Pragmatic language impairment in relation to autism and SLI

Pragmatic language impairment:

A correlate of SLl, a distinct
subgroup, or part of the
autistic continuum?
Dorothy V.M. Bishop

Inilr

lOO

BISHOP

pr:sseRts: r /ith,.language,

difficultiqq rnay ,hav-e :trobl ms.,that:: ext@,beyond the
traditionatr'bound*riei'af'sl-tr, and resemble,those ieen,in: auti , Although
addi_
tionaLBrollemscan arise aq a :eonseqtre*e'e of oral,languagg,*iffi;lties,,we should
L'eware.sf,aut atie&a ing tharpiagmatic Aifnel:l{iei ,iecondar], problens;'Assessmentl'of nonverb,al.cotnrnmieation ean:prilvide,v*lua'ble jn{orna.tion
in'malcingdiagnostie,distinctic*s.: :,,

INTRODUCTION

ol specific language impairment (SLf
emphasise the difficulties that affected children

Accounts

have with mastering the medium of communication: Grammar, vocabulary and, frequently,
phonology are learned with difficulty, so the child
has to struggle to express what he or she wants to
say. Comprehension is often also impaired such

that the child may focus on a few content words
and deduce the meaning from these, but make
errors in understanding spoken language if this

requires complex vocabulary or syntax. It is
typically assumed that the affected child has a
normal desire to communicate and is developing
unremarkably in other respects. As Miller (1991)
put it:
"Children with language disorders evidence
strengths in conversation skills. They are
purposeful and responsive; however com-

munication is limited by their mastery
grammatical form." (p. 6)

ol

Howeveq not all cases of SLI fit this description.
Occasionally one sees children whose problems
are not confined to language form: content and
use ollanguage are also abnormai. pragmatic difficulties are communicative problems that lrave to
do with the appropriate use of language in a given
context. Typically what one sees is a child using
utterances that are syntactically well-formed and
complex, but which don't appear appropriate in
the conversational context in which they occur
(see Bishop, 1997, for a fuller account). In addition, such children may display social and behavioural deficits that resemble those seen in autistic

disorder. Strictly speaking, then, they don't seem
to fit the category of specific langLiage impairment.
Most experts will agree that such children e_ris:
(although I shall argue that their numbers ma1, be
underestimated), but they disagree as to hou. to
respond to them.
A common reaction is to regard tl.re pragmaric

difficulties and nonverbal impairments in social
or imaginative behaviours that are seen in SLI as
secondary to the oral language difficulties. For
instance, Brinton and Fujiki (1993) point our that:

"Because language skills piay a critical role in
social interaction, it seems likely that children with
language difficulties would be at particular risk
lor social failure" (p. i95). These authors argue
that the traditional diagnosis of SLI by exclusion.
with its emphasis on the specifici4; of the impairments in language structure, is misleading and
unrealistic. lmpairments in language do not occur
in a vacuum, but rather have impacts on man).
aspects of development. According to this viewpoint, associated impairments are to be expected.
A contrasting view points to the centrai place

that pragmatic deficits play in autistic disorder.
and proposes that children with serious prag_
matic impairments should be regarded as cases
of autism, or at least be categorised as having a
pervasive developmental disorder, rather than a
specific developmental disorder. Of course. both
views might be valid in different cases. The problem is that, unless we are able to specily just how
and when a language disorder can lead to secondary problems, we confront major diagnostic diffi-

culties every time we see a child who presents
intermediate between
autistic disorder and SLL Furthermore, as Craig
(1993) pointed out, until we establish whether
social interactionai difficulties in SLI represent a

with a clinical picture

primary deficrt in social knowledge or a secondary

6. PRAGMATIC LANGUAGE IMPAIRMENT

101

ofpoor language skills, our attempts

colleagues. Rice, Sell and Hadley (1991) assessed

will be inadequate.
In this chapter, first I review studies that look
at social and pragmatic impairments in children

social interactions in a preschool playgroup for
four groups of children: those with SLI, those
with speech disorders, those with English as a
second language (ESL), and a control group of
normally-developing children. They found close

consequence

at intervention

wrth SLI, considering how frequent these are, and
how far they can be regarded as secondary consequences of structural language limitations. I then
move on to cases of pragmatic language impairment that are more difficult to explain in this way,
because the child's use of language is disproportionately poor in relation to structural language
skills. For such children, a critical question is
whether a diagnosis ofautism or autistic spectrum
disorder would be appropriate. My conclusion is
that there are many children who fall between the
diagnostic options of SLI or autism: their developmental difficulties are not restricted to structural
aspects of language but, on the other hand, they
do not have the full range of pervasive impairr
ments that would warrant a diagnosis of autism.

SOCIAL AND PRAGMATIC DEFICITS AS
SECONDARY CONSEQUENCES OF SLI
Several studies of SLI have documented probiems
with peer relationships and social interaction. For
instance, Paul, Spangle Looney and Dahm (1991)
found that late-talking 2- and 3-year-olds were
impaired on the Socialization Scale of the Vineland Adaptive Behavior Schedule, even if items
involving langtrage were excluded from consideration. Nonverbal items included playing social
games, imitating complex motor routines in play,

using household objects

in play,

and smiling

appropriately. Around one third ofthe late talkers
had receptive language deficits at age 3 years, and

all but one of these children was also impaired
on the Socialization Scale. Such a study tells us
that problems exist, but does not help us sort out
cause and effect. Do the language problems lead
to poor peer relations, or does the child have a
more basic deficit in social cognition or emotional
relations?

Empirical evidence to show that a child's
language level affects peer relationships was
provided in a series of studies by Rice and her

similarities between the SLI and ESL groups: both
were less likely to initiate interactions, and when
they did so, they were more likely to communicate with an adult rather than another child. ln a
related study, Hadley and Rice (1991) found that
children with language or speech impairments
were half as likely to be addressed by their peers
as children with age-appropriate language. When
addressed by other children, they were less likely
to respond. It is sometimes assumed that children
are largely insensitive to the characteristics of
their playmates before the age of 7 or 8 years.
This study contradicts that view, and shows that
even by 3 or 4 years ofage, children are aware of
the communicative level of others, and prefer to

interact with peers who have age-appropriate
language skills. These authors concluded that
there is a negative interactive spiral, whereby a
child becomes increasingly unwilling to engage
in interactions with peers after experiencing lack

of

success.

