This set of recently published research findings is loosely linked by the theme of communication.

The first section discusses the weakness among children with autism, compared to typically developing children, in respect of taking advantage of both auditory and visual information in speech perception ... ie  synchronising the speech sounds and the movement of the lips.

The next section concerns the influence upon the development of theory of mind capacities in children of parental references to, and explanations of, mental states during the reading and discussion of stories.

The final section further explores the nature of hyperlexia in autism compared with reading performance in typically developing children, and in children with ASD but without hyperlexia.


M.J.Connor                                                                                              October 2007


Audiovisual Speech Integration in Autism


The study by Smith and Bennetto (2007) is introduced by a reference to impaired communication as a core feature of autism.  The level of impairment may range from a severe delay in language development to a functional weakness despite apparently adequate vocabulary and expression. 


They note that rapid and fluent perception of speech is an area of particular difficulty for many of the children; and the question is raised whether one significant feature of the difficulty is that of failing to integrate information from the speaker’s voice, lip movements, facial expression, and body language .... (a failure to gain the global picture, one might say).   This multi-modal information aids the recognition and comprehension of what is being said, but the children with autism will commonly be at some disadvantage if they do not access all the cues that are available.


A description is provided of the way in which audiovisual speech perception has primarily been investigated ....  ie  via the “McGurk Effect”. 

The procedure involves the use of single syllable or multi-syllable non-words which are presented either visually (the participant sees the speaker’s lips but hears no sound), or auditorily (the sounds are heard but there is no visual information), or audio-visually (the participant hears the sounds and sees the speaker’s lips). 

When there is a deliberate mismatch between auditory and visual stimuli, the perception may be that of a kind of fusion between the two modes so that, for example, if the sound represents the syllable “ba” but the lip movement represents the syllable “ga”, the listener may perceive the syllable as “da”). 


Meanwhile, it is noted that audiovisual speech perception guides attention and is present in very young children as an influence upon speech production; and that audio-visual speech continues to provide an aid to comprehension among older children and adults in day to day social situations. 


A review of research findings by these current authors produces converging evidence that individuals with autism cannot readily integrate auditory and visual modes of information, not just in terms of these joint cues for understanding speech, but in such skills as matching voices to faces or in forming associations between sound beeps and light flashes. 

However, it has been hypothesised that a deficit in either auditory or visual perception alone could account for the difficulty in cross-modal integration typical of autism.


A practical problem has been the lack of direct exploration of audiovisual integration of speech among participants with autism , although what evidence does exist (eg. Williams et al 2004) suggests that children and adolescents with autism produce fewer fusions during McGurk Effect trials, with the implication that they do not spontaneously take into account the non-matching visual syllable during speech perception.


The authors go on to recognise that most day to speech operates against some degree of background noise ... the bustle of a typical classroom, for example ... so that listeners are required to filter out the ambient noise. 

Further evidence indicates that individuals with autism are relatively weak when it comes to understanding speech against a background noise.  Typically developing individuals are also more able to use visual cues to reinforce the auditory information and, thus, to enhance perception of speech. 


The implication is that, if individuals with autism experience problems both in audiovisual speech perception and in accessing speech against background noise, the combination will produce a cumulative and negative impact upon comprehension in typical settings.


The study completed by Smith and Bennetto themselves set out to determine whether young people with autism can use visual information to enhance an auditory communication presented against background noise; and it was their prediction that these participants would experience difficulties in processing audiovisual speech in this circumstance but that the difficulties would not be explicable in terms of auditory or visual processing deficits alone. 


The participants comprised a sample of adolescents diagnosed with autism and a control sample of typically-developing adolescents matched on chronological age

(mean age 15-8) , gender, full scale IQ (mean IQ 108), and scores of receptive language in the CELF battery.

The stimuli comprised a series of short sentences, no more than 7 words long, containing 3 key words, such as “ the cat jumped over the fence ”, and correct responses were those in which all 3 words were reported.  The sentences were presented either auditorily, via lip-reading, or in audiovisual conditions, and against a background of noise made up of recordings of readings from children’s books but with the meaningful articulations filtered leaving simply a speech-like noise.  

