Journal of Abnormal ChiM Psychology, Vol. 17, No.

2, 1989

Behavioral Correlates of Developmental

Expressive Language Disorder
Marie B. Caulfield, 1 Janet E. Fischel, 4
Barbara D . D e B a r y s h e , 3 and Grover J. Whitehurst 1

The association o f behavior problems with preschool language disorders has

been documented extensively. However, researchers have typically failed to
differentiate subgroups o f language-impaired children, to use observational
data in documenting the behavior disorders, or to study children at the youn-
gest ages. Using a multimodal assessment, this study examined parent-chiM
interaction and behavior problems in a clearly defined subgroup of language-
impaired children, those with developmental expressive language disorder
(ELD). These children exhibit a delay in expressive language compared with
receptive language and nonverbal cognitive skills. Subjects were identified
and studied at the youngest age at which the disorder can be assessed. A group
o f ELD children, averaging 27 months o f age, was contrasted with a group
o f normally developing children, matched f o r age, sex, and receptive lan-
guage ability. Groups were compared on observed parent-child interactions
as well as maternal responses on the Parenting Stress Index, the
Eyberg Child Behavior Inventory, and a behavior-related structured inter-

Manuscript received in final form August 26, 1988.

Portions of these data were presented at the biennial meeting of the Society for Research
in Child Development, Baltimore, April 1987, and at the annual meeting of the Society
for Pediatric Research, Anaheim, California, April 1987. This work has been supported by
NIMH grant no. 1 R03 MH41603 to author Fischel, and by NICHD grant no. 1 R01 HD19245
to authors Whitehurst and Fischel. It has also been supported by grants of equipment from
Commodore Business Machines, Inc., Koala Corporation, and NEC Telephones. We thank
the Department of Pediatrics at the Nassau County Medical Center for the use of their
facilities.
1Department of Psychology, SUNY at Stony Brook, Stony Brook, New York 11794-2500.
2Department of Pediatrics, SUNY at Stony Brook, Stony Brook, New York, 11794-8111.
3Department of Child Development and Family, University of North Carolina at Greensboro,
Greensboro, North Carolina 27412.
~Address all correspondence to Janet E. Fischel, Department of Pediatrics, SUNY at Stony
Brook, Stony Brook, New York 11794-8111.
187
0091-0627/89/0400-0187506,00/09 I989 Plenum Publishing Corporation
188 Caulfield, Fischel, DeBaryshe, and Whitehurst

view. ELD children, when compared with normally developing children, ex-
hibited higher levels o f negative behavior and were perceived as different by
their parents.

The association between language problems and behavior disorders of child-

hood has been documented extensively (e.g., Baker & Cantwell, 1982; Beitch-
man, Nair, Clegg, Ferguson, & Patel, 1986; Cantwell & Baker, 1980; Jenkins,
Bax, & Hart, 1980; Richman & Stevenson, 1977). For example, in a group
of 705 randomly sampled 3-year-olds, Richman and Stevenson (1977) reported
a prevalence of behavior problems of 14~ as assessed by parental responses
on a behavior screening interview. Of those children in the sample who ex-
hibited language delay, 58O7ohad concurrent behavior problems. Similarly,
Jenkins et al. (1980) found an association between low language scores and
behavior problems in a sample of 418 preschool children. In this study, be-
havior problem ratings were obtained from a physician's evaluation of paren-
tal responses on a structured behavior interview.
In a particularly well-designed study, Beitchman et al. (1986) compared
the rates of behavior and emotional problems in a group of 142 children with
diagnosed speech and language disorders and in a normally developing con-
trol group. Frequency estimates of psychiatric disorders for the population
were based on a semistructured psychiatric interview with a subset of chil-
dren from each group. The estimated frequency of psychiatric disorder (i.e.,
at least one DSM-III Axis I disorder; American Psychiatric Association, 1980)
was 48.7~ for the speech- and language-impaired children, while that for
the control group was 12.0~ Axis I diagnoses were then grouped into three
categories: attention deficit disorder, emotional disturbance, and conduct
disorder. The estimated prevalence rates for children in the speech- and
language-impaired group were 30.4o70 for attention deficit disorder, 12.8~
for emotional disturbance, and 5.5~ for conduct disorder. The corresponding
rates for the control group were 4.5%, 1.5%, and 6.0%, respectively.
Clinical reports have also suggested that early language delay is often
associated with behavioral disturbance. Cantwell and Baker (1980) investigat-
ed the incidence of psychiatric disorder in 200 consecutive admissions to a
community speech and language clinic. Fifty-three percent of the clinic sam-
ple was found to have at least one psychiatric disorder as diagnosed by DSM-
III criteria. Thirty children (14~ of the sample) received the diagnosis of
attention deficit disorder with hyperactivity. Other common diagnoses were
avoidant disorder (80/o) and oppositional disorder (5070). Children in speech-
oriented special education classes have also been rated by their teachers as
having more problems than children in regular classrooms as measured by
the Behavior Problem Checklist (Lindholm & Touliatos, 1979).
While the correlation of language and behavior problems has been
documented extensively, research suffers from at least three important limi-
Behavioral Correlates of Expressive Language Disorder 189

