Language is a complex phenomenon and its development can be delayed or disordered. The major forms of developmental language disorder found in childhood broadly parallel the types of acquired communication deficit associated with left and right unilateral brain damage in adults. Different types of developmental language disorder have been identified. For instance Rapin and Allen (1987) suggested several subtypes, including 'phonologic-syntactic syndrome' (a selective difficulty with language form, but normal language content) and semantic-pragmatic syndrome' (use of superficially complex language with clear articulation, but difficulty with the use and understanding of language). The clinical picture of the semantic-pragmatic syndrome appears to change with age: after a severe delay in starting to speak, language use emerges, with echolalia, jargon, and auditory inattention. By age 4 to 6 years, expressive skills may score in advance of comprehension, and the child may use stereotyped language and ask frequent questions. The child aged 7 years and over is likely to use fluent, grammatically complex language but to show minor problems with phonology and syntax; there may be word-finding problems and semantic errors. By this age, comprehension difficulties are not evident on concrete, literal tasks, but the child interprets over-literally and has poor conversation skills with inappropriate language use. There may be poor comprehension and use of nonverbal communication and prosody with poor social skills.

Evidence from neuropsychological investigations of unilateral brain lesions (Smith 1966, Zurif 1980) indicates that damage to the left hemisphere is generally associated with impairment of the canonical components of language form (phonology, syntax and lexical semantics). However, there is more to language than the context-free, componential aspects which are affected by aphasia: other, non-componential, non-literal and context-bound aspects of language are vital to successful communication.

Right hemisphere lesions can result in deficits in such pragmatic skills leading to abnormalities of discourse and conversational management and to impaired understanding of non-literal language such as metaphor and humour (Gardner et al. 1983, Hirst et al. 1984, Code 1987, Bryan 1988). Conversational deficits reported following right hemisphere injury include abnormalities of turn-taking; discourse which is often described as 'rambling' - reflecting topic drift; and abnormalities of non-verbal communication such as poor eye contact (Winner and Gardner 1977, Wapner et al. 1981, Brownell et al. 1983, Brownell et al. 1984). A number of studies have demonstrated a right hemisphere superiority for interpreting prosodic features of speech such as intonation and stress (Dwyer and Rinn 1981). In cases of right hemisphere lesions, impairments are also evident in the comprehension of speech acts which can be used either directly or indirectly - for example, the polite use of a question form such as 'Can you pass the salt?' as an implied request (Hirst et al. 1984).

There is some evidence of a similar pattern of communication impairment in patients who have suffered right hemisphere lesions in childhood. Damage to the right hemisphere which is suffered early in life or which is congenital can give rise to a constellation of deficits characterised by emotional and interpersonal difficulties, shyness, visuospatial difficulties, and inadequate pragmatic skills. Such subjects lack eye contact and do not use normal prosody or gesture. Most also have attentional difficulties (Weintraub and Mesulam 1983, Voeller 1986). This pattern of deficits resembles in some respects that of autistic spectrum disorders (Wing 1988).

Interesting comparisons can be made between the communication deficits observed in some individuals with acquired right hemisphere lesions and those described in children with the developmental 'semantic-pragmatic disorder' (Shields 1991): both groups have relatively intact language form, using fluent, grammatically complex language, but show pragmatic disability in their abnormal language content and use. With regard to developmental disorders of the phonologic-syntactic type, it seems likely that some form of left hemisphere dysfunction must underlie developmental language disorder just as it typically underlies acquired aphasia, but evidence for such brain abnormality in developmental conditions is scarce, although modern imaging techniques are beginning to reveal structural aberration underlying anomalous neurodevelopment (Jernigan et al. 1991).

Comparisons between developmental and acquired conditions should always be made with caution, and there is little evidence, as yet, that developmental language disorders of the phonologic-syntactic type have their origin in left hemisphere dysfunction. However. it does seem plausible that these two broad groups of developmental language disorder (phonologic-syntactic and semantic-pragmatic) may be linked to contrasting hemispheric dysfunction, just as lesions of each of the cerebral hemispheres are known to give rise to contrasting forms of acquired language disorder.

