These summaries focus on two issues.

Firstly, there is reference to the nature and pattern of the commonly-observed regression in skills among a significant minority of children with ASD around the middle of the second year.

Secondly, reference is made to the importance of play skills for ongoing (social) development, with an example of a particular strategy found to be effective for fostering skills in pre-school children with ASD.


M.J.Connor                                                                                                August 2004



Regression in Skills among Children with Autism


There has, over many years, existed the phenomenon, reported by parents and confirmed by clinicians, whereby a significant minority of children later or already diagnosed with autism appear to acquire language and other skills in a largely normal pattern, but who lose such skills during the latter part of their second year.


This phenomenon was recognised by Kanner (1943) and reinforced by the observations of Rutter and Lockyer (1967) …. and it would usefully be noted that this apparent regression has been evident long before the introduction of general vaccination (MMR).


The phenomenon is discussed by Lord et al (2004) who point out that the age of the child when autism is recognised and diagnosed may be quite disparate from the actual age of onset.

Accordingly, clinicians have made use of video records of a child’s development, together with more searching interviews with the parents, and these have served to highlight behavioural anomalies at increasingly early ages, often well before the parents had suspected anything unusual in their children, by which to differentiate children subsequently diagnosed with autism from normally-developing children or children with some developmental delay.


The resulting data from the videotapes and the more specific exploring of parental experiences have demonstrated that some children with autism or ASD do truly lose skills or cease using certain behaviours once used consistently.


The authors are careful to pint out that home videotapes do not usually involve repeated filming in similar contexts, or access to the full range of child behaviours, so that there is still a significant dependence upon parental reports of the skills which children developed and then lost before the formal diagnosis of autism or ASD.


Their own study has concentrated upon the loss of words during the first and second year of life, with Lord et al also careful to acknowledge that the occurrence of this kind of loss does not imply that development was completely normal until that point.

They refer to the work of Rutter and Lockyer (op.cit) who noted that almost half of their sample of 63 children had been described as having had a period of apparently normal development, but they suggested that there probably had been abnormalities and anomalies, albeit not recognised as such by the parents.


A further question concerns the similarities or differences between the skill losses occurring before the age of two years and those occurring after this age. 

These later losses, following a period of normal development across a range of skill areas such as language, social, and motor, have been classified as part of Childhood Integrative Disorder. 

The authors note that it is not known whether the symptoms of this disorder are on a continuum with the earlier losses; although this disintegrative disorder does differ in some ways from autism and ASD in that the prevalence is much lower, the regression occurs later, the outcome is poorer, and in many cases there is a higher association with known neurological abnormalities.


The authors go on to cite existing research findings on the issue of early loss of words, nearly all of which have depended upon retrospective reports of children who were already in school at the time of the information-gathering.


The review of studies involving parents of school age children suggests that the proportion of children later diagnosed with ASD and who experienced  a loss of words before 30 months ranged from 15- 37%, with the majority of evidence suggesting  a level around 20-25%.

When parents were interviewed while their children were younger, either pre-school or just entering formal schooling, retrospective estimates of the age when the losses occurred ranged from means of 17-21 months.

The loss in language was commonly accompanied by a loss in other areas, notably social interest or other communicative skills.


Evidence from paediatricians shows a relatively high rate of seizures and abnormal EEGs in children with word losses, with such symptoms generally more common in children with later regression than in children with losses before 30 months.


In their own study, Lord et al gained information and opinion from parents of children below three years of age referred for possible autism, and from parents of children referred for developmental delay.  Control data from parents of normally-developing children were obtained.

The focus was upon the timing and the nature of word losses in the first two years, with the expectation that the information would be more accurate than that obtained from parents of much older children.


The results confirmed the trend of previous research in that about 25% of children with ASD were described by parents as having used words meaningfully but as losing this skill during the second year of life.

The authors noted from their own evidence and from previous research that the pattern of loss of a small number of meaningful words is not a pattern observed in any other group of children with developmental delay, such as children with Down Syndrome, Williams Syndrome, or Fragile X Syndrome.

There may be some comparability with that rare group of children with acquired epileptic aphasia, but it was held that this pattern of word loss is unique to a percentage of children with ASD.


The significance of this is that a sustained loss of previously known words can be readily identified by a clinician, and the symptom may prove an important marker for a sizeable minority of children with ASD.


Almost all the children concerned were also described as losing social skills, although it was not clear whether the two types of loss occurred more or less at the same time or if one was noted before the other.  Further, parental reports were equally divided in terms of whether the losses were gradual or sudden.