The speakers, presented via large-screen monitors, were 5 women between the ages of 23 and 28 (given existing evidence that young women are most readily understood auditorily and easiest to lip read).


The results provided confirmatory evidence that there typically exists an audiovisual integration impairment in autism. 

While the comprehension of both the target group and the control group improved when visual information was made available, the improvement was much stronger for the typically-developing group.

It was also noted that the adolescents with autism were significantly poorer on the lip-reading task; and regression analyses demonstrated that even after controlling for uni-sensory factors, the between-group differences in audiovisual speech were still evident. 


There were no group differences in responses to the auditory-only condition; but the finding of impaired lip-reading in autism is consistent with existing evidence, as is the finding of greater improvement among the control group when given additional visual information.  Given the large size/significance of this effect, the implication drawn was that this group difference is robust, and that individuals with autism may have a specific impairment in the ability to combine multi-sensory information. 

A further implication may be that of offering further insights into neurological functioning in autism. 

For example, studies of non-autistic adults have indicated that this “additive” audiovisual speech integration involves activation in the brain region defined as the superior temporal sulcus. 

However, in individuals with autism, this brain region is activated differently during a range of social cognition tasks such as perception of faces, or the attribution of mental states, or speech-sound detection. 


Meanwhile, evidence has it that typical perception of audiovisual speech is a matter of temporal synchrony, requiring the integrity of connections between different brain regions.  Imaging studies involving people with autism have shown a decreased connectivity between the superior temporal sulcus and other integrative centres ... and the problems with audiovisual speech integration could be interpreted as a reflection of this circumstance and the associated lack of synchrony.

An alternative hypothesis concerns the activation of mirror neurons, commonly seen as anomalous or under-functioning in autism, which respond not only during self-initiated actions but also, importantly, when observing actions performed by another person.


The current findings, in any event, have implications for daily functioning in that children with autism are likely, compared to peers, to perceive less and to understand less of what is said by their teachers even if attention and verbal conceptual skills are adequate.  Communication and understanding are not to be taken for granted; and  one practical issue is that of the possible benefits of direct instruction to focus attention on the speaker’s lips.

In more general terms, if latent deficits in audiovisual speech underlie atypical or delayed language development, there are implications for the possible benefits of early intervention by which to enhance speech integration among very young children identified with autism and ASD. 


Enhancing Theory of Mind by the Style of “Narrative” Input  


Slaughter et al (2007) describe how parents or other speakers can introduce children to the matter of internal mental states, and how they can be recognised, through the content of conversation and the reading/discussion of story books.

Their review of evidence indicates that those children whose parents often talk about mental and emotional states, during shared play or shared reading or shared examination of pictures, will have a relatively advanced theory of mind ... ie  a capacity to appreciate others’ feelings and perspectives, as well as a relatively early recognition of false beliefs (by which theory of mind capacities can be assessed).


These authors go on to summarise evidence by which to highlight at least two factors which come into play when parents talk specifically about cognitive states and which are linked to their children’s theory of mind development.

Firstly, children need to gain the language by which to formulate and express their beliefs, and the parental talk about mental states and perceptions provides the vocabulary and the syntactic structure for this necessary linguistic underpinning. 

Secondly, it appears that a simple reference to mental states is not sufficient to have an impact upon theory of mind; rather, the parents need to explain how the mental state is represented and how it appears to an observer


Their own study was in two parts.

Part 1 set out to evaluate preschool children’s false belief understanding in relation to the experience of maternal use of clarifying or explaining talk about cognitions and feelings rather than simple references to those types of mental states.  The predictions were that mothers would vary one to another in their use of this kind of clarifying talk, and that preschoolers whose mothers used this talk most frequently would gain the highest scores on false belief tests of theory of mind. 

