Dyslexia:at school: a review of research for the DfES

Recent research and development[1] in dyslexia in relation to children of school age: a quarterly review for the Department for Education and Skills, the British Dyslexia Association and the Dyslexia Institute. Review 1, September 2001

Angela Fawcett, Department of Psychology, University of Sheffield

Disclaimer: The material presented here reflects the views of the reviewer, not necessarily those of the University of Sheffield, the DfES, the BDA and the DI.

In this first review of the series, I will concentrate on recent developments in the theoretical basis of dyslexia. My brief is to concentrate on school aged children.

•explain the theories clearly in language appropriate to a lay audience,

•explain the links between theories

•outline their implications for policy and practice

•consider areas for further research.

Recent research and research currently in progress

I shall introduce these theories in the order in which they have emerged. I shall present the background, followed by examples of interesting recent research within each framework. There have been striking developments in dyslexia in the last ten years, with each theory filling in part of the jigsaw for the complex problem of dyslexia, which can extend far beyond literacy skills and impact on an individual throughout their life. The UK is at the forefront in this theoretical research. In my view there is a synergy between these theoretical developments leading to a more satisfactory explanation of the symptoms of dyslexia than the individual theories on their own. I shall return to the theories in the section dealing with links and emerging themes, where I suggest that the most productive way forward is to work together to design studies where each research group tests out all the theories within the population of children they work with, and within school-based samples.

i)Background - Phonological deficit.

This is the most well developed and supported of the theories of dyslexia. It has been widely researched, both in the UK (York group) and in the US. The US researchers have united in adopting the phonological deficit hypothesis since the early 1980’s, and this united front has led to the investment of more than $15 million annually by the US government, via the National Institute for Child Health and Human Development (NICHD).

There is unanimous agreement that problems with phonology are associated with dyslexia, however, it is becoming clear that phonology is not the only problem. Phonology is a skill underlying the analysis of both spoken and written language; breaking down words into their parts, or segmenting them, so first knowing that ‘cat’ is made up of the onset and rime c-at, and then recognising the individual sounds (phonemes) are c-a-t. Phonological awareness is also used in hearing a sound (a phoneme) and translating it into a letter which represents it (a grapheme). These skills need to develop around the age of 5, if young children are to learn to read successfully. Otherwise, they are limited to reading words they recognise as a whole (orthography) and are limited in their ability to learn new words. There is solid evidence dating from the work of Bradley and Bryant, 1983, that rhyming is impaired in children with dyslexia. In a research programme spanning many years, Snowling and colleagues have investigated phonological deficits over the life span, in particular the ability to read nonsense words, which depend on both the ability to segment and grapheme/phoneme translation.

There is a clear brain basis for phonological difficulties, based on a difference in areas involved in language (the sylvian fissure and the planum temporale). This is found in both the anatomical structure (Galaburda) and the function (work by Chris Frith and colleagues at UCL). Training helps in improving phonological skills, but even high achieving dyslexic adults still show deficits. Despite all the evidence, Frith concludes (1997, p11) “the precise nature of the phonological deficit remains tantalizingly elusive.”

Recent and ongoing research - Phonology

In my view, some of the most interesting studies from the York group fall into 2 groups:

• Predicting which children from an ‘at risk’ population are likely to develop dyslexia (a)

• Longitudinal data on outcomes of childhood disorders (b).

These studies provide evidence that phonology is crucial but not necessarily the whole story, and identify other important predictors of progress.

a) Examining pre-school children with a family history of dyslexia for early indications of dyslexia (Gallagher et al, 2001). There is strong genetic evidence that children with dyslexia in the family have a 50% chance of having difficulties themselves (see Fisher et al, 2001), so this is an important group to target in order to find out more about pre-school development. Letter recognition at 45 months proved the best predictor of literacy at 6 years.

b) A longitudinal study of children identified with language deficits pre-school in terms of their literacy outcomes as teenagers (Snowling et al, 2001). This shows the overlap between Specific Language Impairment (SLI) and Dyslexia, but suggests that around 35% of children with phonological problems develop literacy skills in the normal range, although the gap for less successful children widens. Interestingly, strengths in non-verbal intelligence proved the most protective factor here against continuing difficulties, which may suggest that processing speed is an important factor.

The York group are currently establishing comparative data for phonological and sensory deficits, and naming speed and phonology.

(ii)Sensory Deficit.

This theory is broader than the phonological deficit, because it can potentially account for both visual and auditory deficits in dyslexia. The auditory ‘rapid processing deficit’ was introduced by Tallal in the US, whereas the visual magnocellular deficit comes from Stein and his group in Oxford.