tations. The first limitation is the lack of differentiation across significantly

diverse types of speech and language disorders. For example, data from chil-
dren exhibiting disorders of speech (e.g., articulation problems, stuttering)
and from those with disorders of language (problems in understanding and
using language) are frequently considered together (Beitchman et al., 1986;
Jenkins et al., 1980). Similarly, deficits in receptive skill (comprehension)
and expressive skill (speaking) are often not differentiated (Baker & Cant-
well, 1982; Beitchman et al., 1986; Lerner, Inui, Trupin, & Douglas, 1985;
Richman, Stevenson, & Graham, 1975). Also, speech and language disord-
ers with accompanying intellectual retardation are often undifferentiated from
those without accompanying retardation (Cantwell & Baker, 1980; Cantwell,
Baker, & Mattison, 1980). Finally, language-disordered children who also
exhibit autistic behaviors have been included in some studies (Baker & Cant-
well, 1982; Cantwell & Baker, 1980). It is likely that such studies mask sub-
stantial variation in the degree of association between language and behavior
problems for different subgroups of language-disordered populations.
Describing these variations will be crucial to understanding the role that lan-
guage plays with respect to behavior problems.
When studies have differentiated types of language disorder, interest-
ing results have emerged. For example, Baker and Cantwell (1982) catego-
rized children as exhibiting a pure speech disorder, a speech and language
disorder, or a pure language disorder. The prevalence of diagnosed psychiatric
disorder was much higher for the pure language-disordered group (95%) than
for the pure speech-disordered (29%), with the speech- and language-
disordered (45%) group falling in between. The distribution of developmen-
tal disorders, such as enuresis and mental retardation, followed the same pat-
tern. Given the variation in the relation of language and behavior problems
shown here for broad groupings of speech and language problems, there is
a need for research that identifies the behavioral characteristics of carefully
defined subgroups of language-disordered children. The nature of the lan-
guage-behavior relations can then be examined within those subgroups.
The second shortcoming in the research on language delay and behavior
problems is the primary reliance on parent and teacher report in assessing
inappropriate behavior (e.g., Jenkins et al., 1980; Lindholm & Touliatos,
1979; Richman & Stevenson, 1977). The benefits of multimodal assessment
of problems and, specifically, the importance of including behavioral obser-
vation in clinical assessment have been emphasized in the recent literature
on assessment (Barton & Ascione, 1984; Mash & Terdal, 1981; Nelson &
Hayes, 1979). Barton and Ascione (1984) suggested that behavioral obser-
vations may be less subject to inferences and subjective criteria than paren-
tal report. In addition, direct observations that include the parent may allow
the separation of child effects from parent effects (Barton & Ascione, 1984;
Nelson & Hayes, 1979). Child behavior, adult behavior, and adult percep-
tion are all combined in parent or teacher reports.
190 Caulfield, Fischel, DeBaryshe, and Whitehurst

The third limitation of existing research is that no investigation of the

association of language and behavior problems has included children under
the age of 3 years, despite the reliable occurrence of some language difficul-
ties in 2-year-olds. The lack of research with children who are at the begin-
ning of the developmental course for language and behavior problems severely
restricts our ability to examine causal links between language and behavior.
By studying younger children, it may be possible to find that problems in
one area precede those in the other. Among older children, the temporal clues
to causation are unlikely to be present.
In this study, we focused on one of the earliest diagnosable forms of
specific language disorder, developmental expressive language disorder (ELD;
American Psychiatric Association, 1987). This condition refers to a delay
in the onset of spoken language, with receptive language skill and nonver-
bally estimated IQ in the normal range. There is ample information for ac-
curate diagnosis of ELD in 2-year-olds (e.g., Fischel, Whitehurst, Caulfield,
& DeBaryshe, 1989; Resc0rla, 1984; Zelazo, Kearsley, Stiles, & Randolph,
1985), and estimates of prevalence are as high as 7 to 15% in 2-year-olds
(Rescorla, 1984). Clinical studies of ELD children have described them as
highly oppositional and resistant to parental control (Whitehurst, Fischel,
Caulfield, DeBaryshe, & Valdez-Menchaca, in press; Zelazo et al., 1985). There
are not, however, any systematic studies of the behavior problems of this
group of children.
We used a multimodal assessment to examine parent-child interaction
and behavior problems in ELD children. Behavior observations were used
to supplement maternal interview and questionnaire data in order to obtain
objective information regarding levels of inappropriate behavior in addition
to the mother's subjective ratings. The behavior observations were also used
to obtain data on the mothers' speech to their children and on the strategies
the mothers used to facilitate child compliance.
We compared 2-year-old ELD and normally developing children on the
following measures: (1) observationally derived data on parent-child inter-
action in a play and clean-up situation, (2) maternal ratings on two invento-
ries of child behavior problems, and (3) maternal responses to
behavior-related questions on a structured interview. In view of previous clin-
ical reports, we predicted that ELD children would exhibit more negative
and noncompliant behavior in the observational situation, and that parents
would rate ELD children as exhibiting more problematic behavior.
This study is a necessary step in the systematic study of the relation
of language and behavior problems. If the existence of a language-behavior
relation in children can be observed in very young children for whom lan-
guage delay is incipient, then we will have pinpointed a subgroup of chil-
dren for whom changes in these behaviors can be studied from the time of
earliest observation of that association. Since the rate of spontaneous remis-
Behavioral Correlates of Expressive Language Disorder 191