What was more consistent was the description of the onset of a few spoken words at about the normal time (by 12 months), with a subsequent failure to progress beyond a small number of spoken words, and a loss of what words did exist by around 16-17 months.  There was none of the rapid onset of new words which is common in most children at this time.


The further finding replicated existing evidence in that, among those children who regained speech by the age of 5 years, there was a relatively short time between the loss of words and their return … around 5 months. Thus, the age by which language was fairly stable was much the same for those children with ASD who had experienced the word loss as for those who had not.


The authors noted that the ages of word loss were generally lower than those reported elsewhere.  This may be a function of inaccuracy in retrospective data; but the authors also noted from their own study that the parents of children with a word loss tended to have become concerned at a significantly later age of the child – around 18 months – compared to parents of children who showed no word loss – around 12 months.  This was not something reported in previous research.


Again, the authors speculated that this was a matter of interviewing parents much earlier, and perhaps of some particular characteristic of these children or their parents and linked with their earlier quest for advice from specialist clinics.

Further, it appeared that parents of higher functioning children who rapidly regained language after a word loss tended not to report this information at the time, unlike parents of children with a longer duration of word loss or who never regained speech.


In any event, the conclusion highlighted two issues.


 It was further speculated that this pattern is a matter of anomalous neurological development or cognitive functioning as a result of absent social motivation which impacts negatively upon linguistic or more general cognitive processes related to word learning.


Remaining questions concern what it is about those children with ASD who are able to use early words meaningfully but whose progress is not maintained and these initial skills lost, or about those children who do not show this pattern. 


The issue of language regression and other regression, and its timing, is pursued also by Goldberg et al (2003) who have a particular interest in the matter of gaining accuracy of information about this phenomenon.


They describe how the diagnosis of autism and related disorders has been aided by the development of the Autism Diagnostic Interview –Revised (Le Couteur et al 1989), which can permit the identification of regression in autism during the child’s first five years by providing a clear structure to compensate for the inevitably retrospective nature of parental accounts.

From a review of studies, they note how regression has been reported in previously acquired language and other social or communicative skills, accompanied by the onset of restrictive and repetitive behaviours.  They include a reference to the pattern observed in some cases of the gaining of some spoken words, followed by a plateau period, and then a loss of those words.

It is further noted that the regression may be sudden or gradual. 


The existing evidence also indicates that the mean age for autistic regression falls between 18 and 24 months, but has been detected as early as 15 months.

Estimates of autistic regression per se (ie  not a matter of disintegrative disorder or some other condition) range from 10-50% ; and among those showing regression, loss of all spoken language has been reported in up to 40%.


Goldberg et al also refer to the possible confusion about the actual timing of these events given the confounding of age of onset with age of recognition; and it may be that some children may not have been recognised as showing regression until their second or even third year when, actually, the symptoms may have existed earlier.


It is also held that language delays or setbacks are more likely than social anomalies to lead parents to seek advice and support, possibly because language skills have  more clearly defined set of milestones, and difficulties are more easily detected.

It has been further reported (Siegel 1996) that language problems have coexisted with an absence of imaginary play, and it is this combination which may aid the recognition and diagnosis of autistic spectrum disorders.

In respect of the nature of regression, word loss is most readily noticed, but a significant number of parents have noted losses in non-verbal behaviours and social performance.


The study of Goldberg et al themselves set out to highlight the value of a further measure – the Regression Supplement Form – that allows for the more detailed description of regression, and is complementary to the Autism Diagnosis Interview in gaining data about the nature, timing, and areas of loss in the case of regression, along with the identification of concurrent events.


The participant families numbered 44, drawn from all those participating in a university-based study, and identified through having highlighted a loss of language or other skills in their responses to the diagnostic interview.


The supplementary form focuses upon 18 areas which may have involved a loss of skill, including word expression, direct gaze, orienting to name, social smiling, communicative gestures, etc..  The children concerned were classified as losing spoken words, losing non-language skills, or losing skills in both spoken language and non-language areas.


The results of this study produced a rate of regression in this sample of 38%, largely comparable with other studies.  More than half the children (62%) lost words, but nearly always alongside losses in non-language areas.  The middle of the second year was the typical timing of the regression, with a gradual rather than sudden onset for the non-language losses, but with language losses evenly divided between gradual and sudden onsets.

The parents tended to link the regression to medical factors, such as immunizations.


Many of the children regained the skills in question between two and four years later, with the parents typically attributing this gain to some specific course of intervention

(with the possible implication of a confusion between a temporal association between the regaining of skill and the intervention, and a causal or truly explanatory association).


