The evidence we have reviewed supports a number of important conclusions. Most simply, there are at least two different forms of reading disorder in children: dyslexia (decoding problems) and reading comprehension impairment. Although these disorders are usually defined categorically, it is important to emphasize that reading skills show a continuous distribution in the population and each of these disorders is probably better thought of in dimensional terms – children with dyslexia and reading comprehension impairment are simply at the low end of the distribution of decoding and reading comprehension skills respectively. We have also emphasized that both these different forms of reading difficulty can be seen as reflecting problems with language development, and reading disorders are highly comorbid with diagnoses of language impairment. In addition, reading disorders are frequently comorbid with a range of other seemingly disparate disorders including disorders of attention and motor development. Clinicians therefore should take steps to assess a wide range of potential difficulties in children referred for reading problems, but they should also be clear that many such comorbidities may not be causally related to the reading disorder (this does not deny the potential importance of identifying the other difficulties a child may have and providing appropriate treatment for such difficulties).
It is appropriate to relate these conclusions to the diagnostic categories proposed in DSM-5. Within Neurodevelopmental Disorders, DSM-5 separates Learning Disorders (including Dyslexia and Disorders of Written Expression) from Communication Disorders (including Speech-Sound Disorder, Language Impairment (including Late Language Emergence, Specific Language Impairment, Social Communication Disorder and others).
In the current draft, it is notable that there is no listing for either Reading Comprehension Impairment or for Spelling Disorder. We discuss each of these points in turn before considering the issue of comorbidity. First, it is plain from our review that we disagree with the recommendation:
‘that reading comprehension per se be omitted from DSM-5, because individuals who have specific reading comprehension problems in the presence of good decoding skills, do not meet criteria for dyslexia. Such individuals typically are found to have poor oral language (as in communication disorders). However, specific reading comprehension disorders could be coded under the newly proposed superordinate category of learning disability’.
This recommendation seems to us confusing, because in terms of its diagnosis and treatment, reading comprehension impairment needs to be considered as a related but contrasting disorder to dyslexia.
Likewise, DSM-5 does not propose a separate classification for Spelling Disorder. This decision may reflect the fact that children seldom present for clinical assessment with spelling difficulties in the absence of reading problems. However, there are exceptions (Frith, 1980
; Goulandris & Snowling, 1991
), and particularly in regular orthographies, a double dissociation can be seen between reading and spelling impairments (Moll & Landerl, 2009
). The only reference to spelling difficulty in DSM-5 is as a feature of language impairment where it is ‘banded’ with difficulties in written formulation:
LI affects acquisition and use of spoken language (sound-, word-, sentence, and discourse-level comprehension, production and awareness), written language (reading decoding and comprehension; spelling and written formulation), and other modalities of language (e.g. sign language).
The entry for ‘Disorder of Written Expression’ may well be more specific about spelling difficulty, but at the time of writing, the criteria are still under development. Moreover, there is clearly a danger that the confusion between reading accuracy and reading comprehension, which has been removed in DSM-5 will re-emerge as a confusion between disorders of spelling at the word-level (the ability to recreate the legal sequences of printed words) and of written expression, which draws on a wider range of language skills and control processes. We believe that there is enough evidence to show that disorders of spelling can be observed in isolation from reading disorders and given this, a separate category is warranted.
We also believe that there are important reasons for bringing together disorders of reading and language in a classification system. In addition, given issues concerning the stability of diagnostic criteria that rely upon measuring reading attainments using behavioural tests, there is good reason to include, for each disorder, a cognitive description of the phenotype which will remain the same over development despite possible changes in behavioural manifestation (Morton & Frith, 1995
). As the model proposed by Bishop and Snowling (2004)
suggests, it is productive to consider a dimensional approach to the classification of reading impairments such that oral language difficulties in the phonological domain place a child at risk of decoding deficits (dyslexia) whereas wider oral language difficulties (particularly including semantic and grammatical difficulties) place a child at risk of reading comprehension difficulties. While ideally, a dimensional scheme should incorporate what is known of other risk factors, an important spin-off of the two-dimensional model is that it has direct implications for screening, early identification and intervention. The broad aim of this article has been to situate reading disorders within the context of language difficulties; it will fall to clinicians to make these links if they are not explicitly acknowledged in the classification system.
Clinical and educational implications
We have outlined the case that there are two quite distinct forms of reading difficulty in children (dyslexia and reading comprehension impairment), both of which are quite common; spelling difficulty is a key feature of dyslexia, but such difficulties can also occur in isolation. Dyslexia appears to be caused primarily by an underlying weakness in phonological (speech sound) processing whereas reading comprehension impairment appears to reflect broader language processing weaknesses affecting a wide range of skills including vocabulary, grammar, listening comprehension and narrative skills. The difficulties experienced by children with reading comprehension impairment appear to be on a continuum with children who would qualify for a diagnosis of language impairment. However, we have also argued that considering these two disorders in categorical terms is a limited view; the skills which underlie decoding, spelling and reading comprehension are continuously distributed in the population. The literacy outcome for an individual depends upon the interaction of their cognitive strengths and difficulties, primarily with respect to these dimensions, although it is likely that additional deficits (e.g. in executive skills or in speed of processing) will also play a role.