Gertner, Rice and Hadley (1994) went on to
measure children's peer relations more directly,
using a sociometric method in which children
were asked to nominate which of their classmates
they would like to play with, and which they
would prefer not to play with. Children with ageappropriate language were by far the most popular group, and level ofreceptive language was a
strong predictor of children's popularity. These
authors note that children with English as a second language, like children with SLI, receive fewer
positive and more negative nominations from other
children. They argued, therefore, that the social

difficulties of the SLI group are a direct consequence of their limited language skills, rather
than reflecting some constitutional limitation ol
social cognition that co-occurs with the language
impairment. More recently, Redmond and Rice
(1998) showed that teachers rated children with
SLI as having more social problems than controls
whereas parents did not, and concluded that this

102

BISHOP

was further evidence that social impairments
were situationally dependent, and not an intrinsic
characteristic of the child. Viewed in this 1ight,
abnormal use of language by some children with
SLI could be a secondary consequence of lack
of social experience, which arises as a result of
unsuccessful attempts at communication in the
early years of life.

PRAGMATIC DIFFICULTIES THAT ARE HARD

TO EXPLAIN AS SECONDARY
CONSEQUENCES

There is, however, reason to believe that this is
not the whole story. I sha1l locus on three lines of
evidence that favour the view that, in at least some
children, there are pragmatic and social difficulties
that cannot be explained away as direct or indirect effects ol poor mastery of structural aspects

of

language.

1. Children with disproportionate pragmatic
difficulties: "semantic-pragmatic disorder"

difficulty for any theory that
attempts to explain away pragmatic problems in
SLI as consequences of oral language limitations

Perhaps the greatest

is the heterogeneity of SLI. Some children who
with SLI appear to have particular diffi-

: Rfrpin1s,1Sg6 charaeterisatian of ,scmantic ,, ' ,"'
-.
. -f::g=1*t"ry1| --..-...-

i

r

Verlrosily, , ,,;
Comprehenrion defieits for conneeted speocl
'frlord finding deficits
Atypical word choices
Phoaology and syntax unimpaired

i
I
;
: Inadequate eonversational,skills,
i Speaki*g'aloud,'to'ao ono in'particular , .

.,

:

,

Poor mainlenance of topic
i', Answering besides the peint of e questiou

:
,

and Allen coined the term "semantic pragmatic
deficit syndrome" to describe this subtype. Bishop
and Rosenbloom described a similar clinical picture and, following Rapin and Allen, termed this
"semantic-pragmatic disorder". The most recent
clinical account by Rapin (1996) lists the clinical
characteristics shown in Table 6.1. Although the
UK and US accounts have close similarities,
there is an important difference in how they use
diagnostic labels. Rapin's classification is not
designed simply to subtype SLI, but rather to
characterise communicative problems arising from
any cause. Thus the term "semantic pragmatic
deficit disorder" may be applied to children with

present

known organic etiologies, such as those with

culties with the pragmatic aspects of communication. Their deficits are similar in kind, though
typically milder in degree, than those described
in high-functioning autism. However, there is no
positive correlation between severity of expressive
or receptive language limitations and extent of
pragmatic difficulties. Problems in the appropriate
use of language can co-occur with relatively good
mastery of language form.
In the mid 1980s, two descriptive taxonomies
of developmental language disorders were published one (Rapin & Alien, 1983) in the USA,
and the other (Bishop & Rosenbloom, 1987) in
the UK. Both described a subtype of language
impairment in which structural aspects of language (phonology and grammar) were relatively
intact, but use of language was abnormal. Rapin

hydrocephalus or Williams syndrome, and to those
with autistic disorder, as well as to a minority of
those with SLl. In the UK, "semantic pragmatic
disorder" has been regarded as a subtype of SLl,
and typically incorporates the implicit notion that
nonverbal IQ is normal, there is no known organic
etiology, and the child does not meet diagnostic
criteria lor autism.
Conti-Ramsden, Crutchley and Botting (.1991)
assessed a random sample of all 7-year-o1d children attending language units (special classes for
children with SLI) in England, using both teacher
report and standardised tests. On cluster analysis,
they identified a subgroup of children (cluster 6)
whose language profi1e closely matched Rapin and
Allen's "semantic pragmatic deficit syndrome".

However, the difficulties

of these children only

6,

became evident when teacher ratings were taken
into account. If reliance were placed solely on
standardised test scores, children in this subgroup
looked relatively unimpaired, despite the fact that
the teachers regarded their communicative difficulties as severe. This study emphasises on the
one hand that pragmatic difficulties can be seen in
the context ofadequate structural language skiils,
and on the other that there is a lack of assessments

PRAGMATIC LANGUAGE

IMPAIRMENT

103

made a higher rate of both pragmatically inappropriate and inadequate responses compared to
age-matched controls. This study, then, represented

a direct attempt to distinguish pragmatic problems that could be secondary to structural language limitations from those that could not, and
concluded that for some language-impaired children, the latter type of response was relatively
common.

that pinpoint pragmatic difficulties.

