Beruflich Dokumente
Kultur Dokumente
Frederick Furniss
The Hesley Group, Doncaster and Centre for Applied Psychology,
University of Leicester, Leicester, UK
and
Ann Walter
Derbyshire Children’s Hospital, Derby, UK
Abstract
Introduction
Address for correspondence: Alinda Gillott, Clinical Psychologist, Learning Disability Service, Highbury
Hospital, Highbury Road, Bulwell, Nottingham, NG6 9DR, UK.
et al., 1985). Theory of mind (ToM) relates to the ability to infer other people’s
mental states (such as their thoughts, beliefs, desires and intentions) and to use
this information to make sense of, and predict, behaviour (Howlin et al.,
1999). Thus, impairments in mentalizing ability have implications for both
understanding social interactions and communicating personal needs. Utiliz-
ing tests of rst and second-order theory of mind (Perner and Wimmer, 1985;
Perner et al., 1987), research has consistently demonstrated that children with
autism have a de cit in this area (Baron-Cohen et al., 1985; Baron-Cohen,
1989; Frith, 1989; Yirmiya et al., 1996).
A small number of studies have investigated the theory of mind perfor-
mance of children with speci c language impairments (SLI). Leslie and Frith
(1988) and Perner et al. (1989) compared children with autism with children
with speci c language impairment (affecting comprehension) using rst-order
false belief tasks. In both studies, the group with SLI performed signi cantly
better than the group with autism, with all children but one passing the tasks.
This nding led Leslie and Frith (1988) to conclude that language de cits
could not account for the impairment of mental state concepts in children with
autism. More recent studies have given further attention to the possible
implication of language disorder in the development of ToM de cits, and
have considered whether type of language disorder may be relevant. Prag-
matic, semantic and syntactic aspects of language have all been considered to
be related to theory of mind abilities, although possibly playing different roles
(Astington and Jenkins, 1999).
Children with semantic-pragmatic language disorders have been found to
perform as poorly as children with autism on tasks of false belief and
deception (Shields et al., 1996), and a number of researchers have argued
that this suggests that these disorders exist on the ‘borderlands’ of autism
(Shields et al., 1996; Botting and Conti-Ramsden, 1999). Studies using
children with phonologic-syntactic disorder have found less clear results,
with these children performing better than children with autism, but not always
as well as age-matched controls without any developmental disability (Shields
et al., 1996; Ziatas et al., 1998). One signi cant question with respect to these
children has been whether any impairment in performance on ToM tasks
re ects a real impairment in underlying ToM ability or rather results from
dif culties in managing the linguistic complexity of these tasks. Miller (2001)
systematically varied level of linguistic demand in ToM tasks and found that
children with SLI performed similarly to same-age peers when linguistic
complexity was low, but similarly to younger children matched with them
on language comprehension ability when linguistic complexity was high,
suggesting that linguistic ability rather than underlying ToM ability limits the
TOM ability in children with SLI 3
correct ‘mental state’ answers than the normally developing children, but to
produce a more diverse pattern of errors than the children with autism,
producing some ‘physical state’ answers but also producing many incorrect
mental state answers in which the character’s behaviour is explained as arising
from an inferred mental state but the linguistic complexity of the story results
in an inappropriate inference being made.
Method
Prior to the study, ethical approval was sought from two local National Health
Service Trusts. Following some slight amendments to the original proposal for
the study, full ethical approval was granted.
Participants
Fifteen children with speci c language impairment (phonologic-syntactic)
were compared with 15 high functioning children with autism and 15 normally
developing children. Children were between the ages of eight and 12 years
(mean age across all groups 10.26 years) and were individually matched across
groups for chronological age and gender. Each group consisted of 13 boys and
two girls. Children from the group with SLI received their diagnosis prior to the
study by a speech and language therapist from a speci c language impairment
team. Children with additional language impairments (such as semantic or
pragmatic dif culties) were excluded. Children from the group with autism had
also received a prior diagnosis according to the Diagnostic and Statistical
Manual of Mental Disorders Fourth Edition (American Psychiatric Associa-
tion, 1994) criteria by a clinical psychologist or paediatrician. Children from
the normally developing group were free of any diagnosis of psychiatric or
developmental disorder. Children in all groups had intellectual and reading
abilities within the average range, as assessed by their teacher or therapist. All
participants were assessed using the Communication Domain of the Vineland
Adaptive Behavior Scales – Interview Edition (Sparrow et al., 1984). Mean
standard scores were 100.3 (SD ˆ 17.9) for the typically developing children,
72.4 (SD ˆ 14.5) for the children with SLI and 75.8 (SD ˆ 12.4) for the
children with autism. One-way analysis of variance demonstated the existence
of a signi cant effect of group on the Vineland Communication score
(F(2,42) ˆ 15.2, P < 0.01). Subsequent Scheffé tests con rmed that the mean
Vineland Communication scores of the groups with autism and SLI did not
differ, but that both were signi cantly lower than that of the group of typically
TOM ability in children with SLI 5
Measure
The Strange Stories (Happé, 1994) were designed to test theory of mind in
more able children and adults with autism. The stories concern the different
motivations that can lie behind everyday social utterances that are not literally
true such as sarcasm, pretence, joking and lying.