Part 2 explored this mental state input on the part of the mothers across two groups of mother-child pairs ... those involving typically-developing children and those involving children diagnosed with ASD, in contrast to typical and existing studies which have not involved any participants representing a clinical group.   It was predicted that mothers of children with ASD would be less likely to include mentalistic language in their input.


(It was recognised that reciprocal influences could be operating,  ie  do the children respond to the maternal input, or does the maternal input vary according to the perceived responsiveness of the child ?)


30 preschool children and their mothers took part in the first part of the study; 17 boys and 13 girls ranging in age from 3-2 to 4-9.

The false belief task concerned variations on the familiar scenario of a child moving a toy from its original place to a different place while a second child was out of the room, raising the question of where that second child would be expected to look for the toy on returning to the room. 

Maternal input was analysed for quantity of words used, and for the number of component phrases or sentences that included a mentalistic content ... ie  a reference to the psychological state of one of the characters in a story being read. This input was also coded according to whether there were examples of clarifying ideas, such as offering suggestions about what was in a character’s mind in choosing to do something, or in giving some explanation of why someone should be feeling happy or sad, or in making specific references to discrepancies in the stories between what had really happened and what the absent person thought had happened.  


The results here confirmed that mothers of typical preschool children do vary in the frequency with which they mention mental states while reading or narrating stories, and in their offering causal and clarifying statements.

Further, these variations in the maternal language are linked to individual differences in the children’s performance in theory of mind tests, notably the cognition clarifications and the event-belief discrepancies.  It was concluded, therefore, that it is the maternal tendency to expand on the characters’ thoughts rather than a simple use of “cognitive” words which is associated with enhanced theory of mind. 


In part 2, a different sample of mothers and their children took part in a longer narrative task involving sharing wordless storybooks, and the maternal talk was analysed for references not only to cognitions and feelings in the characters but also for mentions of mental states not directly concerned with the story ... usually references to their own actions and feelings or those of the children.

The sample of children included 24 diagnosed with ASD plus 24 typically developing children matched for language development and verbal IQ. 

The 16 boys and 8 girls in the ASD group ranged in age from 4-3 to 9-3;  while the 14 boys and 10 girls in the other group ranged in age from 3-1 to 6-9.


The results indicated several similarities between the mothers of the two groups in their use of various mental state terms when working on wordless story books with their children. 

Quantity of utterances did not differ, nor did the number of simple mentions of the mental states of characters in the stories.

However, in respect of the content of the narratives constructed, the mothers of children with ASD were less likely to produce clarifying comments about the characters’ cognitive or emotional states.  Further, the children in the ASD group demonstrated that their performance in both perspective taking and false belief tasks was significantly linked to the maternal use of clarifying comments. 


The overall discussion on the part of the authors included a restatement of the finding that, for children with ASD or typically developing children, the most advanced understanding of theory of mind is found among those children whose mothers choose frequently to talk about mental states in a manner designed to clarify what the feeling is and why (not mere mentions of mentalistic themes). 


Among the ASD group, it was the maternal clarifications of feelings that carried the greatest association with test sores concerned with theory of mind and perspective taking.  It was speculated that these children, with their characteristic weaknesses in linguistic and social skills, find affect easier to understand than the content of thinking, possibly because of the additional cues that are available from facial or vocal expressions.


The conclusion reiterated the possibility of bi-directional influences .... mothers may tailor their input according to cues about the child’s interest in and understanding of mentalizing; and the children’s understanding may develop, at least to some extent, as a result of hearing the maternal clarifying talk. 

Further studies are necessary to clarify these issues, but the possible implication to be drawn, one would presume, is for ensuring a high level of exposure for the children with ASD to a high level of adult talk taking a variety of forms, including the elements of clarifying and explaining.  Without this, the child’s existing affective and communicative weaknesses may not receive the compensatory stimulation that could bring about an enhancement, and a Matthew Effect may be the outcome.  


Hyperlexia and ASD                        


Hyperlexia (HPL) refers to a contrast between very high reading accuracy and modest or even poor comprehension. 
Following the recognition of this condition and a description of case studies, it was noted that HPL appeared frequently to be linked with behaviours consistent with autism spectrum disorder.