2. Microanalysis of conversational glitches

3. lmpairments of nonverbal
communication

to understanding pragmatic
is to look in detail at the com-

Of particular interest is nonverbal communication. It is sometimes assumed that children with

municative behaviours that lead to a child being
regarded as a case of "semantic pragmatic disorder". This approach was adopted by Bishop and
Adams (1989), who undertook a classification
of children's conversational utterances that had
been judged "inappropriate". Although some cases
could be accounted for by difficulties in sentence
formulation, semantic selection, or comprehension
of the partner's talk, others were less easy to explain this way. For instance, some children would
provide far too little or far too much information
to the conversational partner, or would use stere-

SLI will compensate for their oral language
deficits by using nonverbal communication. This
is exactly what is seen in children with severe
hearing impairments. Even if not exposed to a
natural sign language, most hearing-impaired children will make healy use of gesture and facial
expression to communicate, and will, if given the
opporhrnity to interact with other hearing-impaired
children, develop a sign language de novo (GoldinMeadow & Mylander, 1998). Use of nonverbal
communication in children with SLI is much
less well documented though an early study by
Bartak, Rutter and Cox (1975) reported that just
over 40oh of a sample with receptive language
impairments did not use gesture. Several clinical
accounts have remarked on difficulties experienced
by children with SLI in the use of gesture and
faciai expression, and difficulties in interpreting
nonverbal cues emitted by others (see Goldman,

Another approach
impairments

otyped utterances, sometimes with abnormal
prosody. More recently, Bishop et al. (2000) have
extended this approach, focusing specifically on
the extent to which children's conversational
responses mesh with an adult's prior solicitation.
An attempt was made to subdivide responses into
those judged to be adequate (meeting expectations), inadequate (not meeting expectations, but
plausibly accounted for by limitations of language
expression or comprehension), and pragmatically
inappropriate (not meeting expectations, but not
readily explained by limitations of language expression or comprehension). After training, raters
achieved reasonable agreement in independently
coding children's responses in this way. and the
method received some validation from the finding
that younger normally-developing children made

a relatively high rate of inadequate responses,
but relatively few pragmatically inappropriate responses. Children with SLI showed wide individual
variation in their performance, but on average

1987, for a brief review).

In the recent

study

described above, Bishop et al. (2000) analysed
conversations between language-impaired children and unfamiliar adults, and compared these
with adult-child conversations with normally
developing children. The language-impaired children had been selected to include some thought

to

have disproportionate pragmatic problems,
and others with more typical SLI. The surprising
finding was that both these groups made less use
of nonverbal responses (e.g. head nodding) than
normally developing children. This could not be
regarded as communicative immaturity, because
younger language-matched control children made

104

BISHOP

Boxplot showing
distribution of nonverbal
responses (absolute
number per conversation)

for four groups of children:
Language matched

controls (LA con), age
matched controls (CA
con), typical SLI (SLl-T)
and pragmatic language

impaired (PLl). The top
and bottom o{ the shaded
bar show the 25th and
75th centiles respectively,
with the central line

o

,6 30
-o
c)

c
o?o
c-'

o

z

1i)

indicating the median,

The error bars show 1Oth
and 90th centiles, with the
range being represented
by circular points. See
Bishop et al. (2000)
lor further data.

the highest use of nonverbal responses

LA con

(see

Fig. 6.1).
Overall then, there are several lines of evidence to support the view that some children have
pragmatic difficuities that are disproportionate and
not readily explained as secondary consequences
of SLI. How, then, should we classify such chil-

dren? One view is that they belong more naturally within the autistic spectrum rather than with
SLI. This view needs to be given serious consideration, given that pragmatic difficulties are a
hallmark of autistic disorder.

CA

con

SLI-T

secondary consequences of poor communication
skills. However, this view was gradually abandoned as research evidence mounted showing that

children with autism had distinctive communisative, social and behavioural difficulties not seen
in other conditions that affected language development. As our understanding of autistic disorder
increased a consensus developed that autism and
developmental language disorder are qualitatively
distinct, a position which is reflected in contemporary diagnostic systems.
The Diagnostic and Statistical Manual (DSM)

of the American Psychiatric
case in point. In

PRAGMATIC PROBLEMS AS
SYMPTOMATIC OF AUTISM

The relationship between autism and developmental language disorder was a topic that was
much discussed in the 1960s and 1970s. At that
time, several experts had suggested that autism
was, fundamentally, a form of language disorder,
and that the abnormal nonverbal behaviours that
are part of the autistic syndrome might simply be

Association

is

a

DSM-IV (American Psychiatric

Association, 1994), autistic disorder is identified
as a pervasive developmental disorder (PDD),
which means that it has two salient characteristics. First, the disorder affects several areas of
functioning. In autistic disorder, there is a triad

of impairments in language, social relationships
and behavioural and imaginative repertoire. Sec-

ond the impairments in these domains go beyond simple immaturities, with children showing
behaviours that would be abnormal at any age.
In contrast, specific language impairment (SLI)

6,

PRAGMATIC LANGUAGE IMPAIRMENT

....no.1ing19*o.o'.uy.i+.*..=:-.:..:iil:l::i:.]:i::..o.5...i
: no phrase speech by 30 monttr.
t"i,.t-u.:..t
it'aiiiiial*r*J*.*ai iaiij.*ia*iei.:',.
,,r,dimiaalsL*d'or attno*roa],,.bgb.blg,..r:::::,::::::ii:tr,,:::::::::::,,,,.,,.','l
.1..t..,*
-"'..tt=:-'..t+it-:,,,.-,,::,
deaf
oiarticulation
.,..l.i:tl..i:i"':l1
ir,ril'.....1:r".:,i...
ever thoueht

delects

,
'

'

*Joruria

i."eq

srereotyped utterances

'

1i;...i':1i...1i.t;

leverj

metaphorical language lever)
inappropriate remarks
no spontaneous chat
,pontun*us char
fails to respond to questions
n.n., used eesrle

is a paradigmatic example of a specific developmental disorder. This is diagnosed when there
is impairment in a single domain of functioning
(in this case language), and the development in
this domain is delayed rather than qualitatively
abnormal.