sion of expressive delay is high among 2-year-olds (Fischel et al., 1989)

future research could be focused profitably on the issue of whether language
problems and behavior problems wax and wane in concert. This is impor-
tant to determining whether their association reflects the operation of a third
variable, such as an underlying neurological dysfunction.
Alternatively, if the association of language and behavior problems is
not present in this group of children, we would have identified a subgroup
of language-delayed children for whom behavior problems are not related
to language delay, or we would have entered the process before the behavior
problems develop. Such findings would warrant longitudinal study of these
children to establish whether behavior problems ever arise and, if so, at what
point. A finding, for example, that the youngest ELD children do not ex-
hibit an elevated rate of behavior problems but do so later in development
would be consistent with a causal model in which language delay produces
behavior problems. Each of these potentially important lines of inquiry de-
pends on establishing the relation, if any, between ELD and behavior
problems in 2-year-olds.

METHOD

Subjects

Subjects were 34 normally developing children and 34 ELD children.

Children ranged in age from 24 to 32 months (mean age = 27.2 months).
Each group consisted of 28 boys and 6 girls. All 68 children were included
in the interview portion of the study; a subset of the children (16 boys and
4 girls from each group) were included in the observational and question-
naire portions.
All children were white, with the exception of two children of Hispan-
ic background. English was the only language spoken in the homes of these
children. Ninety-five percent of the families were from middle- or upper-
class homes, characteristic of the Long Island area. As shown in Table I,
the groups did not differ on demographic characteristics.
No children in either group had any diagnosed hearing deficit or neu-
rological impairment. However, there were significantly more cases of ear
infection in the ELD children's medical histories, as reported by their par-
ents (t(1, 69) = 2.109, p < .05). No child was being treated for an ear infec-
tion, nor did any parent suspect one existed at the time of assessment.
Selection criteria for ELD children included an expressive language stan-
dard score at least 2.5 standard deviations below the mean for chronological
age, as measured by the Expressive One-Word Picture Vocabulary Test
(EOWPVT; Gardner, 1981) and receptive language and nonverbal cognitive
192 Caulfield, Fischel, DeBaryshe, and Whitehurst

Table 1. Demographic Data on ELD and Control Subjects

ELD Normal
Subjects who were firstborn
N 13 19 p > .14 a
Birth weight in ounces
M 127.3 128.5 p > .78 b
SD 20.5 15.1
Number of siblings
M 1.06 .79 p > .22 b
SD .91 .87
Maternal education in years
M 14.2 15.1 p > .ll b
SD 2.3 2.2
Two-parent families
N 31 33 p > .30 a
Number of hours per week in
day care or with baby-sitter
M 9.1 10.3 p > .70 b
SD 12.4 13.9
Family income
Median $30,000 $30,000
"Chi-square test of independence.
bTwo-tailed t test.

ability scores no more than I standard deviation below the mean on the Pea-
body Picture Vocabulary Test-Revised (PPVT-R; Dunn & Dunn, 1983) and
the Leiter International Performance Scales (Leiter, 1976). These criteria are
congruent with the diagnostic criteria for developmental expressive language
disorder in the most recent edition of the Diagnostic and Statistical Manual
of Mental Disorders (DSM-III-R; American Psychiatric Association, 1987,
pp. 45-47). Children 24 to 30 months of age with such a great delay typical-
ly have only a few words in their vocabulary. Assessment of aspects of lan-
guage such as phonology, grammar, and pragmatics was not included in the
evaluation since it is not possible to evaluate these areas adequately in chil-
dren with little or no spoken language. Normally developing children had
expressive, receptive, and nonverbal IQ scores no more than 1 standard devi-
ation below the mean on these same tests, and were matched to the ELD
subjects for sex, age, and receptive language score (see Table II).

