather than in.terms of individual needs

and situations. They were unlikely to allow sufficient time before
a decision to permit the child to assess contingencies or to analyze

a situation. A rather different cognitive panorama is seen by upper
status and low status mothers, and they relay these views directly

to their children. The language deprivation is essentially a depri-
vation of meaning.

This work suggests that verbal models presented to a slum child
in direct verbal interaction are apt to be simple and uncomplicated.
We have already remarked that through TV a slum child may also be
hearing mostly simple models. Thus, unlike the suburban child, the
slum child may be exposed almost entirely to straightforward and
redundant sets of utterances. Such exposure could favor early develop-

ment, and lead to an early appreciation of form class properties,
particularly for very common words and for verbal concepts at a law

level of abstraction. By the same reasoning, a lack of redundance
and possibly the negative transfer generated by the more complex
verbal environment of the suburban preschooler might be a temporary
handicap (seen in his short-lived inferiority to the slum child).

Hierarchical and abstract meanings, partially learned, would not be
very evident in word association measures. As the child continues
to develop and requires more elaborate verbal models to stimulate
growth, however, the suburban youngster begins to profit' from his
rich environment whereas the slum child falls progressively behind
and fails to redeem his early promise.

An uncensored and unlimited exposure to TV, plus a world popu-
lated with persons who speak a simple and uncomplicated idiom, may
be optimum for laying down form class concepts of the most basic and
elementary kind, but a handicap for ultimate development. The quan-

tity and the quality of the slum child's early verbal environment
may be superior to that of the suburban child for development of

language at its most basic stages. The disadvantages only become

apparent subsequently. The syntax that the slum child is not exposed
to may hinder communication later in a most insidious wty by depriv-

ing the child ,of the tools of thought. The plain talk that aids

14



early development may lead to a conceptual poverty that is more than
just a restriction in vocabulary.

Degree of urbanization, or a rural-urban differential, has been
recently reported (Greenfield, Reich, and Olver, 1966) to influence
cognitive growth on a much broader scale than we have been able to
observe directly in the development of form class concepts. Th
difference is one between abstractness and concreteness, and the
authors speculate that it is derived from a differential exposure
to problem solving and communication in situations not supported by
context. In Senegal and Mexico there are differences in abstracting
ability between rural and urban children, and although the difference
is'small, it is similar in nature to larger differences separating
children who have been to school from those who have not. The rural-
urban dimension appears to overshadow any kind of strict ethnic
grouping, because city-living Eskimos are much like urban children
studied elsewhere. The conceptual development of unschooled Wolof
children resembles that of lower class American children, particularly
as differences increase with age. (It is not hard to equate poor
schooling to lack of schooling, especially when absence rates are high
as in slum children.)

Because of this widely observed rural-urban difference in broad
areas of cognitive development plus our own observation of rural-urban
differences in children very closely matched on other variables
(Entwisle, 1966b), the relative in-migration rates of white and Negro
slum dwellers, mentioned earlier, gains favor as an explanation for
white-Negro differentials. Negro slum dwellers in Baltimore are much
more likely than white slum dwellers to have come recently from the
south, largely rural. Patterns observed in rural German children are
also consistent with this rationale.

The slum school's impact cannot be isolated, but its influence
is apparently not sufficient to maintain the advanced rate of develop-
merit found in first-graders. Many studies document the increasing
distance between disadvantaged and other children, and a huge survey
by Coleman (1966) suggests that school quality per se has very little
impart); on verbal skills. Whatever the slum home lacks, the slum school
may be unable to compensate for. Perhaps class-wise dialect cleavage
.prevents effective communication, because middle class teachers have
semantic systems that differ from those of their lower class pupils.
The curriculum perhaps is not designed either to capitalize upon the
advantages of a lower class environment or to compensate for its
deficiencies. The school environment could even be highly favorable
in terms of cognitive factors but cognitive factors may become irrel-
evant as affective factors become increasingly pressing. The slum
child's experience of failure and strong feelings of inablility to
influence the environment (see Coleman, 1966) could completely over-
shadow even optimum cognitive factors.

15



IV. DISCUSSION.

Discussion. Practically nothing is known about the transmission of

linguistic habits from one generation to the next, although the

wide differences in linguistic usage between social classes and sub-
. cultural groups must be initiated early in life. These differences

are far from being matters of simple dialect geography or pronuncia-

tion. They are matters of such far-reaching consequence as different

semantic systems and differing perceptual and cognitive styles.

Evidence is accumulating that non-standard English is a cognitive as

well as a social handicap. It hinders the child in problem solving

efforts as well as in reading, and the deficits seem to accumulate.

The research summarized here is concerned with acquisition of

fundamental concepts about words that will permit simple communica-

tion. Much further work is required to specify exact kinds of

language deficits and their precise relation to environmental

deprivation. For this purpose it seems most expedient to conceive

social class or subcultural membership as a series of specific

experiences to which a child is exposed.

