Language Development in 2-Year-Old Normal and Risk Infants

(1)

Language

Development

in 2-Year-Old

Normal

and

Risk

Infants

Judith Zarin-Ackerman, Ph.D., Michael Lewis, Ph.D., and John M. Driscoll, Jr., M.D.

From the New York Medical College; Educational Testing Service, Princeton, New Jersey; and columbia

Presbyterian Medical Center, New York

ABSTRACT. A variety of language measures was obtained

on two groups of 2-year-old infants matched for social class

but differing in terms of birth conditions. One group, a high

risk group, contained infants who suffered from RDS, birth

asphyxia, hypercalcemia, and hyperglycemia while another

group consisted of normal infants. The results of the

language tests revealed that the high risk group showed

poorer performance than the normal subjects. Other tests of

perceptual-cognitive development revealed little difference

between the groups. The data suggest that the assessment of

early trauma needs to employ a variety of measures,

espe-cially those which are related to the unfolding skills

appro-priate for the particular age group studied. Pediatrics,

59:982-986, 1977, LANGUAGE DEVELOPMENT, HIGH RISK, BIRTH

ASPHYXIA, RESPIRATORY DISTRESS SYNDROME (RDS).

Prenatal and perinatal insults experienced by

the fetus and newborn affect behavior in ways

that are not entirely known. Although

investiga-tors have noted the simultaneous occurrence of

specific birth traumata with certain types of

behavioral deficit in the newborn, causal

hypoth-eses that would predict the character and extent

of the deficit beyond the neonatal period are

rarely offered.

The first step toward proper assessment of the

effects of birth traumata on the growing infant is

the thorough measurement of the existing deficit.

Of consideration is the type and nature of the

measured deficit. Because language is a complex

cognitive function-one of the more significant

events in the development of the human child-it

may be particularly susceptible to CNS insult.

Language deficit by 24 months has been used as a

clinical indication of CNS dysfunction.’

More-over, Lennenberg2’ has reported language

defi-cits in children suffering from CNS insult.

To assess the consequences of early CNS

dysfunction, we were interested in studying

language skills in 24-month-old infants who

suffered from a variety of prenatal and perinatal

insults. Although many reports of behavioral

differences between risk and normal subjects

already exist in the literature, the present

investi-gation sought to improve the quality of this

research in two ways.

Since developmental reports of behavior delays

found in risk populations are frequently

con-founded with sociological factors and since a

significant percentage of perinatal trauma occurs

in the lower socioeconomic segment of the

popu-lation, it is difficult to pin causation of behavioral deviancy on the birth trauma itself or on environ-mental factors such as maternal nutrition,

prena-tal care, education, and/or family life. This is

particularly a problem when a behavior such as

language is being measured, since language may

be a skill that is encouraged and shaped by the

environment. Ryan4 notes that studies of language acquisition consistently fail to take into account

the language the child hears at home. To avoid

(Received March 19; revision accepted for publication June

9, 1976.)

Supported by NSF grant GB28105 to Dr. Lewis, grant

R225-72C to M. Richard Koenigsberger, and grant MCHS 239-07

to Dr. Driscoll.

ADDRESS FOR REPRINTS: (M.L.) Institute for Research in

Human Development, Educational Testing Service,

Prince-ton, New Jersey 08540.

at Viet Nam:AAP Sponsored on September 8, 2020 www.aappublications.org/news

(2)

TABLE I

PREPOSITIONAL ITEMS ON THE INFANT LABORATORY

LANGUAGE TEST

Procedure

METHOD Subjects

NEONATOLOGY SUPPLEMENT 983

these confounding effects, two groups of infants

were seen at 24 months of age who differed in

birth history but who were matched for social

class, birth order, and ethnicity. In this way,

differences found between the groups could be

attributed to the difference in birth experience alone.

The use of multiple language measures is the

other feature of the present study. The literature on language development has started to recognize

that young children’s language competence

de-velops at differential rates in various areas such as

lexicon growth, grammatical structures, etc.