Table II. Group Comparisons of Age Equivalents Expressed in Months

ELD Normal
M SD M SD p~
Chronological age 27.54 2.80 27.09 2.50 > ,333
Receptive volabulary (PPVT-R) 29.23 3.70 29.40 3.63 > ,810
Expressive vocabulary (EOWPVT) 12.14 1.87 31.17 6.84 <,001
Nonverbal cognitive ability (Leiter) 31.71 8.82 31.46 9.74 > .630
aAll p values are for two-tailed t tests.
Behavioral Correlates of Expressive Language Disorder 193

Dependent Measures

Structured Parental interview. Each child's mother was administered

a structured interview concerning the child's medical, developmental, and
language history. A portion of this interview was devoted to questions con-
cerning the child's behavior. Six behavior situations addressed in these ques-
tions and the ratings used arc included in the presentation of results. With
the exception of the questions on feeding and bedtime, the ratings for these
questions were the mothers' responses. For the bedtime situation, a problem
rating was given if the mother indicated that getting the child to go to bed
was a problem, or if the child usually did not sleep through the night or did
not remain in his or her own bed through the night. For the feeding situa-
tion, a problem rating was given if the mother reported that the child did
not eat two or three meals a day plus snacks, or if the mother indicated that
the child's tastes in food were so selective that it created a problem.
Observational Data. Parent-child interaction was observed in a semi-
structured play and clean-up situation modified from the paradigm used by
Vaughn, Kopp, and Krakow (1984). The child was allowed 10 minutes of
free play, at which point the mother was instructed to have the child pick
up the toys and put them away in a set of plastic crates. If the mother re-
quested additional information regarding what to do, she was told to do what
she would do at home. The session ended when all toys had been put away,
or after 6 minutes had elapsed from the time at which the mother was in-
structed to have the child put away the toys, whichever occurred first.
The clean-up portion of the play session was videotaped and coded by
independent observers. While the observers were not told the subjects' lan-
guage status, differences in the samples of language on the videotapes may
have provided this information to the coders. The coders were blind to the
hypotheses of the study.
The presence (i.e., at least one occurrence) or absence of each of the
categories of mother and child behavior was coded within continuous
10-second intervals. The behavior codes, which are listed in Table Ili, in-
clude categories of maternal directive and responsive behavior as well as child
compliance and noncompliance. These categories are exclusive; i.e., no be-
havior is coded as two different categories. Interrater reliability across ses-
sions for each behavior category was calculated using the intraclass correlation
coefficient (Bartko, 1976). Correlations were not calculated for the category
"physical discipline" owing to its very low rate o f occurrence. Reliability es-
timates were obtained for a randomly selected 35~ of the sessions; the mean
intraclass correlation was .90 (range .68-.99).
Parenting Stress lndex. Mothers were administered the Parenting Stress
Index (PSI; Abidin, 1983). Responses on this index involve rating the degree
to which a number o f statements are true for oneself or for one's child. The
194 Caulfield, Fischel, DeBaryshe, and Whitehurst

Table IlL Behavior Categories for the Observational Situation

Child categories
On-task behavior Actively picking up or putting toys away
Refusal Explicit defiance of parental directive either verbally or
gesturally (e.g., shaking head side to side or saying "no")
Verbal communication Talking to self or parent, not including verbal refusal.
Nonverbal communication Vocalizing or gesturing that cannot be coded as refusal or
negative behavior
Toy contact Touching/playing with toys inconsistent with task
Other prop contact Touching objects other than toys
Negative behavior Includes tantruming (crying, screaming, flailing limbs),
aggression (forceful contact with parent), and inappropriate
toy use (forceful or descructive contact with toys)
Child other Child engaged in no behavior included in any of the other
child categories
Parent categories
Task directive Command/request to pick up or put away toys
Nontask directive Command/request not codable as task directive
Praise Positive verbal feedback regarding child behavior
Criticism Negative verbal feedback regarding child behavior
Conversation Child-directed speech not otherwise codable
Modeling/assistance Showing child what to do, helping
Physical discipline Forceful contact with child
Parent other Parent engaged in no behavior included in any of the other
parent categories