When language was considered to be the major deficit in autism, theories of left hemisphere dysfunction were popular (Delong 1978). Prior and Bradshaw (1979) challenged these notions and Fein et al. (1984), failing to demonstrate left hemisphere lesions as critical in autism, pointed out that the deficits in prosody, social use of language and ability to read emotional expression in language found in the communication of children with autism were more likely to be symptoms of right hemisphere lesions. Goodman (1989) discussed the possibility of multiple, co-existing neurological deficits in autism and thought that right hemisphere damage might be particularly implicated in social communication impairments, since he argued that non-verbal communication depends on right hemisphere systems. The idea of right hemisphere dysfunction had been suggested as early as 1960 by Sarvis, who reported the case of a child who developed autistic symptoms following an early lesion in the right temporal lobe. Bishop (1993) reviewed neuropsychological studies of autism and favoured the notion of frontal lobe and limbic system dysfunction.

If right hemisphere dysfunction does contribute to both autism and semantic-pragmatic language disorder, then the relation between the two disorders is a significant issue. There have been suggestions that semantic-pragmatic disorder may form part of the autistic continuum of disorders (Brook and Bowler 1992), since the communication symptoms of more able children with autism resemble those of semantic-pragmatic disorder. It is clear that more research is needed in this area.

In this study, a battery of neuropsychological tests which were selectively sensitive to left/right hemisphere damage was given to groups of children with phonologic-syntactic disorder (P), semantic-pragmatic disorder (S) or high-level autism (A) and to a control group of normal children (C).

Hypothesis A, the primary hypothesis, was that of contrasting hemispheric dysfunction between groups. It was predicted that children with phonologic-syntactic disorder (group P) would show a pattern of results consistent with left hemisphere dysfunction, and those with semantic-pragmatic disorder (group S) would show the reverse pattern, consistent with right hemisphere dysfunction.

Hypothesis B was that group S and group A would show a similar pattern of results in the battery of hemisphere function tests, in line with the view that semantic-pragmatic disorder can be viewed as part of the autistic spectrum.

Method
SUBJECTS
Four groups of 10 children in the age range 7 to 11 years were included in the study.

Two groups with language disorder were selected. These differed in the nature of their language impairment: the first (P) had a disorder of language form (phonology and syntax); the second (S) had semantic and pragmatic difficulties. The two groups were recruited through departments of speech and language therapy in the Yorkshire region. The groups were formed by the local therapist and the first author, rating the children's communication symptoms on checklists using Rapin and Allen's (1987) criteria for 'phonologic-syntactic syndrome' and 'semantic-pragmatic syndrome'. All subjects had specific developmental language difficulties severe enough to cause special educational needs. All had been assessed by educational psychologists and were being educated in facilities for children of normal intelligence. None was known to have established brain pathology: to our knowledge no scans were available and we did not carry out any imaging studies. None had been diagnosed as having autism.

A third group (A) comprised children with high-level autism, some of whom were being educated in mainstream schools, and some in the 'high-ability' class of a school for children with autism. All had been given a medical diagnosis of autism by a paediatrician, and met the DSM-III-R criteria for this diagnosis when rated by their teachers and the first author.

The fourth group (C), a control group, comprised normal children attending a primary school in the same area. They were selected by their headteachers as being in the average range of ability, with no known developmental abnormality and with English as mother tongue.

The subjects ranged in age from 7 to 11 years, but age was controlled by matching across the four groups, each set of four subjects being within 9 months of age of each other. The groups were also matched for socio-economic status (according to the Office of Population Censuses and Surveys). No female subjects with language disorder were available, reflecting the greater prevalence of developmental language disorder in boys (Eme 1979). Since those tests which had norms had been standardised on groups of boys and girls, it was decided to select a similar control group of boys and girls. Subject's details can be found in Table I.