It is now clear that a number of children with dyslexia have problems in processing information coming in via the senses (sensory information). This includes information from both the eyes (visual) and the ears (auditory). This was first identified by Lovegrove, 1993, who found that if you showed a flickering pattern to dyslexic children (think of the lines on a television screen), they needed a more defined flicker before they were aware of it. In more recent work Talcott et al. (1998) showed that dyslexic children needed more information to detect a pattern of stimuli moving together like clouds against a randomly moving background (known as coherent motion sensitivity). Tallaland her colleagues (1993) have claimed that, like language disordered children, children with dyslexia take longer to process sounds which change rapidly. This is tested with high and low tones, or the sounds ba and da, which are only different in the first few milliseconds. Children with dyslexia (and SLI) can’t tell the difference between the sounds if they are presented close together, and this means that they are likely to have problems with phonological awareness.

This type of processing is controlled by the large cells in the magnocellular pathways, which go to a part of the brain known as the thalamus. There are differences in the dyslexic brain anatomy in both visual and auditory magnocellular pathways (Galaburda, Menard and Rosen, 1994; Livingstone, Rosen, Drislane and Galaburda, 1991).

Differences in the visual magnocellular pathway, Stein (e.g., Stein and Walsh, 1997) may cause what is known as ‘visual persistence’ during eye movements. The effect in reading would be that letters in a word drift and blur, because when you try to look at the next letter there is an after image from the previous letter. This explanation seemed to give a good account of the symptoms of blurred vision and letters that move around, which some dyslexic children report. However, it has now become clear that the magnocellular system is not to blame for visual persistence (Stein, 2000). A magnocellular deficit would affect most types of rapid processing, which can be difficult for dyslexic children because it is more demanding for anyone to process material quickly. This is not the easiest theory to get to grips with, because magnocellular deficits cause different problems for vision and audition. In vision, deficits are found for low contrast and/or slowly moving stimuli (Eden et al., 1996; Stein and Walsh, 1997), in audition for rapidly changing stimuli (Tallal, Merzenich, Miller and Jenkins, 1998).

Recent and ongoing research - sensory

Tallal, Merzenich and colleagues (2001) in the US and the Oxford group are producing some of the best work in the area

• Comparing phonological deficits with magnocellular deficits (i)

• Analysing both vision and audition in the same children (ii).

• Sensory development in Infancy (iii).

i) In two recent fMRI studies, Tallal examines the brain basis of rapid auditory processing and links it to phonology, identifying deficits in orthographic (visual pattern matching skills) in dyslexia, in addition to phonological deficits.

ii) Work by Talcott and colleagues (2000, 2001 in preparation) links sensory processing skills to subtypes of reading disability in 350 normal 7-12 year olds. Phonology was tested by reading nonsense words (like ‘torlep’), and orthography by reading irregular words (like yatch). 3.7% of the sample had orthographic problems only (they could read nonsense) and showed delay. 7.4 % of the sample had phonological difficulties only (they could read irregular words) and showed poorer sensitivity to sound frequency; those with difficulties in both tasks (8.9% of the poor readers) showed reduced sensitivity to visual motion. This approach will shortly be applied to children with dyslexia.

iii) Sensory processing in infancy. The Finnish and Dutch infancy studies are examining learning in infancy in terms of both language and sensory development. The work is reviewed in Van der Leij, Lyytinen and Zwarts (2001). In brief the Finnish study looked at the development of 100 children with a family history of dyslexia in comparison with controls in an intensive study from birth. Interesting findings by age 3.5 include the persistence of language deficits and significant associations between motor and language skill development in the familial group. These studies try not only to identify pre-cursors of dyslexia, but also any environmental protective factors which can ‘innoculate’ children against failure. In the US, Molfese (2000) found that infants who were shown to be dyslexic at age 8 could be identified at birth by differences in their brain waves in response to speech and non-speech sounds (see the section on evoked potentials below).

(iii) Speed of processing

Wolf and Bowers (1999) have brought together the phonological and speed problems in the double-deficit hypothesis, which suggests that there are two separate sources of difficulty in dyslexia, phonology and processing speed. Children with both speed and phonology problems have the most severe problems. This is one of the more recently developed theories (for a review see Wolf, 2001).