PSI has two domains, a child domain and a parent domain. The child do-
main includes the following subscales: Adaptability, Acceptability, Reinforces
Parent, Demandingness, Distractability/Hyperactivity, and Mood. Elevat-
ed scores in the child domain are associated with children whose characteris-
tics may be a cause of stress in the parent-child system. The parent domain
includes the following subscales: Depression, Restriction of Roles, Sense of
Competence, Attachment, Relationship with Spouse, Social Isolation, and
Parent Health. High scores in the parent domain suggest that some parent
attributes may be associated with high levels of family stress. The PSI has
been shown to discriminate between normal and clinical samples and is sen-
sitive to changes after clinical intervention (Abidin, 1983).
Eyberg Child Behavior Inventory. (CBI). Mothers also completed the
Eyberg Child Behavior Inventory (CBI; Eyberg, 1980). Responses involve
rating both the frequency and severity of a number of child behavior problems
(e.g., refuses to eat food prepared, physically fights with brothers and sis-
ters, cries easily, hits parents). The CBI has norms established for 2- and
3-year-old children (Robinson, Eyberg, & Ross, 1980) and has been shown
to be sensitive to changes due to clinical intervention (Eyberg & Robinson,
1982).
Behavioral Correlates of Expressive Language Disorder 195

Procedure

Subjects were recruited through announcements in local newspapers and

from pediatrician referrals. The normally developing children were seen for
one session. ELD children and their families were recruited as participants
in an ongoing longitudinal study of developmental expressive language dis-
order. The data presented in this paper were collected at the initial session
with each child. Diagnostic testing was conducted, and children who met selec-
tion criteria then participated in the observational portion of the study. The
structured interview was completed prior to the behavior observation ses-
sion. The mothers completed the PSI and the CBI during the free-play por-
tion of the observation session and while the child was engaged in language
and cognitive testing.

RESULTS

Structured Interview. Data for the interview questions were expressed

as frequency of response for each group; chi-square tests of independence
were conducted on these data. Since previous research supports an associa-
tion between language problems and behavior problems, we predicted that
ELD children would be more likely to be rated as problematic or difficult
in the behavior situaitons that were included in the interview. There were
significant group differences for two of the six behavior situations: ELD chil-
dren were more often rated as fearful or shy in new situations (X2 = 6.84,
p < .01). A greater number of ELD children also had problems related to
bedtime (x2 = 3.84, p < .05). Frequencies of response for the interview ques-
tions are shown in Table IV.
Behavioral Observation. Data were expressed as percentage of 10-second
intervals in which a behavior category was coded as present. Owing to the
large number of behavior categories, we conducted multivariate analyses of
variance (MANOVA) to minimize the probability of Type I error. MANOVAs
were conducted on the maternal and child behaviors separately. We
predicted that child behavior would be significantly different for the delayed
and control groups, with ELD children exhibiting negative or problematic
behaviors in higher frequencies. In light of the paucity of existing data on
the behavior of parents of language-delayed children, we made no predici-
tons regarding the maternal behavior categories.
The child behavior category "verbal communication" was excluded from
the child behavior MANOVA because group assignment was defined by
specific criteria of expressive language skill. Its inclusion in the analysis would
have artificially inflated the overall differences between groups. Instead, the
196 Caulfield, Fisehel, DeBaryshe, and Whitehurst

Table IV. Group Comparisons for the Behavioral Interview Dataa

Frequency Chi
Interview topics ELD Normal square
Discipline
Problem 13 11 .49 p > .492
No problem 19 23
Bedtime
Problem 12 5 3.84 p < .049
No problem 22 29
Feeding
Problem 10 5 2.14 p > .139
No problem 24 29
Shy or fearful in new situations
Problem 19 10 6.78 p < .010
No problem 12 24
Separating from parent
Problem 8 5 1.25 p > .262
No problem 23 29
Dependency on parent
Normal 17 20 3.44 p > .178
Overdependent 7 2
Overindependent 9 12
aOwing to missing data, sums are not equal for all analyses.

data for "verbal communication" were compared between groups using a one-
tailed t test. As expected, this comparison showed that the normally develop-
ing children used verbal communication significantly more than the E L D
children during the videotaped clean-up task (t(1, 38) = 9.80, p < .01).
Results confirmed the hypothesis that ELD children exhibit elevated
rates of problem behaviors compared with matched controls. A summary
of results is provided in Table V. The M A N O V A conducted on the child
behavior data was significant (F(7, 32) = 3.28, p < .01). Univariate tests
of individual child behavior categories showed several group differences. The
delayed group engaged in a higher rate of negative behavior (F(1, 38) = 4.09,
p < .025) than was found for the normally developing group. The negative
behaviors included crying, screaming, hitting, and throwing toys. The aver-
age number of intervals in which these behaviors were coded was 1.7 for
the delayed group (range 0-12) and 0.15 for the normally developing group
(range 0-2).
The delayed group engaged in a higher rate of nonverbal communica-
tion (F(1, 38) = 5.96, p < .01). Also, for ELD children, the behavioral
category "child other" occurred in more intervals (F(1, 38) = 5.60, p < .02).
This category was used when no other codable category of child behavior
was observed; it usually referenced quiet standing or sitting on the child's part.
The M A N O V A conducted on the maternal behavior data was not sig-
nificant (F(8, 31) = 1.69, p < . 15). For exploratory purposes, we conduct-
Behavioral Correlates of Expressive Language Disorder 197