The distinction between specific and pervasive developmental disorders was supported by a
seminal early study by Bartak et al. (1975). These
researchers explicitly compared children with
autistic disorder and those with receptive developmental language disorder on both quantitative and
qualitative measures of communication and behaviour. The initial goal was to identify children
with severe comprehension problems in the context of normal nonverbal ability. Boys aged 5-10
years, with nonverbal IQ of 70 or above, were
recruited from specialist centres for children with

84
t.'.:.'.:l.:'..,i',i,
53

t05

tLt.i65'
:',':::i:1:45'
65

i,;.i:ti.i':,tlt''

.t'tti..'..ii,'..1g$,,.

or
37
32
6J

9
0
0
22

autism and/or language disorders. Once suitable
children had been found, information from case
records was used to subdivide them into those
with autistic disorder and those with receptive
language disorder. A small proportion of children
did not fall clearly into either category and were
kept separate. Table 6.2 shows some of the findings
of the study. In terms of communication there
were some points of similarity between children
with autism and those with language disorder, but
also clear qualitative differences. Two features in
particular stand out: first, almost all the children
with autism were repofied as not using gesture,
whereas this was true for only half of the languageimpaired children. Second, in the children with
language impairment, the communication difficult-

ies could be described in terms of immaturity
and/or lack of communicative skills. In those with

106

BISHOP

autism, there were more qualitative oddities in
the use of language that are not normal at any
age. For instance, children would use words and
phrases with a private, individualised meaning
based on their own personal experiences, such
that the listener who had not shared the child's
experience would have difficulty making sense
of the utterance. Kanner (1946), who used the
term "metaphorical language" to describe such
instances, gave an example ol a child who referred to himself as "Blum" whenever his parents
questioned whether he was telling the truth. This
apparently irrelevant utterance assumed meaning
once the parents realised that the child had been

reading a large advertising sign which stated
"B1um tells the truth". In the study by Bartak
et al., cryptic utterances such as these, which

n

I
I

become comprehensible only when related to the
child's personal experiences, as weil as stereotlped

use of language, were much more common in
children with autism as compared to those with
language impairments. The results may be summarised by saying that whereas both groups had
major difficulties in mastering language form
(syntax and semantics), plus limitations in vocabulary comprehension and understanding of complex sentences, the autistic group had addrtional
impairments in the appropriate use of language,
i.e. pragmatics.
The study by Bartak et al. (1975) was important in establishing that the profile of impairments in
autism cannot simply be reduced to consequences
of poor language comprehension. Over the past
two decades, the conceptualism of autistic disorder has changed with greater emphasis being
placed on impairments of joint attention, social
interaction, and social cognition. These are all
areas where children with autism can be shown to
be deficient, even when compared with control
groups matched on mental age and/or language
level. Furthermore, it is now recognised that
pragmatic difficulties are a hallmark of autistic
communication, and that other language difficulties are a more variable correlate (Frith, 1989).
Bartak et al. (1977') carried out a series of discriminant function analyses on data from the same
sample. Discriminant function analysis is a statisticai method that assesses how accuratelv a set of

measures can be used to assign an individual to

a

specific group. First, the two groups (in this case
those with autism or SLI) are compared on a set
of variabies, and each measure is assigned a
weighting which reflects how well it discriminates
between the groups. The weighted measures are
then summed for each child; those scoring abrove
a cutoff are allocated to category A, and those
below the cutoff to category B. If there is little
overlap between the groups on the measured char-

acteristics, then there should be close agreement
between this categorical assignment and the original allocation ofcases to groups. That is exactly
what was found by Bartak et al. (1977). Very few
children had scores that were at all ambiguous.
Overall then, the study by Bartak et al. (1975)
was seen as supporting the distinction between
autistic disorder and SLI, with pragmatic difficulties being a characteristic difficulty of the former
group. but not lhe latter.
Does this mean, then, that all children with
autistic-like pragmatic dilficulties shouid be regarded as autistic? There are two lines of evidence
that caution against leaping to that conclusion.
First, the studies concerned with dilferential diagnosis of autism and SLI, while noting the different clinical pictures of these disorders, aiso reveal cases that are hard to categorise as one or
the other. Second, studies ofchildren who are iden-

tified as having semantic-pragmatic disorder reveal that only a subset of them appear to have
signilicant autistic features in nonlinguistic domains.

STUDIES CONTRASTING AUTISTIC
DISORDER AND SLI: EVIDENCE FOR
INTERMEDIATE CASES

Bartak et al. (1975) noted that while the data
showed a clear differentiation between autism and
developmental language impairment, five "mixed"
cases were important in demonstrating an area

of overlap containing children who show some
features of both conditions. On the discriminant
function analysis, about one fifth of the children
could not be unequivocally classified usually

6,

PRAGMATIC LANGUAGE

IMPAIRMENT

107

because data from the child's clinical history told
a different story than the contemporary information. Some children had moved from an autistic
profile to a language-impaired one, whereas others
changed in the opposite direction.
Cantwell, Baker, Rutter and Mawhood (1989)

disorder, they nevertheless had difficulties in
social behaviour that were similar to those of the
autistic children. They commented on the artificiality of absolute diagnostic distinctions within
a population of cognitively delayed language-

carried out a follow-up in middie childhood of
29 of these children. These investigators found
that the earlier, ciear differentiation between the
groups became somewhat blurred at the later
age. Although many aspects of communication

mediate cases has been emphasised further in
studies of parents and siblings of people with
autism, which find that although core autism is
rare in these relatives, milder difficulties, often
involving only one or two of the elements of the
autistic triad are common (Bolton et al., 1994).
This has led to the notion of a "lesser variant" of

and behaviour tended to improve in the languageimpaired group, there was a subset of children for
whom peer relations deteriorated notably. Cantwell
et al. commented: "the finding of marked diffi-

in friendships in some children whose
language was improving raises questions regard-

impaired young children. The existence of inter-

autism being employed in genetic studies.

culties

ing the traditional view that the socioemotional
problems are just secondary features that have
developed as a response to. . . having a language
handicap" (p.29). Furthermore, at follow-up, 28%
of the language-impaired group showed ritualistic
behaviour,

3

6% had stereotyped mannerisms,

2I

%o

used metaphorical language, and 31% produced
stereotyped utterances. All of these are behaviours that are not typically regarded as part ofthe
clinical picture of SLI but are seen in autism.