Regression was most commonly reported in word use, direct gaze, orienting to name, spontaneous imitation, and response to social overtures.  Evidence indicated that loss of language alone was much less common than either a loss of non-language skills or a combination of spoken words plus non-language skills, even if the loss of words is more readily noticed.


The authors referred to problems with a reliance upon parental reports, although linking those reports with home videotapes increased the reliability of data; and, even if there was a time gap between the actual loss of skill and the point at which parents became aware of developmental anomalies, it was confirmed that the average stage for the regression was around the middle of the second year.


Responses to the supplementary form also indicated that skills reflecting higher levels of cognitive, social, or linguistic functioning tended to be lost only rarely; with the majority of the children who lost non-language skills having no spoken words in the first place.

The authors commented that the frequent reference by parents to the link between regression and immunization was not unexpected given the publicity surrounding the claims set out in the Wakefield et al (1998) study which was still current during the course of this present study.


Goldberg et al acknowledge the relatively small sample size as a limitation of their research, as well as incomplete data from some respondents which made for some difficulty in placing children in the appropriate group in terms of the nature of regression observed.

Nevertheless, they hold that this Regression Supplement Form offers both clinicians and researchers a useful tool for gaining more specific and more reliable data by which to examine the nature and timing of regression which appears to be a relatively common phenomenon among children identified with ASD.


Play Skills


This section offers a confirmation of the significance of (social) play for children with autism, both in terms of identifying the disorder in the first place or differentiating it from other pervasive disorders, and of a focus for intervention. 

One particular approach to develop social interaction through enhancing play skills is set out.


Firstly, one can refer to the work of Jordan (2003) who restates how a lack of varied or imaginative play is a key feature of autism.

She recognises that it may be difficult to provide a specific definition of play, and that the range of behaviours that could be regarded as a form of play are many and various.  

The implication is that a child with ASD might be perceived as not able to play when, actually, the child in question and the adult observer may differ in their view about what play entails.  The observer may only recognise play when it is of a conventional form, and the kind of activities modelled or taught may not be diverting or truly playful for that given child with ASD.


Nevertheless, there is a long tradition whereby play is seen as important because it enables children to learn and practise new skills or roles in a secure setting, and to establish interaction with other children.


Similarly, Jordan continues, play reflects the development of cognitive, social, and linguistic skills, as illustrated, for example, by the link between cognitive growth and the move of play through the stages of sensory exploration, repetitive play, constructive play, and purposeful and problem solving play.

Meanwhile, play may start as a series of individual activities, with the child learning about cause and effect, but gradually will involve an observation of the activities of other children, and a mutual responsiveness, leading to shared play or even competitive play.


Play is the medium through which initial interactions with care-givers are established, and where are laid the foundations for the motivation for contact with other people, as well as learning about others’ reactions. 

It is the social and mutually rewarding play that forms the basis for friendships and relationships involving cooperation and trust, and giving and taking.


The question is raised by Wolfberg (1999) whether the child needs to learn how to play (in a conventional and potentially interactive way) in order to socialize, or to learn to socialize in order to play.


Certainly, there are limitations in the development of pretend play, but a stimulus for intervention is the finding that the child with ASD can engage in such play if there is a structure established and cues provided.


Jordan’s review of evidence leads to her suggestion that it is the social aspects of the play that are affected.  The child may be capable of symbolic play, but it remains simple and repetitive in the absence of joint play with others which would lead to a greater variety or flexibility of play themes.

This is important given the probability that a failure to share pretend play with other children is linked with later social isolation and the failure to develop social skills.


Jordan’s final point concerns how best to facilitate play development, and a further review of studies suggests that caregivers may be too directive and too anxious to elicit observable play.  

The child with ASD may be more responsive to the play and play overtures of other children, including siblings, who are less directly demanding upon the child or less intrusive in their approaches.


(Relevant to this theme is the finding from a recent study - Wong et al [2004] – which highlighted the validity of the Checklist for Autism in Toddlers [CHAT] in identifying autism.  In particular, it was found that key areas were joint attention, pretend play, social relatedness, and social referencing.  Significant questions concerned whether the child can imitate the parent, pretend to talk on the telephone or take part in any other pretend game like caring for dolls, point to things of interest, check the adult’s reaction, share attention, etc..  In other words, the critical issues appear particularly to include pretend play and social communicatory behaviours.)