Table V. Group Differences on Behavioral Observation Measures

Percentage o f intervals coded
ELD Control
Behavior categories M SD M SD F
Child behavior M A N O V A 3.278 c
On-task behavior .627 .288 .652 .245 .089
Nonverbal communication .408 .189 .265 .181 5.964 b
Refusal .040 .094 .035 .055 .040
Toy contact .444 .265 .417 .285 .103
Other prop contact .117 .206 .126 .177 .023
Negative behavior .048 .095 .004 .014 4.088 a
Child other .016 .027 .001 .006 5.603 b
Maternal behavior M A N O V A 1.693
Task directive .566 .216 .590 .265 .110
Nontask directive .108 .096 .130 .112 .449
Praise .181 .122 .219 .146 .827
Criticism .064 .050 .045 .064 1.089
Conversation .259 .208 .410 .181 6.014 b
Modeling/assistance .171 .205 .205 .180 .316
Physical discipline .010 .022 .000 .000 4.497 ~
Parent other .234 .236 .146 .160 .316
~p < .05.
~p < .025.
~p < .01.

ed univariate tests on the maternal behavior data. These tests showed that
mothers o f normal control children engaged in higher rates of non-task-
oriented conversation (F(1, 38) = 6.01, p < .02), and mothers of ELD chil-
dren were more likely to use physical discipline (F(1, 38) = 4.49, p < .04).
Questionnaire Data. Data were expressed as a summary of ratings for
the questions in each subscale. MANOVAs were conducted separately on
the scales in the PSI Child Domain, the PSI Parent Domain; and the CBI.
Given previous data from parental reports of child behavior problems, we
predicted that ELD children would receive higher stress ratings on the PSI
Child Domain, and that ELD children would receive higher frequency and
severity scores on the CBI. No predictions were made regarding the PSI Par-
ent Domain.
Results for the PSI supported the hypotheses. The MANOVA for the
Child Domain was significant (F(6, 33) = 2.63, p < .05). Univariate Ftests
showed that ELD children received significantly less favorable ratings on the
Acceptability subscale (F(1, 38) = 5.85, p < .025). This scale addresses paren-
tal expectations concerning the child's physical, intellectual, and emotional
characteristics. The multivariate analysis showed no group differences on
the CBI (F(2, 37) = .020, p < .99). There were no group differences for
the Parent Domain of the PSI (F(7, 32) = .49, p < .837).
198 Caulfield, Fischel, DeBaryshe, and Whitehurst

DISCUSSION

While a high frequency of behavior problems for children with language

disorders is firmly documented, the degree of association of language and
behavior for specific types of language disorders has rarely been examined.
The general finding that language-delayed children exhibit higher rates of
problem behavior has been supported here for a carefully defined group of
children with ELD. ELD children in this study exhibited more negative be-
havior in an observational situation and were rated by their mothers as more
shy and as exhibiting more problem behaviors at bedtime, when compared
with matched non-ELD controls. They were also rated by their mothers as
less acceptable in their physical, emotional, and intellectual characteristics.
These results indicate not only that behavioral differences are evident in chil-
dren who are delayed only in their expressive language but also that such
differences are observable in children as young as 2 years of age, an age group
not heretofore represented in research on language disorders and behavioral
disturbance.
Since we have shown the existence of behavior problems concurrent
with developmental expressive language disorder at 2 years of age, further
research following the course of both of these problem areas is warranted.
The rapid changes in language skill possible in this group of children (Fischel
et al., 1989) make this an ideal population for studying possible causal
models of the relation of language and behavior. It will be important to in-
vestigate whether the severity of language problems and behavior problems
coincide across time. One might also study the effects of intervention for
one of these problems on the other; i.e., does intervention targeting either
the language or behavior problems decrease or eliminate the other?
Note that differences in the occurrence of problem behavior between
ELD and normally developing children were more marked on the observa-
tional measures than on the parental report measures. Had we employed only
the self-report measures common to previous studies, the conclusions drawn
would be very different. While the questionnaire and interview data show
some differences between the ELD and normally developing children, these
data are enhanced markedly by the documentation of differences in observed
behavioral interaction, with higher rates of negative behaviors and nonver-
bal communication.
This raises the question of why parents did not rare their ELD children
as having elevated rates of problem behavior when such effects were manifest
on the observational measure. Clinical experience with parents of ELD chil-
dren suggests that parents may take their child's language delay into account
in their evaluation of the child and, as a result, not rate their child's behavior
as aberrant. Parents may attribute their child's negative behaviors to "frus-
Behavioral Correlates of Expressive Language Disorder 199