Mawhood (1995) studied the same sample in
adulthood and found that several cases who had
been regarded as having a receptive language
disorder continued to show evidence of social
impairment and restricted interests. Overall, then,

both the original study by Bartak et al., and
subsequent follow-ups of this sample confirmed,
that, even when the most detailed and careful diagnostic process is applied there are children who,
are difficult to categorise unambiguously as cases
of autistic disorder or SLI.
In recent years, the notion ofa sharp boundary
between autistic disorder and language impairment has come under increasing pressure. In a
study of preschool children with mental handicap
and very limited language skiils, Lord and Pickles

(1996) contrasted social behaviour, nonverbal
communication, and repetitive behaviours in
autistic vs nonautistic cases. They concluded that,
although none of the nonautistic group had ever
been thought to have a pervasive developmental

AUTISTIC FEATURES IN CHILDREN WITH
SEMANTIC- PRAGMATIC DISORDER

Although there has been much debate as to
whether "semantic-pragmatic disorder" is part
of the autistic continuum (e.g. Lister Brook &
Bowler, 1992; Boucher, 1998, and associated
commentaries on Boucher's paper), there has been

little hard evidence on which to base an argu-

A recent study using neuropsychological
tests and measures of social cognition showed
close overlap in test profile between a group with
"semantic pragmatic disorder" and children with
ment.

high-functioning autism, both of whom contrasted
with children with a more typical form of SLI
(Shields, Varley, Broks, & Simpson, 1996a, b).
However, other studies, including those reviewed
above, show that many children with pragmatic
difficulties fall short of meeting diagnostic criteria
for autistic disorder. So we are left with the question of how to categorise language-impaired
children who have pragmatic difficulties which
sometimes, but not always, occur in association .
with mild abnormalities of social interaction or
restricted interests.

Part of the problem confronting researchers
who wish to address this issue is the lack of
clear criteria for defining "semantic-pragmatic dis-

order", coupled with lack of suitable diagnostic
instruments for objectively documenting pragmatic

difficulties. With most forms of SLI it is possible

108

-

BTSHOP

to devise operatioital dejrnitions" sclecting

chil_

dren. for instance. whose stanc-iarclisecl scores on
language tests fall belorv some specifiec,l lcvei.
However, the clinical leatures shown in Table
6.1 are not easy to assess by availabie tests. As
noted above. Conti-Ramsden et al. (1997) tbuncl
that language tests did not reveal a distinctive
pattern of deficits in children .,l,ho r,r,ere thought
to have semantic-pragr-nattc disorcier: it was only
rvhen tcacher impression ',vas taken into account
that the dilfrculties of these childrcn could be
characterised.

One w.ay ol respr:nding to this situation is to
make teachcr ratin-qs more objective. This approach
rvas adopted by Bishop (1998) in developing rhe
children's comnrLrnication checklist (CCIC). This
study had tu'o related goals. On the one hancl- it

rr

aimed to develop a trore objective way of assess_
ing childrent pragmatic impairments using ratings
by teachers or other prolessionals r.r,ho kneu, the

child uell. The rationale for adopting this approach

was that many praglnatic deficits are rare in
occurrence. and/or dilficLrlt to elicit in a clinical
e

t
d
n

i
ti
,f
T

fi

I

f,

!
l

icl

&n

problems. using instruments clevelopecl

lor

the

diagnosis of autism. For the present, though, our
data are consistent u'ith those of other str,rdies
in suggesting that there is n'ic1e variation in the
specific clinical profiles shown by children rvith
pragmatic difficulties, and it would be pr.emature
to conclude that all of them have a mild fonl ol
ar-rtistic disorder.

iarly might. therefor-e" be in a better position'to
evaluate these aspects of behaviour than a professional rvho sees the child fbr a sin-eie assessment.
A second goal ol this study ,as to consider hou,
strongly pragmatic impairments uere associated

u"ith impairments in other dornains, especially
problems rvith social interaction and restricted
interests ol the kind that are character.istic of

autism. In this study, a subset of chilclren liorn
the suru'ey by Conti-Ramsden et al. (1997) were
assessed on a checklist that included scales

skills, social interaction. and restricteci interests.
Ttems rvere retained

in the checklist ilthey shou,ed

acceptable ler,'els ol agreement betu'een independ_
ent raters. Diagnostic inlonnation ftom school
records was contpared u'ith checklist results. Chil_
dren nhose records indicateci a definite or possible diagnosis of "semantic pragmatic disorcler.'
did as expectecl obtain lower scores on the pragmatic scales. None of these children hacl a dia_
gnosis olautism, but a subset rvas thought to have

possible or definite autistic features or Asperger

#
,il

another subset ol the semantic pragmatic group
had ner.er been gir,en any autistic spectr-um dia_
gnosis. Thesc cliildren did not drfler fi.om a .,typical
SLi" group in terms of their scores on the social
and intelests scales. This study, then, supportecl
the vieri' that, .hile pragmatic in.rpairments and
other autistic features tencl to co-occllr. there are
children with si-enificant pragmatic difficLrlties
who do not have any marked difficulties in thc
domains olpeer relations or interests. The check_
list gives only a cursory exanrination of these
arcas, and in fllture u'ork it is planned to do a
much nore detailcd ar.ralysis of autistic f-eatr-rres
in language-inrpaired children with pragmatic

ent.

assessing aspects ol strr_rctural language, pragmattc

/

j

assessntent because they are contextually depencl_
A person u.ho interacts rvith the child reslr_

syndronte. The latter group shorved somc impaii._
n'rellts on the social ancl interests scales. Horvever.

IS THERE A SEPARATE SYNDROME OF
SEMANTIC- PRAGMATIC DISORDER?

I have argued that one sees children u,ith prag_
matic difllculties that arc not ji-rst secondary to
structural language problems, yet who do not
appear to meet criteria lor autism either. So where
should tliey be classified'J One solution rvould be
to propose "semantic pragmatic disorder" as a
separate diagnostic entity, distinguished from both

SLI and autistic disorder. I shall argue that this is
not a satislactory solution, because there is little
evidence that the features described in Table 6.1
fornr a colrerent syndrome. Rather. it seems as
though pragmatic dilficulties are a variable cor_
relate ol SLI: they can be flor-urd in children
who meet the clinical descriptions of semantic_

pragmatic disolder and rvho use fluent. complex
ianguage, but they can also be found in other
childreri u,ho have more typical structural language
problems. They may be associated with semantic

6.