Over time, HPL has become characterised by superior word reading ability above the level of reading comprehension, and greater than predicted by the general verbal or cognitive functioning of the individuals concerned.

Further, many of the studies reviewed in the literature by Newman et al (2007) continue to recognise the early onset of this imbalance, a compulsive interest in reading at the expense of other means of communication, and the high degree of comorbidity between HPL and ASD.  


This is not to say that HPL is a condition which impacts only upon children with ASD and there remains uncertainty whether HPL is generalisable to individuals with a range of cognitive profiles and levels.  On the other hand, there remains the possibility that a greater understanding of HPL might provide some insights into the nature and course of ASD, with some speculation whether the very high word reading skills may prove a marker for a more positive prognosis among individuals with ASD in terms of measured cognitive ability, expressive language, and social relatedness.


It has been argued (Frith and Snowling 1983) that the semantic weaknesses associated with HPL are similar in nature to the deficits observed in the oral language of children with ASD.  However, the interest and skills in reading among children with HPL raise the possibility of using these characteristics as a means of strengthening the meaningfulness of both oral and written language.


Another line of research with children diagnosed with high functioning autism or ASD has noted a pattern of stronger word reading skills in contrast to deficits in language comprehension and abstract reasoning, leading to the question whether HPL is associated with a distinct subtype of ASD among individuals across a wide range of ability levels or only among those whose measured ability falls within a narrower range (good average and above ?).


In any event, Newman et al recognise that a study of HPL could increase the knowledge about reading and reading difficulties in general.

In typically-developing children, word reading and reading comprehension are closely correlated, but HPL involves a dissociation between these two elements ... even to the point where increasing the difficulty of the text or altering the word order do not bring about a reduction in the reading fluency among children with HPL ... and a greater insight into this disconnection would inform existing models of reading, including the relatively new field of comprehension difficulty.

It is their view that children who demonstrate both HPL and ASD could act as unique participants in studies concerned with comprehension weaknesses which appear independent of word reading skills. 


A practical problem is the low incidence of HPL with researchers seeking to examine an isolated skill occurring in a small percentage (no more than 10%) of children with ASD. 

Existing evidence has also shown that the high word-reading skill which can be observed among young children tends to even out by around the age of 10 years so that children identified with HPL ultimately do not read any more, or better, than their peers with ASD. 

In other words, the research has involved a limited sub-population of children, and findings typically relate to single case studies or to very small sample sizes.


A fundamental question is whether word reading among children with HPL develops in the same way as “normal” reading albeit taken to a high level by the consistent practice, or whether the children pursue some different sequence of component skills to achieve their word reading fluency.


Is hyperlexic reading largely a matter of visual skills and a matter of pattern recognition (as illustrated by the study of Cobrinik [1982] in which boys with HPL showed greater speed and accuracy than controls in deciphering incomplete words) ?

Are hyperlexic reading and normal reading both reliant on phonemic processing to identify single words ... as suggested by evidence that readers with HPL make fewer errors with regular words than with irregular words; and that they are able to complete phonological awareness tasks, such as phoneme-deletion exercises, as well or more accurately, than typically-developing children. 


The study by Newman et al set out to investigate differences in performance on reading tasks among children with ASD and HPL, and two control groups.  The first control group included age and gender matched ASD children without HPL; and the second control group included typically developing children matched by gender and word reading level. 

The specific aims were to confirm that HPL is characterised by strong single word reading relative to comprehension; to confirm that strong decoding and weaker visual memory skills contribute to the differential performance on word recognition tasks across the 3 groups (with the prediction that the ASD +HPL group would show decoding skills as good as those of the typical children and better than those of the ASD group without HPL); and to investigate for differences between the target group and the control groups in respect of component skills of decoding such as rapid naming, phonological awareness etc,  and of vocabulary knowledge which is correlated with good comprehension. 