tration" at being unable to communicate and, consequently, expect less of

the child than they might of a nondelayed child of the same age.
Such sensitivity to the child's language delay is consistent with findings
regarding the parent-child language interactions of ELD children. White-
hurst, Fischel, Lonigan, et al. (1988) found that parents of ELD children
gauge their speech to their children on the basis of the children's level of ex-
pressive language rather than the children's receptive language level. This
suggests that parent behavior is greatly affected by the child's delay. If par-
ents' perceptions of their child's abilities are affected similarly, the lack of
problem behaviors as identified by parent report may reflect the lowered ex-
pectations of the parent rather than the actual behavior of the child.
An alternative explanation of why parents did not rate the ELD chil-
dren as problematic is that the higher rate of negative behavior found in the
observational data may be a mild precursor to a developing problem that,
if its course is unaltered, will be noticed by the parent at a later time. Given
the high rates of parent-reported behavior problems found in older children
with language problems, one would predict the development of clinically sig-
nificant problems in at least a subset of these children.
The usefulness of behavioral observation in discriminating ELD and
normal children indicates that this type of assessment may be particularly
important when dealing with early forms of behavior problems. In addition
to showing significant differences in negative behavior for the language-
delayed and normal children, the behavioral observation data have proven
useful for clinical prediction within the language-delayed sample. Fischel et
al. (1989) found that the percentage of intervals coded as "child other,"
indicating the amount of time the child was standing or sitting quietly, was
positively associated with expressive language ability 5 months after the ini-
tial assessment. This behavior category was included for the purposes of be-
ing exhaustive in coding and not for any predicted association with language
status or language change. Presumably, "child other" is inversely related to
high activity levels that may be problematic. While further observational work
is necessary to understand this finding more completely, these findings do
suggest is that subtle behavior differences that may go unnoticed by the par-
ents are associated with language progress.
The finding that mothers of ELD children were more likely to rate their
children as shy or fearful warrants some discussion. One might consider this
contradictory to the argument that these children are at risk for the kinds
of externalizing behavior problems to which we have referred. However, it
is important to consider what behaviors constitute shyness or fearfulness in
young children. We have found through clinical experience with these chil-
dren that they are often hesitant to use what little language they have. Such
reticence may be considered to reflect shyness or, alternatively, opposition-
200 Caulfield, Fischel, DeBaryshe, and Whitehurst

ality. Further investigation of mothers' criteria for ratings of shyness may

shed light on this.
We have found that 2-year-old E L D children display elevated rates of
problematic behavior. This finding is important in expanding our understand-
ing of the relation between behavior and language delay by demonstrating
that problematic behavior need not involve either intellectual or receptive
difficulties. I f language delay causes problem behavior, the present research
points directly to the child's level of expressive ability as the mediator. The
child's inability to express need verbally may lead to problem behavior as
an alternative communication system, as in the model proposed by Carr and
Durand (1985). The present research does not rule out the possibilities that
problem behaviors lead to language delay, or that both are a product of a
third variable. However, our finding of problematic behavior in children
whose delay is specifically and only expressive requires that these alternative
causal models be refined. For instance, why should problematic behavior
lead to expressive delay but not receptive delay? Perhaps most important,
by demonstrating that behavior problems can be measured in a very young
population of ELD children who have a highly variable developmental course,
we have provided a potentially powerful basis for future attempts at under-
standing the relations between language skills and problematic interpersonal
behavior.

REFERENCES

Abidin, R. (1983). Parental stress index (PSI): Clinical manual, form 6. Charlottesville, VA:
University of Virginia.
American Psychiatric Association. (1980). Diagnostic and statistical manual o f mental disord-
ers (3rd ed.). Washington, DC: Author.
American Psychiatric Association. (1987). Diagnostic and statistical manual o f mental disord-
ers (3rd ed.-rev.). Washington, DC: Author.
Baker, L., & Cantwell, D. P. (1982). Psychiatric disorders in children with different types of
communication disorders. Journal o f Communication Disorders, 15, 113-126.
Bartko, J. J. (1976). On various intraclass correlation reliabilitycoefficients.Psychological Bulle-
tin, 83, 762-765.
Barton, E. J., & Ascione, F. R. (1984). Direct observation. In T. H. Ollendick & M. Hersen
(Eds.), Child behavior assessment: Principles and Procedures. New York: Pergamon Press.
Beitchman, J. H., Nair, R., Clegg, M., Ferguson, B., & Patel, P. G. (1986). Prevalence of
psychiatric disorders in children with speechand language disorders. Journal o f the Ameri-
can Academy o f Child Psychiatry, 25, 528-535.
Cantwell, D. P., & Baker, L. (1980). Psychiatric and behavioral characteristics of children with
communication disorders. Journal o f Pediatric Psychology, 5, 161-178.
CantweU, D., Baker, L., & Mattison, R. (1980). Psychiatric disorders in children with speech
and language disorders. ChiM Psychiatry and Development, 12, 195-206.
Carr, E. G., & Durand, V. M. (1985). The social-communicativebasis of severebehaviorproblems
in children. In S. Reiss & R. R. Bootzin (Eds.), Theoretical issues in behavior therapy
(pp. 219-254). New York: Academic Press.
Behavioral Correlates of Expressive Language Disorder 201