Language might not be a unitary construction but

rather a conglomerate of skills slowly coalescing

into a general language competence. It seems

reasonable, therefore, to employ multiple

mea-sures when assessing language growth. This may

be particularly appropriate when assessing risk

populations where growth and development in

any one area may be decidedly different from

another. The present study used two tests of

language development that yielded five specific

language measures. These measures are obtained

easily and could be administered in the pediatri-cian’s office. As such they constitute a quick and relatively easy measure of the infant’s language ability.

Thirty-six risk infants who had suffered a multi-tiide of perinatal difficulties, including

respira-tory distress syndrome, birth asphyxia,

hypercal-cemia, and hyperglycemia, were given a battery of

language tests at 24 months of age as part of an

ongoing, high risk follow-up project at Babies

Hospital, Columbia-Presbyterian Medical

Cen-ter, New York. The risk group included

Cauca-sian, Black, and Hispanic infants from the

Washington Heights area of Manhattan and

Bergen County area of New Jersey. A control

group of 34 normal, term infants who were

matched to the risk group for sex, social class, and

birth order, was given the same language battery

at the Infant Laboratory, Educational Testing

Service, Princeton, New Jersey. These infants

came from the Princeton and Mercer County

areas and were from two ethnic

back-grounds-Caucasian and Black. The normal

sample was screened on testing for any known

difficulties experienced in the prenatal, perinatal, or postnatal periods or all three. Members of both groups were limited to the lower socioeconomic levels (ie, levels 3, 4, and 5 on the Hollingshead

Manipulated Acted out

1. The block goes into

the cup.

2. The dog goes in front 1. (Name of child), go in

of the chair. front of the chair.

3. The plate goes under 2. (Name of child), go

un-the block. der the table.

4. The block goes out of

the cup.

5. The dog goes behind 3. (Name of child), go

be-the chair. hind the chair.

6. The plate goes on t#{252}p 4. (Name of child), go on

of the block. top of the table.

scale6). Both the control and experimental groups

were evenly matched across social class.

The first task was a modified Peabody Picture

Vocabulary Test (PPVT). In addition to the

stan-dard comprehension task, the children were given

a production task. Using the same format for

each, the child was asked to label one of four

pictures on a page. The two different forms of the test were used for the two tasks, the production

task always preceding the comprehension task.

The number of correct responses before five

consecutive failures was the measure obtained for

both tasks.

The second task was the Infant Laboratory

Language Test (ILLT) which is comprised of two

parts-knowledge of prepositions and adjective

contrasts. Table I lists the prepositions that

comprise the test in order of their presentation.

They include “in front of/behind,” “into/out of,”

and “under/on top of.” Two different types of

responses were used to measure the child’s

know!-edge. The children were asked to manipulate

small objects presented to them on a table near

where they were sitting. All objects were clearly identified for the subject before the test item was

administered. The second response measure

con-sisted of having the subject “act out” the

instnic-tion given by the experimenter. This required the

child to use his body rather than objects to test his

knowledge. For example, it was said, “Johnny (the

subject), go on top of the table. “ Only the two

pairs of prepositions “in front of/behind” and

“under/on top of’ were used for the body

acting-out task. This body acting-out response was

included in order to determine whether

two-year-olds had prepositional knowledge when they were

(3)

TABLE II

ADJECTIVE ITEMS ON THE INFANT LABORATORY LANGUAGE

TEST

1. 2.

Show me which one is the dry cloth.

Show me which one is the small ball.

3. Show me which one is the clean sock.

4. Show me which one is the short stick.

5. Show me which one is the wet cloth.

6. 7.

8.

Show me which one is the big ball.

Show me which one is the long stick.

Show me which one is the dirty sock.

the subject of the manipulation rather than

another object.

The second part of the ILLT is presented in

Table II. This task involved knowledge of

adjec-tive contrasts: “dry/wet,” “small/big,” “clean/

dirty,” and “short/long.” The examiner placed

two objects that were representative of the two

contrasting adjectives in front of the subject. For

example, a clean sock and a dirty sock were

presented and the subject was asked to show the

examiner which was the dirty one. On another

trial (see Table II for order), the subject was asked

to show which was the clean sock.