There is evidence of speed problems for dyslexic children in almost all areas, including those where rapid sensory processing is not needed. This has been known since the 70’s based on ‘Rapid Automatized Naming’ tests (RAN - Denckla and Rudel, 1976), in which dyslexic children show speed deficits in simply saying the names on a page full of simple pictures (or colours). Problems are found even when language is not involved, so children with dyslexia are slower to simply press a button when choosing between a high and a low tone (Nicolson and Fawcett, 1994). In a study which measures the speed of brain waves to see whether the problem lay in registering the tone (sensory) or categorising it as high/low, (using EEG evoked potentials Fawcett et al., 1993) slowed central auditory information processing was found (see recent cerebellar research below). In educational terms, this means that children with dyslexia need longer to read a word that is familiar to them (van der Leij and van Daal,1999) and this may lead to a strategy of trying to process large chunks of letters in reading, rather than breaking the word down phonologically in order to read unfamiliar words. This approach makes heavy demands on working memory, and limits the new words which can be tackled.

Recent and ongoing research – double deficit

The double deficit has become a major focus for research, with the majority of papers presented at the 2001 conference including naming speed. This is largely because it has been clear to practitioners for some time that accuracy is not enough to make a child a good reader, and that they clearly need to develop fluency as well. However, it is not surprising that dyslexic children are slower to name letters and numbers, because they find it more difficult to acquire these in the first place. Some of the most interesting research looks at picture naming, which less predictably is also slowed in dyslexia. The most diagnostic test is the RAN, because it tests the following areas; speed of access to the name of the picture, articulation, eye movements to the next picture, ability to keep your place, and maintain concentration for the whole set of stimuli. However, there are differences in the way some tests are administered which makes them not directly comparable. A standardised methodology is needed to allow comparison in further research.

(iv) Cerebellar Deficit

In the early 1990s, the Sheffield group found that dyslexic children in their panel had severe problems with a wide range of skills, including balance (Fawcett and Nicolson, 1992; Nicolson and Fawcett, 1990); motor skill (Fawcett and Nicolson, 1995b); phonological skill (Fawcett and Nicolson, 1995a) and rapid processing (Fawcett and Nicolson, 1994,b). Many of these skills were not language based, suggesting that the phonological deficit could not explain all the problems in dyslexia. Looking at all the data together (Nicolson and Fawcett, 1995a; Nicolson and Fawcett, 1995b), it was clear that most of the children showed problems ‘across the board’, rather than with different profiles suggesting sub-types. This pattern of difficulties was in line with the dyslexic automatisation deficit hypothesis (Nicolson and Fawcett, 1990), that dyslexic children have problems in fluency for any skill that should become automatic with extensive practice. This hypothesis could explain dyslexic symptoms in phonological skills, in reading, and in other skills, but did not attempt to specify which brain structure was involved.

Problems in motor skill and automatisation point to the cerebellum, an area at the base of the brain known to be associated with motor skill, but largely dismissed in dyslexia because until recently there were no known links between the cerebellum and language. There is now clear evidence that the cerebellum is involved in both language and cognitive skill, including specific involvement in reading (Fulbright et al., 1999). The human cerebellum has evolved enormously, becoming linked not only with the motor areas at the front of the brain, but also some areas further forward in the frontal cortex, including Broca’s language area.

It seemed to the Sheffield group that cerebellar deficit could possibly explain the range of problems suffered by children with dyslexia, and so this idea was tested out, looking first at behaviour and then at the brain. First, it was shown (Nicolson, Fawcett and Dean, 1995) that dyslexic children showed a pattern of poor performance on time estimation and normal performance on loudness estimation that Ivry and Keele (1989) found only in cerebellar patients. This time estimation task did not involve rapid processing; the children simply had to listen to 2 tones with a second between them, with the first tone also lasting a second, and the second tone either longer or shorter. The control task was very similar, but with the tones either louder or softer. Secondly, it was found that children with dyslexia showed a range of classic cerebellar signs (Fawcett and Nicolson, 1999; Fawcett, Nicolson and Dean, 1996) with problems in muscle tone and balance in 80-90% of the children tested. Direct evidence of cerebellar deficit came from a PET scan study which found that dyslexic adults did not show the normal pattern of activation when performing a motor sequence learning task, with only 10-20% the expected level of activation compared with controls.

Recent research – Cerebellar

This falls into 2 groups:

• Learning (i and ii)

• Anatomy (iii)

i) An analysis of how dyslexic children learn (Nicolson and Fawcett, 2001) when asked to blend two simple button pressing tasks, suggests performance can become automatic, but strikingly, a ‘square root rule’, suggests that this takes longer in proportion to the square root of the time normally taken to acquire a skill – see (ii) and Areas for further research below.

ii) A recent study on learning suggests that there may also be abnormalities in fundamental learning processes such as classical conditioning, habituation, response ‘tuning’ and error elimination in an eye blink conditioning study (Nicolson et al, 2001, under review).

iii) Using the brain bank that Galaburda used to check out the phonological and sensory deficits, evidence has been found for differences in the anatomy of the dyslexic cerebellum (Finch et al, 2001, under review).