PRAGMATIC LANGUAGE

IMPAIRMENT

1Og

language use

e^€

Notional distribution of
impairments in language

phonologicg

* E**
synactrc € e*-^*
deficit €e€ *€
€*e€ e*

structure and language

+

€€

phonologic-syntactic def icit

€€

*
=s

lems or unusual vocabulary.
What is the evidence to support these assertions? One source of data comes from studies
I conducted in the 1980s, using experimental
methods to assess aspects of pragmatic functioning in language-impaired children. The goal

to contrast children who had the clinical

features of semantic pragmatic disorder with other

children with more typical SLI. To this end, I
developed various tasks that were designed to
pinpoint cognitive processes thought to underlie
pragmatic problems, anticipating that these would
clearly differentiate typical SLI from ',semanticpragmatic disorder". However, assessments that
were devised to be sensitive to deficits in cases
of semantic-pragmatic disorder typically revealed
unexpected deficits in some children with more
typical SLL This proved to be the case with
tests of referential communication skill (Bishop
& Adams, 1991) and story comprehension (Bishop
& Adams, 1992). In the study already described

by Bishop et al. (2000), which used

and those with semanticpragmatic deficit. Good
skills are shown as +,
and deficits as -.

semantic* Pragmatic

*edeflcit

difficulties, but can also be found in children who
do not appear to have either word-finding prob-

was

use, according to a model
that allows for opposite
profiles in those with

language structure

derailed

analysis of more naturalistic conversational data,
the same picture emerged. There was a subset of
children who were prone to make pragmatically
inappropriate responses, and who were also

abnormally restricted in their use of nonverbal
responding. However, these children are not necessarily the ones identified by teachers as having
characteristics of semantic-pragmatic disorder.

Some of them had structural language limitations
as well.

Clinical classifications have stressed the contrast between semantic-pragmatic disorder and
more typical SLI, which typically corresponds
to what Rapin (1996) would term phonologicsyntactic deficit disorder. On this basis, one
anticipates a double dissociation between the two
kinds of impairment, as shown in Fig. 6.2. However, in a sfudy using an earlier version of a teacher

checklist, Bishop (2000) found results more
consistent with the scenario shown in Fig. 6.3.
Rather than the predicted inverse relationship
between impairments

of

language

form

and

pragmatic difficulties shown in Fig. 6.2,there was
a small positive correlation between the two
domains. Although there were many children
who did resemble the profile in Table 6.1, with
relatively good mastery of syntax and phonology
and poor pragmatics, and others who showed the
opposite picture, there were yet other cases who
had a "double deficit", with poor ratings on language form and language use. Taken together,

such results begin

to

question the notion of

semantic-pragmatic .disorder as a separate syndrome of language impairment. Pragmatic diffi-

culties are a common, though by no means
universal, correlate of language impairment. A
further result from this study was that there was
no positive correlation between semantic and prag-

matic difficulties, semantic impairments being
common in all language-impaired children.

110

BtsHoP

€-s*

€-

swntactic s-€

deficit
Alternative model
of distribution of

€
€

l+

€-

phonologic-@*

€

€
e€
€

e€
6€

impairments, showing

€

additional subgroup

with impairment in both

€

language structure and
language use.

language structure

e
€
€

€
e

*€

€ e€€
€

+

semanticpragmatic
deficit

social use
language
structure

Model depicting
dissociable impairments

in language structure,
social use of language,
and interests, and how
particular profiles map

onto existing clinical
categories.

Bishop (1998, 2000) suggested that the notion
of a distinct syndrome of semantic pragmatic disorder may have come about just because pragmatic

problems are much more obvious and hard to
explain away when they occur in the context of
good formal language skills. Quite simply, children who speak clearly and in complex sentences
but who use strange language draw attention to
themselves, because on the one hand we can
understand what they are saying, and on the other

hand we don't expect odd utterances from a
child with apparently mature language skills. Data
from both conversational analysis and from teacher
checklists, however, suggest that there is a wide
range of pragmatic competence in children with
more limited mastery of language form. Howeveq

we automatically tend to make allowance for
pragmatic difficulties in such a child assuming
that this is a secondary consequence of limited
powers of expression and comprehension.
Findings such as these do raise questions
about a term such as "semantic pragmatic deficit

syndrome". The label "pragmatic language impairment" seems preferable. It does not imply that
semantic and pragmatic problems

will necessarily

co-occur, and it is easier to accommodate to a
more dimensional view of language impairment,
which treats pragmatics as one domain in which
communication may be impaired.
The set diagram shown in Fig. 6.4 illustrates
this more dimensional view of communication disorders. The three sets have some correspondence

6.

t

with the traditional triad of autistic impairments,
but are shown as frequently being dissociated.
Different children wiil have different combinations
of symptoms, varying in severity. Autistic disorder
is diagnosed when a child has major impairments
in all three domains. Asperger's disorder, defined
in DSM-IV as a pervasive developmental disorder in which structural language skills develop
normally, is shown as involving a combined deficit
in the social use of language and restricted interests. Children who have pragmatic problems in the
context of relatively good language form would
in this diagram, fit in the area in the top nght of
the figure, where only the "social/pragmatic', circle
is involved. Howeveq as the diagram emphasises,
there are no clear boundaries between this disorder and autistic disorder, on the one hand and
SLI on the other. Furthermore, there are children
who have a mixed picture ol problems with language structure and pragmatics.