PROCEDURE
Each subject was tested individually by the first author in their own school or home. A battery of tests was selected with the aim of comparing strengths and weaknesses of the groups, seeking signs of contrasting hemispheric dysfunction. Some of the tests selected had been designed for children and had norms: others were adapted from those known to be reliable in indicating hemispheric differences in adult subjects with brain lesions. The contents of the test battery are listed in Table II and explained in the Appendix.

Results
STATISTICAL TREATMENT OF DATA
A reasonable way of testing the predictions would be to examine: (I) in the left hemisphere test battery, whether group P in general did less well than groups S, A and C; and (ii) in the right hemisphere test battery, whether groups S and A in general did less well groups P and C. These tests were effected by conducting a profile analysis (Morrison 1990, Johnson and Wichern 1992) to discover whether there are general differences between the groups, followed by more specific between-group comparisons. All data were rendered commensurable by conversion to percentages if necessary.

__________________________________________________________________________
Group        Age          Handedness     Sex       Socio-Economic Status1
          Mean  SD        Right Left  Male Female      Range   Mode
__________________________________________________________________________
S        105.9 19.78      9     1     10   0           l-4     2
P        104.4 19.44      9     1     10   0           l-4     2
A        108.9 18.54      8     2      9   1           l-4     2
C        106.3 19.84      8     2      5   5           l-4     2
__________________________________________________________________________

______________________________________________________________________________________________
Right hemisphere bias                            Left hemisphere bias
______________________________________________________________________________________________
RH1 - British Ability Scales: block design       LH1 - British Picture Vocabulary Scales
RH2 - Coral Blocks                               LH2 - Test of Reception of Grammar
RH3 - Benton Line Orientation Test               LH3 - British Ability Scales: Similarities
RH4 - Benton Face Recognition Test               LH4 - British Ability Scales: Digit Span
RH5 - Design Learning Test                       LH5 - British Ability Scales: Reading
RH6 - Postural Expression Test                   LH6 - British Ability Scales: Word Definition
RH7 - Rivermead Memory Test: Faces               LH7 - Word List Learning Test
RH8 - Visual Object and Space Perception         LH8 - Verbal fluency. Sound Test
______________________________________________________________________________________________

Tests on assumptions
With sample sizes as small as 10, tests on assumptions were unlikely to give very clear conclusions. However, tests on univariate normality in each sample were performed by computing Pearson correlation co-efficients from quantile-quantile plots of ordered scores against standard normal quantiles and testing at the 0.10 significance level. For both batteries of tests, the normality assumption was rejected for approximately half the samples and other samples gave high correlations mainly by virtue of the substantial number of tied scores. Transformation of the data to logits did little to improve the situation, and the large numbers of tied scores, of course, remained.

Tests on the equality of the covariance matrices were conducted using the test devised by Box (1950), although since the distribution of its statistic depends strongly on the assumption of multinormality its results are likely to he of dubious validity. Since the number of observations fell below 20 in each sample, and the number of variables exceeded five, Box's statistic distributed as F was computed in addition to the more usual X² statistic. All statistics exceeded the critical values at p<0.0l, suggesting that equality of covariance matrices could probably not be assumed.

The above tests showed frequent violations of assumptions with regard to departures from univariate normality and inequality of covariance matrices, and also with regard to the large number of ties in the data. However, the samples were very small, and because of this the assumptions could not be adequately tested. All these considerations, taken

together, suggested that parametric multi-variate analyses, which rely on these assumptions, were probably inappropriate. It was therefore decided to use randomisation tests (Edgington 1980, Manly's 1991) as the basis of the analyses. Such an approach would still make it possible to compute multivariate analyses, such as profile analysis, but would overcome the problem of violation of parametric assumptions by establishing significance by randomisation.

RIGHT HEMISPHERE TEST BATTERY
For the right hemisphere battery of results, running a randomisation test on the ANOVA version of profile analysis, we found significance levels of p=0.0019 (Groups effect), p<0.000l for the repeated-measures (Tests) effect and p=0.6416 for the interaction (Parallelism) effect.