Both parts of the ILLT-the two preposition

measures and the adjective contrasts-were

scored on the 4-point scale. A score was assigned to each pair of responses, not to each individual

response. A score of “3” indicated that both

responses of a pair were correct. A score of “2”

indicated that one response was correct whereas

the other response was incorrect but not a repeti-tion of the first. A score of “1” indicated that one

response was correct whereas the other response

was a repeat of the first. A score of “0” indicated that both responses were incorrect. For example, to the contrast “short/long,” the subject received

a “3” if he said “short/long,” a “2” if he said

“short/something else,” a “ 1” if he said “short/

short” and a “0” if both responses were wrong.

Although different E ‘s administered the

lan-guage tasks to the two groups of subjects, both E ‘s

were trained together and their reliability

ob-tamed on a group of ten subjects. The number of

agreements over all items of all tests as a ratio of

number of agreements plus number of

disagree-ments indicated percentages of consistency

ranging from .89 to .97 for the ten subjects.

RESULTS

Table III presents the mean data for all

lan-guage measures.

PPVT. Normal subjects showed significantly

better comprehension than risk subjects

(t = 2.49, P < .05, 2-tail). When production

scores were compared, normal subjects also

showed better production scores than risk

subjects (t = 2.40, P < .05, 2-tail). Thus, both

types of vocabulary measures revealed a superior

performance by the normal infants.

ILLT, prepositions. When the mean scores for

the manipulated prepositional items were

com-puted, no differences between the normal and risk

groups were found. Subjects scored an average of

approximately 2.65 out of a possible score of 9.00.

Similar results were found for the body acted-out

prepositional items. Both the normal and risk

groups scored an average of “ 1” out of a possible

“6,” with the normal subjects performing only

slightly better than the risk subjects. No

differ-ences between the males and females were shown

for either group on any of the prepositional

items.

ILLT, adjective contrasts. The mean data for

the adjective contrasts revealed that the normal

subjects scored twice as high as the risk subjects

(t = 2.29, P < .05). Out of a possible “ 12,” the

normal subjects scored a mean of “6’ ‘ whereas the

risk subjects scored a mean of “3.”

Language competence: A general “G.”

Pearson Product Moment correlations were

computed among all the language measures used

in the study to determine if the various language tasks were related. Table IV presents these corre-lations for the normal and risk subjects. Included

are the two Peabody scores and the three scores

from the ILLT. No significant differences

be-tween the normal and risk data were found. In

both groups, PPVT comprehension and PPVT

production were positively related to one another

(.58, P < .001). However, although comprehen-sion was positively related to every other measure

of language competence, production was not.

In-terestingly, the acting-out of prepositions was not related to the manipulation of objects in response

to prepositions. The correlation between these

two measures for both groups was low and

nonsignificant.

DISCUSSION

The language tasks presented in this study

appear to be only moderately related to one

another. Although comprehension and

produc-tion are related, it is only comprehension that is

related to adjective and prepositional skills. This

is not surprising since comprehension skills are

necessary to demonstrate adjective and

preposi-tional skills. Across all tasks, there is some

(4)

TABLE III

NEONATOLOGY SUPPLEMENT 985

MEANS AND STANDARD DEVIATIONS FOR THE PEABODY AND INFANT LABORATORY LANGUAGE TEST r-No. PPvT1 Comprehension Production .

X SD X SD

ILL1

Prepositions

#{176}

Adjectives

X SD

(Manipulate) (Act Out)

---SD SD Total Normal Risk 34 36 12.5#{176} 7.39 5.5 5.4 7.6#{176} 4.23 3.3 4.5

2.7 1.2 1.5 1.61

2.6 1.39 0.8 1.79

6.0#{176} 2.94 3.0 3.85 Male Normal Risk 17 18 12.2 6.93 6.0 7.61 6.7 3.55 3.3 5.56

2.5 1.83 1.5 1.77

3.0 6.89 1.3 2.75

6.2 2.85 3.3 3.76 Female Normal Risk 17 18 12.9 8.03 5.2 3.34 8.5 :3.54 3.3 3.78

2.9 1.58 1.7 1.50

2.6 2.03 0.5 1.27

6.1 3.13

2.8 8.70

OP < .05 t test.

tPPVT = Peabody Picture Vocabulary Test.