Actually there is little research directed at social class or

subcultural differences in American children's language behavior

that cannot be explained by the well-documented association between

socioeconomic status and intelligence level. The rare samples where

IQ is controlled find class differences to be small but to increase

with age (Cherry-Peisach, 1965; Deutsch and Cherry-Peisach, 1966).

A specific cognitive deficit that might be class-related is a low

level of auditory discrimination. Some data of ours indicate that

children who live in very noisy environments may not develop the

requisite discrimination abilities to learn to read well. Recent

work stresses process rather than status variables as components

of social class, i.e. the educational aspirations of parents for

their children vs. income level (Bloom, 1964). Although income

level is correlated with school attainment, it is easy to see how

the two variables could be only indirectly related. Educational

aspirations are only realistic if income is sufficient to finance

continuing education.

There is no direct evidence available from our work relative to

genetic influences. That very young children are not retarded,
though, whereas older slum children are, suggests that it is chiefly

environmental factors which are responsible for social class

differences in language ability on our measures. Genetic influences

as far as rural children are concerned are not so easily disposed of.

Rural children may be drawn from a reservoir of persons with lower-

than-average verbal talents because of selective migration. That

is, over the years rural persons with the highest verbal abilities

may have been most successful at making a rural-to-urban move, and

those that remain have relatively low verbal ability. Although

our data are certainly incapable of testing this notion, one would

expect that inter-class mobility might be selectLa in the same

manner as rural-urban mobility. Persons capable of upward mobility

16

11.0401.7.,



by reason of superior verbal skills should rise from the slum
population or the blue collar population to higher class strata,
and thus lead to interclass differences in the urban and suburban
groups. Such differences are not found, of course, especially
at first-grade when such effects should be most apparent. The
genetic argument as an explanation for the rural-urban differen-.
tials thus loses some cogency although it cannot be completely
discounted.

Finding of strong rural-urban differences on word association
measures, and of minimal differences between urban groups, is
congruent with results from a whole series of studies on cognitive
style by Bruner and his co- workers cited earlier.= They find rural-
urban differences that resemble differences noted between schooled
and unschooled children, and inter-cultural differences appear small
when degree-of-urbanization is held constant. One reason advanced
for the rural-urban divergencies is the greater abstractness of
urban life which forces practice of verbal models not supported by
context. Our observation of a falling-behind by slum children
between first- and third-grades, and the evidence of impoverished
verbal models to which the slum child attends, point to a similar
cause for the slum child's rising verbal deficiencies.

Further study in more detail is now being undertaken to see how
the responses themselves differ from one group to another. (It is
known, for instance, that Amish children at third-grade not only give
fewer paradigmatic responses but give more syntagmatic responses.)
Labeling is generally considered a less complex mental activity than
categorizing, and an examination of superordinate responses, now under
way, may show that groups of the same level in paradigmatic develop-
ment nevertheless differ in conceptual structuring. Also the total
lexicon available for each group may give some clue as to the percep-
tual efficiency of the group as well as the extent of overlap between
semantic systems.

Finally, it should be noted that our findings are entirely
consistent with Chomsky's (1965) notion that exposure in no way
decrees the direction which language acquisition takes, whatever its
effect on rate. All the groups we have studied, including German-
speaking, show growth in paradigmatic responses to about the same
asymptotic levels, and the levels,of development from one group to
another are fairly stable even though groups differ in rate. If
Chomsky's notion is correct, slum children may just not receive
sufficient exposure to certain kinds of language ever to attain
an asymptote in other areas of verbal competence. It could be true
as well that there are stages in grammatical development where
exposure to instruction could be "most efficient" for children
whose extra-school exposure is insufficient, and this hypothesis is
amenable to experiment..

17



V. SUMMARY

Children's word associations, which are closely related

to grammatical competence and verbal development, seem to evolve

.through similar stages for several different American subcultural

groups. The groups do differ in rate of evolution, however, and

slum children are advanced at first grade and behind at third

grade. The intelligence of the child and the geographical loca-

tion of his home (urban vs. rural) are both influential factors.
School differences and differential exposure to mass media are

considered in relation to social class.

18



REFERENCES CITED

Bailyn, L., Mass media and children: A study of exposure habits
and cognitive effects. Psychol. Monogr., 1959, 73, Whole no.
471, 1-48.

Bernstein, B. Linguistic codes, hesitation phenomena, and
intelligence. Language and Speech, 1962, 5, 31-46.

Bloom, B.S. Stability and change in human characteristics. New
York: John Wiley, 1964.

Brown, R. and Berko, J. Word association and the acquisition of -
grammar. Child Developm., 1960, 31, 1-14.

Cherry-Peisach, E. Children's comprehension of teacher and peer
speech. Child Developm., 1965, 30, 467-480.

*Chomsky, N. Aspects of a theory of syntax. Cambridge: M.I.T.
Press, 1965.

Coleman, J.S. Equality of Educational 222ortunity. Washington,
D.C.: Office of Education, 1966.

Deutsch, M. and Brown, B. Social influences in Negro-white
intelligence differences. 3. soc. Issues, 1964, 20, 24-35.

Deutsch, M. and Cherry-Peisach, E. A study of language patterns.
The Instructor, 1966.