The Strange Stories consist of 12 short vignettes, each accompanied by a
picture and two test questions: a comprehension question (‘was it true what x
said?’) and a justi cation question (‘why did x say that?’). The 12 story types
comprised lie, white lie, joke, pretend, misunderstanding, persuade, appearance=
reality, gure of speech, sarcasm, forget, double bluff and contrary emotion. In
each Story a character says something which is not literally true and the
participant is asked to explain why the character said what he or she said.
Answers given in response to the ‘why?’ question were evaluated for
correctness and for reference to mental states or physical (literal) states.
Answers classi ed as mental state answers included ‘he doesn’t want to hurt
her feelings’ (white lie), ‘she was being sarcastic’ (sarcasm) and ‘she’s just
pretending’ (pretend). Answers classi ed as physical state answers included ‘it
looks like a telephone’ (pretend), ‘because her hair was short’ (joke) and
‘because it’s raining’ (sarcasm). Answers classi ed as incorrect included those
which made references to mental states inappropriate to the context of the story
such as ‘to get her jealous’ (pretend) and ‘he wanted to play a trick’ (forget).
Answers were scored on a 3-point scale: 2 points for a correct mental state
answer, 1 point for a physical state answer, and 0 points for an incorrect answer.
Interrater agreement
The data were coded by two raters, the rst author and a Consultant Clinical
Psychologist who was blind to group status. Interrater agreement was calculated
using Kappa. Kappa for all stories across all three groups was found to be 0.77,
the range across stories being 0.64–0.87. Disagreements between raters were
discussed until a consensus scoring was agreed.
Procedure
The rst author and child sat at a low table where the child was able to see and
read the stories. The set of Stories was introduced as follows: ‘I am going to
6 Child Language Teaching and Therapy
read you some stories and I’d like you to listen carefully. I will then ask you
two questions at the end of each story’. After each Story was read out aloud to
participants, they were asked the two test questions. The rst question ‘was it
true what x said?’ was treated as a test of comprehension and was not scored.
(The second justi cation question was not asked until comprehension was
established). The second question ‘why did x say that?’ was then asked and the
participant’s answer was recorded verbatim.
Other measures, the results of which are reported elsewhere (Gillott et al.,
2001), were also administered.
Results
The median number of mental state, physical state and incorrect responses
given by children in each group are presented in Table 1. Kruskal–Wallis
analyses of variance showed signi cant between-group differences for the
number of mental state answers given (w2 ˆ 7.4, df ˆ 2, P < 0.05) and for the
number of incorrect answers given (w2 ˆ 8.7, df ˆ 2, P < 0.05), but not for
the number of physical state answers given. Comparisons between the
groups were made using Mann–Whitney U tests. For the number of mental
state answers given, signi cant differences were found between the normally
developing groups and groups with autism (U ˆ 54.0, P < 0.05), and between
the normally developing and language impaired groups (U ˆ 65.0, P < 0.05)
but not between the language impaired groups and groups with autism
(U ˆ 88.0, NS). For the number of incorrect answers given, signi cant
differences were found between the normally developing groups and groups
with autism (U ˆ 47.0, P < 0.01) and between the language impaired groups
and groups with autism (U ˆ 62.5, P < 0.05), but not between the normally
developing and language impaired groups (U ˆ 91.0, NS).
Normally
developing SLI Autism
(n ˆ 15) (n ˆ 15) (n ˆ 15) w2
Mental state responses 8 (1.5) 7 (1.5) 6 (3.5) 7.4*
Physical state responses 2 (1) 4 (1.5) 2 (1.5) 4.9
Incorrect responses 1 (1) 2 (1) 3 (2.5) 8.7*
*P < 0.05.