TABLE III
Right hemisphere battery: one-way ANOVAs

_________________________________________________________________
Test                F value of         Significance level by
                    actual data         randomisation test
_________________________________________________________________
BAS Block Design       2.28                  0.097 (ns)
Corsi Blocks           6.25                  0.002
Line Orientation       1.32                  0.281 (ns)
Face Recognition       4.05                  0.015
Design Learn           2.84                  0.055 (ns)
Postural Expression    7.27                  0.001
Rivermead Face Memory  2.87                  0.046
VOSP                   6.34                  0.002
_________________________________________________________________

TABLE IV
Right hemisphere battery: two-tailed p values

_______________________________________________________________________________
Test                    P vs S  P vs A  P vs C  S vs A  S vs C  A vs C
_______________________________________________________________________________
(BAS Block Design)1     0.098   0.316   0.348   0.632   0.032   0.115
Corsi Blocks            0.0063* 0.0012* 0.477   0.841   0.037   0.014
Line Orientation)1      0.162   0.358   0.486   0.836   0.091   0.211
Face Recognition        0.580   0.162   0.068   0.381   0.019   0.0024*
Design Learn)1          0.233   0.113   0.308   0.814   0.039   0.014
Postural Expression     0.135   0.040   0.077   0.500   0.0048* 0.0008*
Rivermead Face Memory   0.058   0.021   1.0     0.936   0.119   0.067
VOSP                    0.022   0.030   0.368   0.877   0.0002* 0.0016*
________________________________________________________________________________

We could conclude that the profileswere parallel and accept the first result that the groups differed significantly. (The significance of the repeated-measures effect will be ignored, differences between the mean scores across the tests are not of interest.) The graph of the profiles (Fig. I) showed clearly that groups S and A were approximately coincident and that their mean results were lower than those of groups P and C.

To examine differences between the groups within each of the eight tests in the right hemisphere battery separately, F statistics from individual one-factor independent-samples analyses of variance were computed in randomisation tests. They produced the results shown in Table III.

Modified Fisher protected least significant difference (PLSD) tests on individual comparisons gave the results presented in Tables IV and V. Asterisks to indicate statistical significance were attached only if the p value was less than a value deriving from application of the Bonferroni inequality, excluding comparisons for which the means were not expected to differ. This gave p =0.05/4 = 0.0125 as the critical value in both tables. The comparisons between P and C, and between S and A, are included merely to show that, if one tests these differences even against the 0.05 'per comparison' error rate, none of them is significant.

Where a difference had been predicted, its direction had also been predicted. We could therefore use one-sided significance tests for the four sets of predicted differences. Table V presents the one-sided significance levels, against the modified Bonferroni p value of 0.0125. Since no predictions about differences were made for the comparisons of P with C and S with A, one-sided tests were not computed in these cases.

TABLE V
Right hemisphere battery: one-tailed p values

_____________________________________________________________________________
Test                         P vs S  P vs A  P vs C  S vs A  S vs C  A vs C
_____________________________________________________________________________
(BAS Block Design)1          0.048   0.162                   0.017   0.059
Corsi Blocks                 0.0034* 0.0008*                 0.019   0.082*
(Line Orientation)1          0.079   0.180                   0.048   0.111
Face Recognition             0.286   0.086                   0.0098* 0.0016*
(Design Learn)1              0.111   0.057                   0.022   0.0071* 
Postural Expression          0.066   0.021                   0.0032* 0.0006*
Rivermead Face Memory        0.026   0.012*                  0.061   0.035
VOSP                         0.010*  0.015                   0.0001* 0.001*
___________________________________________________________________________

_________________________________________________
Test  	  		One-tailed  Two-tailed
_________________________________________________
(BAS Block Design)      0.0083*     0.0174*
Corsi Blocks            0.0001*     0.0002*
Line Orientation)       0.0277*     0.0561
Face Recognition        0.0064*     0.0116*
(Design Learn)1         0.0056*     0.0105* 
Postural Expression     0.0003*     0.0005*
Rivermead Face Memory   0.0018*     0.0038*
VOSP                    0.0001*     0.0001*
_________________________________________________