Deutsch, M., Malirer, A., Brown, B., and Cherry, E. Communica-
tion of information in the elementAra school classroom. 'Coopera-
tive Research Project No. 908. Office of Education, U.S. Dept.
of Health, Education and Welfare, 1964.

Di Vesta, F.J. A normative study of 220 concepts rated on the
semantic differential by children in grades 2 through 7. J. genet.
Psychol., 1966, 109, 205-229.

Entwisle, D.R. Word associations of young children. Baltimore:
The Johns Hopkins Press, 1966 (a).

Entwisle, D.R. Developmental sociolinguistics: A comparative
study in four subcultural settings. Sociometry, 1966, 29,
67-84 (b).

Entwisle, D.R. Form class of children's word associations. J.
verb. Learn. verb. Behay., 1967, in press (a).

Entwisle, D.R. Developmental sociolinguistics: Inner city
children. in preparation, 1967 (b).

19



REFERENCES CITED

Entwisle, D.R., Forsyth, D.F., and Muuss, R. The syntactic-paradig-

matic shift in children's word associations. J. verb. Learn. verb.

Behay., 1964, 3, 19-29.

Entwisle, D.R. and Muuss, R. Word associations of rural German

children. J. verb. Learn. verb. Behay., in press, 1967.

Ervin, S.M. Changes with age in the verbal determinants of word

association. Amer. J. Psychol., 1961, 74, 361-372.

Greenfield, P.M., Reich, L.C., and Olver, R.R. On culture and

equivalence: II. in Studies in cognitive growth.
Bruner, J.S.,

Olver, R.R., and Greenfield, P.M. (Eds.), New York: Wiley, 1966.

Hess, R.D. and Shipman, V. Early experience and socialization of

cognitive modes in children. Child Developm., 1965, 36, 869-886.

Keller, S. The sacial world of the urban slum child.
Amer. J.

Orthopsychiatry, 1963, 33, 823-831.

Labov, W. The Social Stratification of English in New York City.

Washington, D.C.: Center for Applied Linguistics, 1966.

Lee, E.S. Negro intelligence and selective migration:
A

Philadelphia test of the Klineberg hypothesis. Amer. soc. Rev.,

1951, 16, 227-233.

Meumann, E. Intelligenzprilfungen an kindern der volksschule.
Z.

exp. Padagogik, 1905, 1, 35-101.

Palermo, D.S. and Jenkins, J.J. Frequency of superordinate

responses to a word association test as a function of age.
J. verb.

Learn. verb. Behay., 1963, 1, 378-383.

Palermo, D.S. and Jenkins, J.J. Changes in the word associations
of

fourth and fifth grade children from 1916 to 1961. J. verb. Learn.

verb. Behay., 1965, 4, 180-187.

Riegel, K.F., Riegel, R.N., Smith, H.E., and Quarterman, C.J.

An analysis of differences in word meaning and semantic structure

between four educational levels. Mimeo. Dept. of Psych., University

of Michigan, 1964.

Rosenzweig, M.R. and Menahem, R. Age, sexe et niveau d'instruction

comme facteurs determinants dans les associations de mots.
L'Annee

Psychol., 1962, 62, 45-61.

Schramm, W., Lyle, J., and Parker, E.B. Television in the lives

of our children. Stanford, Calif,: Stanford University Press, 1961

20



REFERENCES CITED

Schramm, W. (Ed.) The effects of television on children and
adolescents. New York: UNESCO, 1964.

Templin, M.C. Certain language skills in children. Minneapolis:
University of Minnesota Press, 1957.

Thorndike, E.L. and Lorge, I. The teacher's word book of 30,000
words. New York: Bureau of Publications, Teachers College,
Columbia University, 1944.

Woodrow, H. and Lowell, F. Children's association frequency tests.
Psychol. Monogr., 1916, No. 22, 1-110.

C..

.:



1
See Acknowledgments,

2
See Acknowledgments,

FOOTNOTES

paragraph one.

paragraph two.

3
The exact percentage is unknown because many children attend

public kindergarten and then go to Catholic parochial first-grade.

The total first-grade enrollment exceeds the kindergarten enroll-

ment in spite of the departures to parochial school. There is

also a net out-migration from the school population which has

averaged over 1 per cent in recent years.

4
These data are taken from census tracts where inner city

schools are located. The tracts do not coincide exactly with the

boundaries of areas that schools draw their pupils from, but

neighboring City tracts are closely related. Census data are'for

1959, and our school survey was carried out in 1965-66.

5Some fragmentary data cited by Meumann (1905) suggests that

German children around the turn of the century may haVe similarly

given sequence-type responses at considerably later ages than modern

German children (see Entwisle and Muuss, 1967).

6We tried, unsuccessfully, to study slum children with hearing

deficiencies, with the idea that their exposure to language would

be greatly reduced. Hearing losses Spring from many causes, but

only those due to conduction defects would be suitable for study.

It is a commentary on the success of modern public health measures

that we could not locate enough children with hearing losses to form

samples, althoLSh twenty years ago this probably would have been an

easy task.

22