PRAGMATIC DIFFICULTIES AND PDDNOS

A dimensional view may be useful in capturing
the fulI range of clinical phenomena, but it can
create problems in practice. Quite simply, diagnostic labels are important in ensuring that
children obtain access to suitable education and
intervention. A11 too often, the ciinical reality is
that many children have complex constellations
of developmental difficulties, but provision tends
to be directed towards the textbook categories.
Thus one may be confronted with a stark choice
between an educational placement for children
with autism, which is geared mainly to the needs

of

low-functioning children

with

difficult-to-

manage behaviour, or one for children with SLI,

where the emphasis is on conventional speech
language therapy.

is increasingly being
such complex cases is pervasive

One diagnosis thaf
adopted

in

Developmental Disorder Not Otherwise Specifie<J,
or PDDNOS (see, e.g. O'Hare, euew, & Aitkin,
1998). This category was introduced in DSM-IV

to be used for cases where there is subthreshold
symptomatology, or where there is significant

PRAGMATIC LANGUAGE

IMPAIRMENT

111

impairment in only one or two of the domains of
the autistic triad. However, there are problems with

this label. PDDNOS was originally intended as
a default diagnosis to be used in rare instances
when a child just fell short of diagnostic criteria
for autism. It is vaguely described with no clear
defining criteria, and potentialiy incorporates a
ofclinical profiles. It is too nonspecific
a diagnosis to be useful for identilying the kinds
of services a child requires (though the label
itself may serve a function in ensuring that the
huge range

child obtains access to some kind of services).
It may be that, as PDDNOS becomes an increasingly popular diagnosis, better services will be
developed for high-functioning children who do
not have major behaviour problems but who do
have complex difficulties. Some speech-language
therapists have developed considerable expertise
in working with such cases (e.g. Brinton & Fujiki,
1989; Gallagher, 1991), but many professionals
stiil are seriously challenged by their complex
deficits. Although there is some work concerned
with facilitating nonverbal communication in children (e.g. Nowicki & Duke, 1992), this has been
developed independently by clinical psychologists
and there has been little application to languageimpaired children.
My own view is that it would be premature to
conclude that children with pLI either do or do
not properly belong in the autistic spectrum in
our eurrent state of knowledge. Future research
on etiology or underlying cognitive processes may
give clearer answers. Meanwhile, an important
message from the research to date is that we need
to look carefully at pragmatics when assessing a
child with language impairment, and not automatically assume that poor use of communication
is a secondary symptom. To date we have been
seriously limited by the lack of reliable and practical assessment tools. Methods such as the checklist described by Bishop (1998) may provide a
start in alerting practitioners to cases where the
child has a wider range of communicative difficulties. In addition, video analysis of nonverbal
communication looks like a promising method for
identifying the child whose communication difficulties extend beyond oral language. I suspect that
ultimately we may need to abandon the sharp

112

BISHOP

distinction betr,r'een autism and SLI. and mor,e to
a more quantitative approach, u.hich depicts PLI
as literally intermediate betrveen autistic disorder
anci SLl, rather than belon_eing with one or the
otl.rer. One challenge fbr the future is to devise
more appropriate pror.ision for children with such
complex problems.

REFEREhIOES

Bishop. D.VM.. Chan, J.. Hartley.. J.. Adams. C.. &
'eir" F-. (2000). t'onr,ersational responsir,eness in
spccific langr-rage irnpairment: Eviclcnce of disproportionate pl"agmatic dtfficulties in a sr.rbset of
clrldren. Developtneri drul Ps.t'rhopathology, 12,

177 199.
Bishop. D.llvl., & Rosenbloorn. L. (i987). Classification of childhood language disorders. hr &'. Yule &
N{. Rutter (Ed,s')" Language clet,elopmenr und disorders; Clinics in Developntentul Medir:ine fdoubie
issue]. London: MacKeith Press.
Bolton, P, MacDonald- H.. Pickles, A., Rios. P. Goode,
S., Cror','son, M., Bailey. A., & Rutter, M. (1994).

American Psychiatric Association. (.1994). Diagrutstic:
cold .tatistical ntutuul of' mentctl clistnrlers, llh
eclition. (DSM IIt1. Washington. DC: Amelican

A case-control family history study of auttsm.
Jotu'nul oi Child Psycholagt, uncl Psychiutr.y, 35,

Psychiatric Association.
Bartak. L.. Rutter, M.. & Cox, A. (1975). A comparative stud,v of infantile autism and specific clevelopmental receptive language disorder: I. The chilclren.
Brilish .lottrnctl o.l' Psrchian'.t," I 26, 127 145.

Boucher, J. ( I 998). SPD as a distinct diagnostic entity:
Logical considerations and directions for future
resealch. Inlerncttional,Iournal oJ' Longuage and

Bartak. I-., Rutter. M..

parative stud-v

of

&

Cox.

A. (1977). A

infantile autism and

comspecific

clevclopmental rcceptive langr-rage disorders. IIL
Discliminant function analvsis. Journal of Autisnt
antl Childhood Schi:ophrenia.7. 383 396.

Bishop. D.VM. (1997). Uncoutntott t.rnderstanding.
Det,elopntent und clisorcler.s of langtruge comprehen-

sion in childreil. Hove, UK: Psychology Prcss.
Bishop. D.VM. (1998). D,^r,elopment of the children's

communication chccklist (CCC): A rncthod fbr
asscssing elualitatile aspccts of conmunicative

ffi

ilf
illt
liil
fl
u:_

rl

ir.npairment in childlen. Journa| of Chiltl P,stclto!og.t.
und P.st.'chianl. J9. 879 892.
Bishop. D.VM. (2000). What s so specral about Asperger
syndromc? The need fbr furthcr explolatton of thc
bolder'lands of autism. In,{. Klin. F.R. rolkmar. &
S.S. Sparror'r' (Eds). Ly;erger "s1:pllyouru (pp. 25:1
277). Nov Yori<: Gr-rilfbrd Pless.

Bishop. D.{N1., & ,A.dams. C. (1989). Conr.,ersarioniLl

qi

{p

Ln

71 81.
(1989). Conyerscrtionol

L'ontntttnic:atictn Disorclers" JJ,

Brinton. 8..

& Fuiiki. M.

ttluntrgenent u,i.th language-impuiretl children:
P ru g n1

ct t i

c

a s. e.

s

meil t ctntl

i

n te

rv

enlion. Rockville.

MD: Asperr Publishers.
Brinton, 8., & Fujiki. M. (1993). Language. social

slrills, and socioemottonal behavior. Language,
Speech, and Hetu'ing Sert,ices in Schools, 24, 194
1

98.