TOM ability in children with SLI 7
Discussion
In this study, normally developing children gave more correct ‘mental state’
responses to the Strange Story questions than either children with phonologic-
syntactic language impairments or children with autism, with the latter two
groups performing similarly in this regard. Unexpectedly, however, the groups
did not differ overall in the number of ‘physical state’ responses given. Rather,
the children with autism gave more incorrect responses than the normally
developing children and children with language impairment, with no signi -
cant difference between the normally developing children and those with
language impairments in the number of such responses. Our results suggest
that children in the group with autism responded as if they did indeed have a
general ‘Theory of Mind’ insofar as they explained the behaviour of characters
in the Strange Stories in ways which suggest that they attribute mental states to
the characters, but that they were unable to accurately describe the mental state
likely to result in the behaviour described. The children in the language
impaired group produced a more heterogeneous pattern of responses, attribut-
ing correct mental states no more frequently than the children with autism, but
also attributing incorrect mental states no more frequently than the typically
developing children.
In contrast to the ndings of Shields et al. (1996) and Ziatas et al. (1998), the
children with phonologic-syntactic language impairment in this study did not
succeed on a ToM task more frequently than the group with autism. There may
be several reasons for this nding. It may be that there are substantial theory of
mind de cits in children with phonologic-syntactic disorders which are more
evident with the more complex Strange Stories than with the rst- and second-
order theory of mind tasks used in previous research with this population. We
cannot, however, exclude the possibility that the apparent similarity between
the performances of the children with autism and children with SLI results from
different impairments, with the performance of the children with autism being
limited by impairments in theory of mind ability, but the performance of the
8 Child Language Teaching and Therapy
that the groups with autism and SLI performed similarly on this measure does
however suggest that the communication dif culties of the children with SLI
were no greater overall than those of the children with autism and therefore
that the dif culties of the language impaired group on the Strange Stories task
cannot be entirely attributed to purely linguistic dif culties.
Heterogeneity in the group of children with SLI may also have contributed
to the pattern of results seen. All children in the group with language
impairments had received a diagnosis of phonologic-syntactic language
impairment, which has been established as a robust language disorder
subgroup (Conti-Ramsden et al., 1997; Conti-Ramsden and Botting, 1999).
However, Chapman (1991) has argued that children with language disorder
can be variable in their language skills; and in fact, although Conti-Ramsden et
al. (1997) found psychometrically robust subgroups of SLI, they found in their
later study that individual children’s dif culties do change over time (Conti-
Ramsden and Botting, 1999). Therefore, ndings in the present study may be
attributed to other unreported or newly emerging language problems, such as
pragmatic or language comprehension dif culties, in some members of the
language impaired group. Further research exploring theory of mind abilities
in children with different speci c language dif culties would provide some
clari cation on this issue.
Finally, the possibility must be considered that the patterns of ToM abilities
we observed might be accounted for simply by between-group differences in
intellectual ability. Although all the children in our study were evaluated by
teachers and therapists as having intellectual abilities within the normal range,
we undertook no formal intellectual assessments. We cannot therefore rule out
the possibility that overall performance IQ level might be higher in the group
of children with autism than in the group of children with SLI, and that such
an advantage in intellectual ability may have enabled the children with autism
to compensate for intrinsically greater dif culties with ToM so as to equal the
performance of the children with SLI in giving correct mental state answers.
In conclusion, the aim of this study was to explore theory of mind ability in
children with phonologic-syntactic language impairment. Early research
suggested that children with language disorders are not impaired in their ability
to mind read (Leslie and Frith, 1988; Ziatas et al., 1998) and that this
impairment is speci c to children with autism (Baron-Cohen et al., 1985;
Frith, 1989). More recently, ToM impairment has been indicated in some
groups with SLI, although without clear results (Johnston et al., 2001; Miller,
2001). In the present study, children with SLI performed similarly to children
with autism in making appropriate mental state attributions to explain the
behaviour of characters in the Strange Stories task. We cannot exclude the
10 Child Language Teaching and Therapy
Acknowledgements
This work was undertaken by Leicestershire and Rutland Healthcare NHS Trust
(or former Trust) which received a proportion of its funding from the NHS
Executive; the views expressed in this publication are those of the authors and
not necessarily those of the NHS Executive. The authors wish to thank the two
anonymous reviewers for their helpful editorial comments.
References