The most direct test of the hypotheses (which overcomes the disadvantage that comparisons within tests are not independent) was to compute specific contrasts (again using randomisation tests) between the S and A groups combined, on the one hand, and the P and C groups combined, on the other. The results, giving one- and two-sided p values, are presented in Table VI. For each test in the battery, only one contrast was being calculated. so 0.05 could be used as the critical level for significance. In all of the one-tailed comparisons. and all but one of the two-tailed, the combined P and C groups scored significantly better than the combined S and A groups.

Group S scored significantly lower than group P on visual-nonverbal short-term memory (Corsi Blocks) and on Visual Object and Space Perception (VOSP) (see Table V). Groups P and C combined scored significantly higher than groups S and A combined on all tests in the right hemisphere battery (Table VI). There were no significant differences between group P and the control group, but group S scored significantly lower than the controls for Face Recognition, Postural Expression, and VOSP (see Table V).

On tests targeting the right hemisphere, therefore, the group with semantic-pragmatic language disorder did show weaknesses which were not found in the group with phonologic-syntactic language disorder.

Fig. 2. Mean scores of children in four experimental groups on tests of left hemisphere function. For abbreviations. see Fig 1. Error bars show ±ISEM.

TABLE VII
Left hemisphere battery: one-way ANOVAs

______________________________________________________________
Test                   F value of      Significance level by
                      actual data       randomisation test
______________________________________________________________
BPVS                      9.38               0.0003
TROG                      5.27               0.005
BAS Similarities          8.18               0.0005
BAS Digits                3.57               0.0227
BAS Reading               5.11               0.005
BAS Word Definition       7.42               0.0006
Word Learn                0.20               0.8936 (ns)
Verbal Fluency (Sound)    1.56               0.2155 (ns)
______________________________________________________________

LEFT HEMISPHERE TEST BATTERY
For the left hemisphere battery of results, running a randomisation test on the ANOVA version of profile analysis, we found significance levels of p=0.0002 (Groups effect), p<0.0001 for the repeated-measures (Tests) effect and p<0.0001 for the interaction (Parallelism). In this case we could not conclude that the profiles were parallel (as shown by the third significance level), and there was therefore some doubt about the scope of the first result that the groups differed significantly. (As before, the significance of the repeated-measures effect will ignored because differences between the mean scores across the tests are not of interest.)

TABLE VIII
Left hemisphere battery: two-tailed tests

________________________________________________________________________
Test                     P vs S  P vs A  P vs C  S vs A  S vs C  A vs C
________________________________________________________________________
BPVS                     0.349   0.815   0.0013* 0.277   0.0001* 0.0048*
TROG                     0.885   0.324   0.0033* 0.214   0.0003* 0.044
BAS Similarities         0.735   0.169   0.0007* 0.323   0.002*  0.017
BAS Digits               0.201   0.421   0.0003* 0.712   0.095   0.074
BAS Reading              0.0022* 0.022   0.0003* 0.850   0.364   0.700
BAS Word Definition      0.0068* 0.392   0.072   0.090   0.0003* 0.014
(Word Learn)             0.89    0.808   0.574   0.940   0.567   0.435
(Verbal Fluency (Sound)) 0.29    0.449   0.523   0.662   0.471   0.155
________________________________________________________________________

The graph of the profiles (Fig. 2)showed clearly that group C scored higher than the other groups, but that the size of this difference varied across the tests in the battery. Our prediction was different from that result: we had predicted that group P would have poorer scores than the others. With the exception of British Ability Scales (BAS) Digits and BAS Reading, the means of group P seemed to fall within or close to those of groups S and A.

F statistics from individual one-factor independent-samples analyses of variance were computed in randomisation tests of between-group differences in each of the eight tests in the left hemisphere battery separately. The results are shown in Table VII.