One example of a strategy to enhance play and interaction is described by Thomas and Smith (2004) whose initial concerns reflected much of the foregoing, viz, that without focused play, the child with ASD may remain socially isolated and not be included within the play and interactions of peers.


The intervention is being used in a number of the early years settings in the LEA in question, and combines a play script with tabletop mirroring using two sets of identical toys (“Identiplay”).


The approach is based on existing evidence (see, for example, Beyer and Gammeltoft 2000) for the effectiveness of specific teaching of play sequences through a structured series of steps.

The three targeted pre-school children, all identified with ASD, were involved in an intervention over a two-week period, with each child receiving five minutes of daily input working on specific table top play.  Carer questionnaires were completed before and after the intervention, and videotaping of the child’s play and interaction was maintained, in order to assess change over the course of the two weeks.


The measures used included “duration” – the amount of time spent engaged in specific types of play; “repetitiveness” – the number of occurrences of each play activity; and “diversity” – the number of changes between different play activities.

The carers’ questionnaire focused upon play behaviours, including social elements, and examined the extent of repetitive play, the extent to which the child watched other children, and reactions to the presence of other children.


The equipment included a table divided in two offering a clear area for the child’s toys and the adult’s toys and two sets of the same toys.  The adult begins by setting out the toys in both halves of the table, and watches the child’s reactions.  The adult then follows the child’s lead and joins the child’s play, imitating his actions and sounds, adding to the play and teaching/modelling actions, but returning to mirroring the play sequence of the child if the child did not respond or appeared resistant.


The findings showed that the children did increase the amount of time on appropriate play with the toys, and played more (and more purposefully) in free play times, being able to include at least part of the taught play sequence. Further, there was more time spent in playing alongside other children in parallel play or in playing with the other children.  Social interaction, as reflected in eye contact and verbal communication, was also seen to increase.


The authors held that the outcomes confirmed the importance of direct teaching for the enhancing of play and interaction, with the identiplay setting providing a good balance of teaching and mirroring, and giving the child the structure and confidence by which to increase their range of actions.


The social changes are considered to reflect the improvements in functional play which encourages peers to join the play of the targeted children.  Further, the increase in play skills put the targeted children more on a par with normally-developing peers, providing the confidence for greater interaction in the way of eye contact and verbal communication while also encouraging the other children to approach the children with ASD and play alongside them.

The reduction in the repetitive sensorimotor and self-stimulating behaviours would also make the targeted children more attractive play companions, with the implication that the improved behaviours and play skills are maintained.


The authors acknowledge that the play input within their study was only a small part of the whole experience of the children in question, with other opportunities and stimuli for play and social activity. 

Nevertheless, the observed nature of the changes were said to reflect closely the kind of behaviours modelled in the table-top play that it seemed reasonable to claim a positive benefit from the specific input in the identiplay intervention.


Accordingly, it was concluded that play scripts with the mirroring style of interaction do provide a means of developing play and allied skills in young children with ASD, making them more approachable in the eyes of peers and fostering social interaction and development.



                         *          *          *          *          *          *          *         *



M.J.Connor                                                                                                August 2004






Beyer J. and Gammeltoft L.  2000   Autism and Play.   London : Kingsley Publishing


Goldberg W., Osann K., Filipek P., et al   2003   Language and other regression.   Journal of Autism and Developmental Disorders  33(6)  607-616


Jordan R.  2003   Social play and autistic spectrum disorder.  Autism  7(4)  347-360


Kanner L.  1943   Autistic disturbances of affective contact.   Nervous Child  2  217-250


LeCouteur A., Rutter M., Lord C., et al  1989   Autism Diagnostic Interview.   Journal of Autism and Developmental Disorders  19  363-387


Lord C., Shulman C., and Dilavore P.  2004   Regression and word loss in autistic spectrum disorders.   Journal of Child Psychology and Psychiatry  45(5)  936-955


Rutter M. and Lockyer  L.  1967   A five to fifteen year follow-up study of infantile psychosis.   British Journal of Psychiatry  113  1169-1182


Siegel B.  1996   The World of the Autistic Child.   New York : Oxford University Press


Thomas N. and Smith C.  2004   Developing play skills in children with autistic spectrum disorders.   Good Autism Practice  5(1) 53-57


Wolfberg P.  1999   Play and Imagination in Children with Autism.   New York : Teachers’ College Press


Wong al  2004.  Findings on the use of a screening test for autism reported in Medscape Medical News – 5 August 2004.  Original article in Pediatrics 114 166-176


This article is reproduced by kind permission of the author.

© Mike Connor 2004.

Back to NAS Surrey Branch Welcome Page