The groups comprised 20, 20, and 18 individuals respectively (mostly boys), with a mean age of 10-4, 12-3, and 9-9.


The results of the study confirmed the findings of previous research in demonstrating that hyperlexic reading is characterised by a significant discrepancy between single word reading and reading comprehension. 


The single word reading skills of the ASD+HPL group appeared to be similar to those of the typically-developing children, and greater than those of the children in the ASD without HPL group.

However, further analysis revealed the effects of age differences.  The youngest hyperlexic children in the sample were aged around 5 years, hardly expected to achieve much of a score on a word reading test, and whose reading, therefore, appeared precocious in comparison with the norm.

When this effect was controlled, the HPL+ASD group were shown to be performing at a level similar to children in the ASD but not HPL group, and more poorly than the typically-developing children. 


The test of non-word reading indicated that this was a strength among children in the ASD+HPL group and suggested that children with HPL rely on phonological decoding, like typical readers, rather than on visual and whole-word memory. 

This consistent finding was in contrast to the variable results in tests of non-word reading reported in existing research studies.


The tests of visual memory and of rapid naming produced no differences in scores among the three groups.


One unexpected result was that, in respect of reading fluency, the ASD+HPL group performed similarly to the ASD control group and more poorly than the typically reading group ... when existing findings had suggested that children with HPL read significantly more quickly than normal readers. 

The authors sought to explain this discrepancy in terms of the nature of the current reading task compared to what had gone before ... ie the current test of reading fluency was one that required a degree of comprehension (with the child required to read a sentence and circle yes or no according to the perceived truth of the sentence).

In other words, when required to examine the content of the text, HPL readers are significantly slower in their reading. 


In their concluding comments, Newman et al emphasised the unusual pattern of early

acquisition of (hyperlexic) reading accuracy, but confirmed that exceptional skills in word recognition observable at a young age appear to even off by the age of 10 years to an average level, with other children able to catch up.

Even so, ASD+HPL children older than 10 years have better single word reading and word attack skills than those shown by their ASD peers without a history of HPL.

They suggested that this progression may reflect the general language performance of children with ASD in respect of better “mechanical” skills than abstract and reasoning skills ... compounded by an uncertainty among professionals about how to capitalise upon the early word reading strength. 


The authors regard word recognition as an isolated skill that is preserved in individuals with HPL, perhaps developed at a young age as a result of deliberate, even obsessive, practice.  However, apart from the trend towards intensive practice, these children appear to be using the same processes (phonological and orthographic mapping) as typically developing children when seeking to recognise words.

The lack of a positive correlation between word reading and text comprehension, which is observed among both typical readers and dyslexic readers, is seen as consistent with the generally poorer oral language skills of children with ASD, especially in the more abstract elements such as pragmatics and comprehension. 



                                    *          *          *          *          *          *



M.J.Connor                                                                                              October 2007                                                      







Cobrinik L.  1982   The performance of hyperlexic children on an incomplete words task.   Neuropsychology  20  569-577


Frith U. and Snowling M.  1983   Reading for meaning and reading for sound in autistic and dyslexic children.   British Journal of Developmental Psychology  1  329-342  


Newman T., Macomber D., Naples A., Babitz T., Volkmar F., and Grigorenko E.  2007   Hyperlexia in children with autism spectrum disorders.   Journal of Autism and Developmental Disorders  37  760-774


Slaughter V., Peterson C., and Mackintosh E.  2007   Mind what mother says; narrative input and theory of mind in typical children and those on the autistic spectrum.   Child Development  78(3)  839-858


Smith E. and Bennetto L.  2007   Audiovisual speech integration and lip-reading in autism.  Journal of Child Psychology and Psychiatry  48(8)  813-821


Williams J., Massaro D., Peel N., Bosseler A., and Suddendorf T.  2004   Visual-auditory integration during speech imitation in autism.   Research in Developmental Disabilities  25  559-575         

This article is reproduced by kind permission of the author.

© Mike Connor 2007.

Back to NAS Surrey Branch Welcome Page