The results of Bonferroni-modified Fisher PLSD tests, again by randomisation, are given in Table VIII. Only two-tailed tests are presented here, since the prediction that group P would have poorer scores than the other three groups clearly failed in the majority of cases. Furthermore, since these comparisons are a posteriori and all six possible comparisons are being made, the more stringent Bonferroni probability of p =0.05/6 = 0.0083 has been adopted as the criterion for significance.

In the case of BAS Word Definition, group P did significantly better than group S, not worse as predicted. However, this result was not out of line with the general finding that the means of group P were often similar to those of groups S and A. The more specific prediction that group P would do worse than group S seems, from the graph, to be upheld only in the case of BAS Digits and BAS Reading. In fact, the BAS Digits difference is clearly non-significant (p=0.201), and although the BAS Reading difference is significant (p=0.0022), the next test in the battery (BAS Word Definition) also gave a significant difference (p=0.0068) but in the opposite direction.

For the left hemisphere test battery, modified Fisher PLSD tests showed significant differences between the two groups with language disorder, group P scoring significantly lower than group S on BAS Reading (as predicted), but group S scoring significantly lower than group P on BAS Word Definition (see Table VIII).

For this test battery there were several significant differences between group P and the controls, with group P scoring significantly lower than the controls on sight reading (BAS Reading), auditory-verbal short-term memory (BAS Digit), verbal conceptual abstraction (BAS Similarities), receptive vocabulary (BPVS), and receptive syntax (TROG).

Group S also scored significantly lower than the controls on British Picture Vocabulary Scales (BPVS), Test of Reception of Grammar (TROG), BAS Word Definition and BAS Similarities, but not on BAS Digits. Thus the skill of sight-reading did split the groups in the way predicted, but other linguistic tests suggest that group S (and group A) may have dysfunction of both hemispheres or, alternatively, that some of the tests tapped behaviours which require bilateral processing.

CONCLUSIONS
Hypothesis A, the hypothesis of contrasting hemispheric function between the two groups with language disorder, group P and group S, was largely supported.

There was also support for hypothesis B, that semantic-pragmatic language disorder forms part of the autistic spectrum of disorders, in so far as there were no significant differences between groups S and A on either the right or left hemisphere test batteries. There was clear close coincidence between groups S and A in the right hemisphere tests battery profile.

Discussion
The results support the view that right hemisphere dysfunction (or bilateral dysfunction) is implicated in developmental semantic-pragmatic language disorder and in high-level autism. The right hemisphere test battery showed a consistent pattern of groups S and A scoring significantly lower than groups P and C.

The results from the left hemisphere test battery are less clear, with the control group showing better results than groups P, S and A overall, but variable contrasts between groups P and S+A. The implication of these findings may be that the tests used tap skills which are not based entirely in the left hemisphere. This may be evidence of bilateral distribution of components of the language system (Brownell et al. 1990. Howard et al. 1992) or for the involvement of bilateral processing in using the language system.

Some language skills differed between groups S+A and P, notably reading and word definition. Children with disorders of language form often experience difficulty with reading, whereas children with high-level autism are often 'hyperlexic', showing a dissociation between performance and understanding for written language with good mechanical reading ability but poor comprehension of what is read (Bishop and Rosenbloom 1987). Word definition and similarities are both linguistic tasks, but word definition is a lexical task with implicit rules, which involves accessing a complex set of conceptual and semantic ideas. In contrast, similarities is a more convergent task, involving superordinate hyponymic relationships and an analogical, formulaic task with overt rules. The pattern of differences found for these two 'linguistic' tasks was not predicted but might be explained by hypothesizing that the right hemisphere does play a part in accessing multi-componential lexical knowledge which is linked to world knowledge, and in staying within a cognitive context. Such an explanation could be validated by comparing performance on these two tests by subjects with acquired lesions of the right hemisphere. It seems possible that the involvement of higher non-linguistic cognitive functions in 'linguistic' tests clouds the distinction between 'left hemisphere' and 'right hemisphere' tests.