Dunn, L. M., & Dunn, L. M. (1983). Peabody Picture Vocabulary Test-Revised (PPVT-R).
Circle Pines, MN: American Guidance Service.
Eyberg, S. (1980). Eyberg Child Behavior Inventory. Journal of Clinical Child Psychology, 9, 29.
Eyberg, S., & Robinson, E. (1982). Parent-child interaction training: Effects on family func-
tioning. Journal of Clinical ChiM Psychology, 11, 130-137.
Fischel, J. E., Whitehurst, G. J., Caulfield, M. B., & DeBaryshe, B. D. (1989). Language
growth in children with expressive language delay. Pediatrics.
Gardner, N. F. (1981). Expressive one-wordpicture vocabulary test. Novato, CA: Academic
Therapy Publications.
Jenkins, S., Bax, M., & Hart, H. (1980). Behaviour problems in pre-school children. Journal
of Child Psychology and Psychiatry, 21, 5-17.
Leiter, R. (1976). Supplement to general instructionsfor the Leiter International Performance
Scale. Chicago: Stoelting.
Lerner, J. A., Inui, T. S., Trupin, E. W., & Douglas, B. A. (1985). Preschool behavior can
predict future psychiatric disorders. Journal of the American Academy of Child Psy-
chiatry, 24, 42-48.
Lindholm, B. W., & Touliatos, J. (1979). Behavior problems of children in regular classes and
those diagnosed as requiring speech therapy. Perceptual and Motor Skills, 49, 459-463.
Mash, E. J., & Terdal, L. G. (1981). Behavioral assessment of childhood disturbances. In E.
J. Mash & L. G. Terdal (Eds.), Behavioral assessment of childhood disorders. New York:
Guilford Press.
Nelson, R. O., & Hayes, S. C. (1979). Some current dimensions of behavioral assessment. Be-
havioral Assessment, 1, 1-16.
Rescorla, L. (1984, April). Language delay in two-year-olds. Paper presented at the Interna-
tional Conference on Infant Studies, New York.
Richman, N., & Stevenson, J. (1977). Language delay in 3-year-olds. Acta Pediatrica Belgica,
30, 213-219.
Richman, N., Stevenson, J. E., & Graham, P. J. (1975). Prevalence of behaviour problems
in 3-year-old children: An epidemiological study. Journal of Child Psychology and Psy-
chiatry, 16, 277-287.
Robinson, E. A., Eyberg, S. M., & Ross, A. W. (1980). The standardization of an inventory
of child conduct problem behaviors. Journal of Clinical Child Psychology, 9, 22-29.
Vaughn, B., Kopp, C., & Krakow, J. (1984). The emergence and consolidation of self-control
from eighteen to thirty months of age: Normative trends and individual differences. Child
Development, 55, 990-1004.
Whitehurst, G. J., Fischel, J. E., Caulfield, M. B., DeBaryshe, B. D., & Valdez-Mencbaca,
M. C. (in press). Assessment and treatment of early expressive language delay. In P.
Zelazo & R. Barr (Eds.), Challenges to developmental paradigms: Implications for the-
ory, assessment and treatment. Hillsdale, N J: Erlbaum.
Whitehurst, G. J., Fischel, J. E., Lonigan, C., DeBaryshe, B. D., Valdez-Menchaca, M. C.,
& Caulfield, M. B. (1988). Verbal interaction in families of normal and expressive
language delayed children. Developmental Psychology, 24, 690-699.
Zelazo, P., Kearsley, R., Stiles, K., & Randolph, M. (1985, April). Parentalfacilitation of ex-
pressive language. Paper presented in G. J. Whitehurst (Chair), Early expressive lan-
guage delay: Identification, correlatesand treatment. Symposium conducted at the biennial
meeting of the Society for Research in Child Development, Toronto, Canada.