D.. Baher. L., Rutter. M., & Manhood. L.
989). Infantile autism and developmental r'eceptive

Cantu'el1.
(

1

dysphasia: A comparative fbllor,v-up into micldlc
childhood. Jout'nul of Atttisnt crntl Developmentul
Disortlers. 19, 19 31.
Conii-Rarrsclen. G.. Crutchley. A., & Botting. N.
( 1 997). The extcnt to ri,hich psychornctric tests diflerentiate subgroups of children with SLl. ,loLtntu/
o/ S1teeth, Lctnluage. und Hecu.ing Reseurch, 10.

765 777.
Clraig, il K (1993). Social skills of chiich'en u,ith
specific language impairment: Peer lelationships.

characteristics ol children r.r'ith sernantic pragntarrc

Lctrtguttge, Speech, und

disolder. II. What leatures lead to a judgen-rent of
inappropliac.v'.) B ri t i h J o m n l o f D i.s o rtl er "s o /' C.- o n t

21.2A6 215.

s'

ntrulicLtrion. 21"

u

21I

263.
Bishop, D.{M.. & Aclams, C. (1991). What clo reftr'entral commrLnication tasks measure? A stucl1' of
chiIdren rvith specific iangr-Lage impairment. App|ied
Psvt'hr.tlingttisllc,s, 11.

1

99 215.

Bishop, D.1M.. & Adams. C. (1992). Conrprehension
problems in children with specilic language impairrnent: Literal and inf'erential meaning. Jotnurl o.;f
Specc:h

f;

877 900.

untl Itearing Reseat't-lt. -t-t. 119 129.

I{eoring

Sen.'ice-s

in

St'lt.ool.s.

Frith. U. (1989). A ner.r. look at langua-qe and comrnunication in arLtisr-n. B ri ti s h Journ o / of' I)i..t rde rs
ct

o.f Contuuurit'ution.

GaJlaglrer.

T

)1.123

150.

(Ed.). (.1991l. Prugntdtic. of lunguoge;
Szin Dicgo: Singr-rlar Pub-

(llinicttl prct('ti(e rs.sae.r.

lishing Group.
Gertner, B.L.. Rice. Nlt.L.. & Hadlel'. PA. (199,+).
Influence of corrlnunicatire cornpetenee on feer
preferences in a pleschool classloom. Journul of'
Speech untl Itettrin,g Reseut'c.h, 3i. 913 923.

6. PRAGMATICLANGUAGE IMPAIRMENT
it
.ii

;
0
;$l

Goldin-Meadow, S., & Mylander, C. (1998). Spontaneous sign systems created by deaf children in two
cultures. Nature, 39 I, 219281.
Goldman, L.G. (1987). Social implications of language
disorders. Reading, Writing and Lenrning Disabilities, 3,119-130.
Hadley, PA., & Rice, M.L. (199i). Conversational
responsiveness of speech and language impaired
preschoolers. .

Journal

of

Speech and Hearing

Research, 34, 1308-1317 .
Kanner, L. (1946). Irrelevant and metaphorical language
in early infantile autism. American Journal of Psychiatry,103,242 246.

&

Lister Brook, S.,

Bowler, D. (1992). Autism by

another name? Semantic and pragmatic impairments

in children. Journal of Autism and Developmental
Disorders, 22, 6l-82.
Lord, C., & Pickles, A. (1996). Language level and
nonverbal social-communicative behaviors in autistic
and language-deiayed children . Journal of the American Academy of Child and Adolescent Psychiatry,
35,1542 1550.

l

Mawhood L. (1995). Autism and developmental language disorder: Implications from a follow-up in
early adult 1ife. Unpublished Ph.D. thesis, University of London, UK.
Miller, J. (1991). Research on language disorders in
children: A progress report. In J. Mil1er (Ed.).
Research on child language disorders (pp. 3-22),
Austin, TX: Pro-Ed.
Nowicki, S., & Duke, M.P. (1992). Helping the child
who doesn't

$

ft

in. Atlanta'. Peachtree Publishers.

113

O'Hare, A.E., Quew, R., & Aitken, K. (1998). The identification of autism in children referred to a tertiary
speech and language clinic and the implications for
service delivery. Autism, 2,171 180.
Paul, R., Spangle Looney, S., & Dahm, P.S. (1991).

Communication and socialization skills at a1es 2
and 3 in "late-ta1king" young children. Journal of
Speech and Hearing Research, -t4, 858-865.
Rapin, I. (1996). Developmental language disorders:
A clinical update. Journal ofChild Psychology and
Psychiatry, 37, 643-655.
Rapin, L, & Allen, D. (1983). Developmental language disorders: Nosologic considerations. In U.
Kirk (Ed.), Neuropsychology of language, reading,
andspelling (pp. 155-18a). New York: Academic
Press.

& Rice, M.L. (1998). The socioemotional behaviors of children with SLI: Social
adaptation or social deviance? Journal of Speech,
Language, and Hearing Research,4l,688 700.

Redmond" S.M.,

Rice, M.L., Sell, M.A., & Hadley, P.A. (1991). Social
interactions of speech and language impaired children. Journal ofSpeech and Hearing Research,34,

1299 1307.
Shields, J., Varley, R., Broks, P., & Simpson, A. (1996a).
Hemispheric function in developmental language
disorders and high-level autism. Developmental
Medicine and Child Neurology, 38, 473-486.
Shields, J., Varley, R., Broks, P., & Simpson, A. (1996b).
Social cognition in developmental language disorders
and high-level auttsm. Developmental Medicine and
Child Neurology, 38,487 495.

F
llt

Speech and language
impairments in children:
Causes, characteristics, intervention
and outcome
edited by
Dorothy V.M. Bishop
University of Oxford, UK

Laurence B. Leonard
Purdue University, lndiana, USA

!el: fl {:

Pragmatic language impairment in relation to autism and SLI

More Related Content

Similar to Pragmatic language impairment in relation to autism and SLI (20)

More from Dorothy Bishop (20)

Recently uploaded (20)

Pragmatic language impairment in relation to autism and SLI