One might ask why, if a lateralised brain dysfunction does underlie the various forms of developmental language disorder, intrahemispheric and interhemispheric re-organisation do not occur, in view of findings of plasticity of function in the young brain. First, the notion of plasticity derives from studies of children with acquired brain lesions (Basser 1962), and the majority of children with developmental language disorder, and indeed the subjects of this study, have no postnatal history of overt cerebral insult. Hence an event that 'triggers' massive functional re-organisation might not occur. In addition, more recent studies of acquired brain lesions in children have shown that the extent of the plasticity in the young damaged brain is not as great as originally thought (Van Hout 1990), and even left hemisphere lesions suffered before the age of one can result in some degree of persistent language impairment (Vargha-Kadem et al. 1985).

The overall results strongly support the second hypothesis, that semantic-pragmatic language disorder forms part of the autistic continuum. Indeed, there were no significant differences between group S and group A on any of the tests. The two groups were selected by clinical diagnosis and none of group S had been diagnosed as having autism, but it is possible that, with the growing understanding of the nature of disorders of the autistic spectrum, such children may be seen to belong to the same broad clinical group, as suggested by Brook and Bowler (1992).

The results of this study lend support to the idea that different forms of developmental language disorder may he linked to contrasting forms of hemispheric dysfunction. In particular, it seems likely that phonologic-syntactic language disorder is linked to left hemisphere dysfunction and semantic-pragmatic language disorder to right hemisphere (or bilateral) dysfunction.

The study also indicates a link between semantic-pragmatic language disorder and high-level autism with the similar performance on both batteries of tests suggesting similar patterns of brain dysfunction in these two groups. A second study will compare the same four groups of children on tests of social cognition in order to further investigate the possibility that semantic-pragmatic language disorder forms part of the autistic spectrum of disorders.

Appendix
RIGHT HEMISPHERE BIAS TESTS:
RH1 - British Ability Scales (BAS), Block Design (Elliott et al. 1977): tests ability in constructing block designs: a visuo-spatial task. Damage to the right parietal lobe is known to produce dysfunction on such tests (McFie 1975, LeDoux et al. 1977).
RH2 - Corsi Blocks: a test of visual short-term memory (Milner 1971), which is a right hemisphere skill (De Renzi and Nichelli 1975).
RH3 - Benton Line Orientation Test (Benton et al. 1983): a test which is highly sensitive to right hemisphere (parietal lobe) function in adults (Benton et al. 1978).
RH4 - Benton Face Recognition Test (Benton et al. 1983): a test which is sensitive to right hemisphere lesions in adults (Benton and Van Allen 1968).
RH5 - Design Learning Test (Coughlan and Hollows 1985): together with the word learning test, this procedure was designed for use with adults. It provides a measure of the ability to learn a visual design, over five trials. Nonverbal memory and learning are right hemisphere skills (Newcombe 1969, Coughlan 1979).
RH6 - Posture Expression Test (Spence 1982): a test designed for young people, but without norms. The perception of facial and gestural emotion is a right hemisphere skill (Cicone et al. 1980).
RH7 - Rivermead memory test: faces (Wilson et al. 1990): a test of visual memory for faces, tested by recognition. rather than by recall. Facial perception skills can be disturbed by lesions of the right hemisphere (Benton and Van Allen 1968).
RH8 - Visual Object and Space Perception Test (VOSP) (Warrington and James 1991): a battery of eight tests, four of object perception and four of space perception, which are sensitive to right hemisphere lesions in adults (Warrington and James 1991). Those testing object perception are more sensitive to temporal lobe damage: those testing space perception are more sensitive to parietal lobe damage.

