WILSTAAR - the Ward Infant Language Screening Test, Assessment, Acceleration and Remediation - provides a scheme for identifying and treating infants aged eight to ten months who are showing early signs of language and cognitive difficulty. Its approach is common in community health services: a simple screening assessment, where professionals (usually health visitors) ask parents about delays and deficits their children are exhibiting, followed by confirmatory assessment and a remedial programme, usually delivered by speech therapists.

Sally Ward published evidence that children who failed the screen tended to have language and cognitive problems at pre-school age, while a simple programme, which enriched parents' interactions with such infants, normalised the children's development. These reports attracted a good deal of critical comment and scepticism. The main concern was whether these infants really had psychological difficulties or whether the apparent improvements attributed to the intervention programme were really due to a lack of difficulties in the first place, or to minor delays that the children spontaneously grew out of. Inaccurate screening of this sort is known as 'false positive' identification.

The second concern was whether the children's home environments were responsible for their psychological difficulties, as Ward supposed. Press reports highlighted her observation that television sets were constantly on during home visits but no systematic evidence was published. Our study's goal was to provide an independent evaluation of WILSTAAR's effectiveness and, in the process, to address the questions raised by its critics.

We asked 10 National Health Service Trusts using WILSTAAR to refer cases to us for study following the WILSTAAR screen, but before the intervention began. They were asked to refer infants who failed the screen ('at-risk' cases) and infants of the same age (within a month) and gender who passed the screen ('not-at-risk' cases). In total, five WILSTAAR schemes referred 80 at-risk and 72 not-at-risk infants to us. In each group, 21 infants were excluded for a variety of reasons. A further 18 mothers from each group refused to take part and 11 at-risk and three not-at-risk cases dropped out later, giving 30 infants per group for comparison.

Following the normal WILSTAAR procedure, each infant's difficulties (or lack of difficulties) were confirmed using the Bzoch & League Receptive-Expressive Emergent Language Scale (REEL). Like the WILSTAAR screen, the REEL measures a parent's (usually the mother's) report of her infant's listening and expressive language abilities, but is more comprehensive. As well as using it to confirm the group differences when infants entered our study, we repeated the REEL after the study was over, to establish whether the mothers changed their minds.

Each infant was assessed in two main ways. First, measures of the home environment were made by a trained researcher during two home visits on successive days, each lasting two hours. As well as demographic details, measures of both the social and physical features of the homes were made. Infant television exposure was observed during the visit and mothers were interviewed about exposure throughout the preceding day. A sound-level meter was used to sample the sound level in the vicinity of the infant. Mother-infant interaction was observed in the spontaneous situation, and video-recorded during a standard play task, which allowed the reliability and validity of the observation measures to be confirmed.

A week after the home visits, each infant's attention abilities were measured using a standard, validated task, the Distractibility Task. This measures an infant's ability to focus attention on toy play, and to resist distraction by a series of 'distractor' slides shown in the background. We were interested in whether the development of infants' attention skills at this age acts as an 'organiser' of their social-psychological and behavioural abilities, which drives their language skills forward. We expected that 'at-risk' infants would perform poorly on this task.

The findings were in keeping with this expectation and some, but not all, of Ward's original ideas. The not-at-risk infants were quicker in deducing that the distractor slides were of little importance; they also spent more sustained periods of time focused on toy play. The at-risk infants were slower and less effective in learning to use the information available in the task to manage their attention and interactions with the environment.

The at-risk infants did not have noisier home environments, or spend more time exposed to television. Rather than an excess of noisy stimulation, the chief shortcoming in the at-risk infants' environment was a dearth of social interaction. Mothers of at-risk infants spent significantly less time interacting with their children than mothers of not-at-risk infants. This lack of social interaction was reciprocated by the at-risk infants, who babbled substantially less than not-at-risk infants during observations of their spontaneous behaviour.

The design of this study, which assessed infant behaviour for the first time at 10 months, does not allow the origins of this dearth of social interaction to be understood. Since there is evidence that attention and language difficulties can have genetic origins, the lack of spontaneous babbling in the at-risk infants may well be the result of infant vulnerabilities, which make them unrewarding social partners, leading their parents to interact less with them. Equally, the at-risk infants' lack of babbling may reflect low levels of social stimulation from their caregivers. The demographic details collected did not give any indication of general social deprivation or risk in the families of at-risk infants. Their parents tended to be well-educated, Caucasian, middle or lower-middle class, and in stable relationships, as did the not-at-risk infants' parents. Both infant groups were looked after predominantly by their mothers and no group difference in amount of maternal employment was found.

Although the findings do not identify the origins of the at-risk infants' difficulties, they clearly point to deficiencies in social interaction and communication, involving both infant and parental contributions, as maintaining the infants' difficulties at this age. They are consistent with the idea that attention abilities, which emerge at this age, provide the building blocks for social-cognitive and language abilities, and that these building blocks are deficient in infants who fail the WILSTAAR screen. These findings contribute to our understanding of the process of infant development and provide a conceptual framework for further research. The idea that social-attention abilities drive language development forward is beginning to receive support in other studies.

Turning to the study's implications for WILSTAAR and health services, an important question is whether the infant difficulties identified are transitory, self-resolving delays, or the early signs of serious and protracted deficits. Although this study was not designed to answer this question conclusively, several findings indicate that the WILSTAAR screen is inaccurate in its current form. First, 25 per cent of at-risk infants referred to the study by WILSTAAR projects had to be omitted because of our exclusion criteria. Some of these criteria, such as an infant being prematurely born, need to be recognised within the screening process, since such infants' limitations may be due to immaturity and so resolve themselves when they reach the target maturational age. Other cases we excluded, such as those where English was not a parent's first language, pose special challenges, since this could provide both an explanation for an infant's delay and a justification for an intervention programme. In a multicultural society, the implication is that WILSTAAR policies need to provide for such cases in order to minimise false-positive identification, whereas these policies were not effective in the practice of the WILSTAAR projects included here.

A further finding was that some 30 per cent of mothers who identified their infants as having problems on the WILSTAAR screen considered that the infants did not have problems when re-questioned using the REEL interview a week later, or two weeks later after taking part in the Distraction Task. Although this finding can be interpreted in a number of ways, a discrepancy of this size clearly questions the reliability of the original maternal assessments.

Taken together, these findings support concerns about the applicability of WILSTAAR for routine clinical use. It appears likely that many 'at risk' cases selected by the WILSTAAR screen are false positives, and that this will produce an inflated view of the effectiveness of the WILSTAAR intervention programme. Less clear is whether these inaccuracies are resolvable, by including more extensive and accurate assessments within WILSTAAR, or are an inevitable consequence of assessing young children at a time when their development is in a state of flux. A potential advantage of early assessment is that it may lead to a more effective intervention. However, this has to be balanced against the likelihood that both false positive and false negative identification rates are higher the younger an assessment is made. It is possible that the optimum trade-off between assessment age and accuracy will involve delaying assessment until after age eight to ten months, currently targeted by WILSTAAR.

Finally, although these service implications require further evidence, it is worth noting that the practical application of the present findings needs not await this work. As NLT-supported initiatives such as Talk to Your Baby have recognised, the core of the WILSTAAR intervention - encouraging social communication between parents and infants - is not complex and does not need to be confined to at-risk cases. How this is best accomplished, and how it should be included in national schemes such as Sure Start in the UK, are important topics for policy makers.

Subscribe to Literacy Today