IJLLT = Infant Laboratory Language Test.

evidence that the risk infants displayed language

deficit when compared to normal infants.

Differ-ences between the two groups were found in their

comprehension and production as well as

prepo-sitional and adjective knowledge. These deficits

could not be a function of social class since this variable was controlled.

The etiology of this language dysfunction is still

questionable. It may be that the original trauma

continues to act on the developing organism.

Damage to the central nervous system may affect

such cognitive operations as information storage,

organization, and retrieval as well as the more

motorically dependent behaviors of speech

pro-duction and auditory speech coordination.

Alter-natively, the original insult may affect the parent-child relationship, causing the parents to behave in a manner that leads to the observed deficit. For

TABLE IV

PEARSON PRODUCT MOMENT CORRELATIONS BETWEEN LANGUAGE MEASURES#{176}

PPVT ILLT

Prepositions Adjectives Comprehension Production

(Manipulate) (Act out)

PPVT comprehen-sion PPVT production ILLTII prepositions (Manip) ILLT prepositions (Act out) ILLT adjectives . . . .581 -.11 .30 .30 .58t . . . .01 .681 .31 .46t .49t .24 .29

. . . .19

. 18 . . .

.28 .31 .44t .28 .49t .29 ..

#{176}Thenormal subjects’ correlations are to be found in the upper right portion of the table

whereas the risk subjects’ correlations are to be found in the lower left. N 34 for normal, 36

for risk infants.

tP < .01.

:1:P< .001.

§PPVT = Peabody Picture Vocabulary Test.

IILLT = Infant Laboratory Language Test.

(5)

example, the early insult may lead to a lack of

linguistic activity on the part of the infant which

in turn may cause the parents to curtail their

linguistic interaction with the infant. It may be

this parental behavior rather than the original

insult that produces the observed deficit.

These findings have a general implication in

that they suggest that “risk” has to be measured

around the unfolding skills of the infant. These

skills can be both within a domain, as in different

language skills, or across domains as in

percep-tual-cognitive or language skills. Moreover, it is reasonable to assume that behavioral skills vary in the level of demand they place on the developing

central nervous system and as such, we should be

alert to the possibility of differential deficit. By

employing multiple behavioral measures over a

period of time, continuing behavioral deficit can

be detected in the risk infants that could go

unnoticed with a less comprehensive test. Within

the set of language skills included in this study,

there are differences in the subjects’

per-formances so that not all tasks revealed

risk-normal differences. In an earlier report7 on data

for these same subjects, we observed their

percep-tual-cognitive skills at three age points (3 to 6

months, 9 to 12 months, 18 to 24 months). Their

attending behavior indicated no difference

between the risk and normal infants by 24

months. Thus, whereas their perceptual-cognitive

behavior revealed a “catch-up” phenomenon,

their language behavior, as reported in this paper,

revealed a continuing deficit. The total effect of

perinatal insult on cognitive behavior will be fully

assessed only when further testing in the areas of

thinking, memory, and language is continued over

a significant period of time.

REFERENCES

1. James S: Personal communication, 1973.

2. Lennenberg EH: The Biological Foundations of

Language. New York, Wiley, 1967.

3. Lennenberg EH: On explaining language. Science

164(3880): 635-643, 1969.

4. Ryan J: Early language development: Towards a

communication analysis. In, Richards MPM (ed):

The Integration of a Child into a Social World,

London, Cambridge University Press, 1974.

5. Nelson K: Structure and strategy in learning to talk.

Monogr Soc Res Child Dev 38(1-2, Serial No. 149),

1973.

6. Hollingshead AB: Two-factor index of social position.

Unpublished manuscript, 1957.

7. Zarin-Ackerman J, Lewis M, Driscoll J: Patterns of

visual fixation in the sick, premature infant: A

longitudinal study in the first two years of life.

Paper presented at the meeting of the Society for

Research in Child Development, Denver, April,

1975.

(6)

1977;59;982

Pediatrics

Judith Zarin-Ackerman, Michael Lewis and John M. Driscoll, Jr.