LEFT HEMISPHERE BIAS TESTS:
LHI - British Picture Vocabulary Scales (Dunn et al. 1982): a test of receptive vocabulary, with norms for British children. The comprehension of word meaning is frequently involved in aphasic disturbances of language, and has been shown to be impaired in patients with lesions of the left hemisphere (Coughlan and Warrington 1978, Damasio 1981, Benson and Geschwind 1985).
LH2 - Test of Reception of Grammar (Bishop 1982): a test of syntax comprehension, with norms for British children. Syntactic comprehension can be impaired by lesions of the left hemisphere (Caplan 1987, McCarthy and Warrington 1987, Caplan and Hildebrandt 1988).
LH3 - BAS Similarities (Elliott et al. 1977): a linguistic test, tapping convergent semantic skills of categorisation, which may be poor in aphasia and are associated with the left hemisphere. Word-finding skills can be impaired by lesions of the left hemisphere (Hecaen and Angelergues 1964).
LH4 - BAS Digit recall (Elliott et al. l977): a test of short-term auditory memory, known to be a left hemisphere skill, and poor in adults with aphasia, and in children with disorders of language form (DeRenzi and Nichelli 1975, Warrington et al. 1986).
LH5 - BAS Reading (Elliott et al. 1977): a test of sight reading (simple decoding, rather than comprehension) of single words, with norms for British children, known to be a left hemisphere skill. Acquired dyslexia (the disruption of reading) is associated with lesions of the left hemisphere (Greenblatt 1973, Friedman and Albert 1985).
LH6 - BAS Word Definition (Elliott et al. 1977): a verbal test, which may involve right hemisphere semantic skills (Joanette et al. 1988).
LH7 - Word List Learning (Coughlan and Hollows 1985): the verbal equivalent of the design learning task, this measures the ability to learn a list of words over five trials. Memory for auditory-verbal material can be disrupted by lesions of the left hemisphere (McFie and Piercy 1952, Newcombe 1969, Coughlan 1979).
LH8 - Verbal Fluency (Sounds): a test of the ability to generate words, beginning with a certain sound. Such tests are thought to be sensitive to anterior lesions of the left hemisphere (Benton 1968).

SUMMARY
Two groups of children with contrasting types of developmental language disorder (phonologic-syntactic and semantic-pragmatic) were compared with a group of children with high-level autism and with a control group of normal children on a broad battery of neuropsychological tests, known to be sensitive to left-right hemisphere damage. Significant differences found between the groups suggest contrasting forms of hemispheric dysfunction.

RÉSUMÉ
Fonction hémisphérique dans les troubles développementaux du langage et autisme de haut niveau Deux groupes d'enfants avec des types contrastants de trouble développemental du langage (phonologique-syntaxique et sémantique-pragmatique) ont été comparés à un groupe d'enfants avec autisme de haut niveau et avec un groupe contrôle d'enfants normaux sur une large batterie de tests neuropsychologiques, connus pour leur sensibilité aux atteintes de l'hémisphère gauche-droit. Des différences significatives ont été trouvées entre les groupes, suggérant des formes contrastées de dysfonction hémisphèrique.

ZUSAMMENFASSUNG
Hcmisphärenfunktion bei entwicklungsbedingten Sprachsörungen und hochgradigem AutismusZwei Gruppen von Kindern mit unterschiedlichen entwicklungsbedingten Sprachstörungen (phonologisch-syntaktisch und semantisch-pragmatisch) wurden mit einer Gruppe von Kindern mit hochgradigem Autismus und mit einer Kontrollgruppe gesunder Kinder anhand emer Reihe von neuropsychologischen Tests verglichen, von denen man weiß. daß sie rechts-links Hemisphärenstörungen erfassen.

RESUMEN
Función hemisférica en las alteraciones del desarrollo del lenguaje y autismo de alto nivelDos grupos de niños con tipos contrastados de alteración del desarrollo del lenguaje (fonolódgico-sintáctico y semántico-pragmático) fueron comparados con un grupo de niños con autismo de alto nivel y con un grupo control de niños normales utilizando una amplia bateria de pruebas neur-psicológicas, sensibles a la alteración hemisférica izquierda-derecha. Las diferencias significativas encontradas entre los dos grupos sugieren formas contrastadas de disfunción hemisférica.