Assessment of reading disorders.
For many years, dyslexia was considered to be a ‘specific’ learning difficulty which affected reading (and spelling), but not other general cognitive abilities. Accordingly, it was defined using a ‘discrepancy’ approach: that is, measured reading attainment should be below the level expected based on age and IQ (Rutter & Yule, 1975
). However, building on a variety of evidence, the role of IQ in explaining reading (decoding) problems has been rejected (e.g. Hatcher & Hulme, 1999
; Shaywitz, Fletcher, Holahan, & Shaywitz, 1992
; Stanovich & Siegel, 1994
). In line with this, DSM-5 explicitly states that reading attainment need not be out of line with general cognitive ability (although it is noteworthy that ‘the term dyslexia might be used
where there is evidence of a discrepancy’– this caveat is important to accommodate cases of dyslexia in high-functioning individuals in whom, reading deficits are mild or compensated as we have argued in relation to the ‘broader phenotype’).
It follows that DSM-5 recommends reading should be assessed as follows:
Multiple sources of information are to be used to assess reading, one of which must be an individually administered, culturally appropriate and psychometrically sound standardized measure of reading and reading-related abilities.
Our review highlights the need to ensure that tests of reading fluency and comprehension are included to ensure the nature of the disorder is clarified – we note here that problems of reading fluency are far more common in readers of the European alphabetic languages than are problems of reading accuracy, reflecting the relative ease of learning to decode in these orthographies. Furthermore, the recommendation does not detail ‘reading-related abilities’; we would argue that information critical to intervention should be gathered from tests of phonological awareness and related skills, such as rapid naming (particularly important for dyslexia) and oral language skills (particularly important where there are problems of reading comprehension impairment) (Muter & Snowling, 2010
). A single word spelling test, separate from written expression, is also important to differentiate dyslexic difficulties from problems of language expression and or motoric skills.
We would note that ‘diagnosis’ inevitably involves placing arbitrary divisions on what is essentially a continuous distribution of reading skills in the population. Although, administratively it may sometimes be useful and convenient to categorize children as being ‘dyslexic’ or a ‘poor comprehender’, in reality, there are in the population continuous variations in reading and other cognitive skills underlying a ‘spectrum’ of reading difficulties. This has implications for the validity of diagnostic labels (Rutter, 2011
). The longitudinal stability of a ‘dyslexia’ diagnosis is low (Shaywitz et al., 1992
) and this is true at all levels of severity of the disorder (Wagner, Brown Waesche, Schatschneider, Maner, & Ahmed, 2011
). Moreover, there is little agreement between traditional behavioural definitions of dyslexia and definitions which use ‘response to intervention’ as a metric (Brown Waesche, Schatschneider, Maner, Ahmed, & Wagner, 2011
). Such findings caution against the use of a diagnostic category to describe an individual whose learning disorder is prone to change with age and in response to intervention (Fletcher et al., 2007
Finally, assessing reading and related skills only touches the surface of some of the complex issues surrounding the diagnosis of reading disorders. Following a recent review for the UK government, Rose (2009)
stated that ‘Dyslexia....is best thought of as a continuum, not a distinct category, and there are no clear cut-off points. Co-occurring difficulties may be seen in aspects of language, motor co-ordination, mental calculation, concentration and personal organization, but these are not, by themselves, markers of dyslexia.’ Thus, while there is good agreement that dyslexia significantly impedes learning, continuities and co-morbidities with other language and learning disorders complicates identification, assessment and diagnosis. A similar argument could be made for reading comprehension impairment.
Interventions for children’s reading difficulties. The different causes of dyslexia and reading comprehension impairment clearly imply that different forms of intervention will be required to help children with these different forms of reading disorder. The different forms of intervention required to reinforce the importance of the correct assessment and diagnosis of children’s reading difficulties.
There is now a considerable body of evidence about effective interventions for children with dyslexia, and a gradually increasing body of evidence concerning interventions for reading comprehension impairment (for recent reviews see Duff & Clarke, 2011
; Fletcher et al., 2007
; Snowling & Hulme, 2011
). As might be expected from the theoretical framework we have discussed, children with dyslexic difficulties benefit from teaching that directly targets learning spelling-sound relationships and helps to overcome their phonological difficulties (Bus & van Ijzendoorn, 1999
; Torgesen, 2005
), whereas children with reading comprehension impairments require interventions that work on the oral language skills that underlie the condition (Clarke, Snowling, Truelove, & Hulme, 2010
; National Reading Panel, 2000
A natural question is whether, if children could be identified early as being ‘at risk’ of reading difficulties, interventions could prevent the development of dyslexia and/or reading comprehension impairment. The evidence we have so far is limited, but does suggest that later reading problems might at least be reduced by suitable early interventions. There is a long history of studies showing that early phonological training can benefit the development of decoding skills (Bradley & Bryant, 1978
; Lundberg, Frost, & Petersen, 1988
). One of the first randomized trials to address this issue was conducted by Bowyer-Crane et al. (2008)
who evaluated early interventions to ameliorate both decoding and language comprehension difficulties. This study compared a Phonology with Reading programme (P + R) to an Oral language Programme (OL) delivered by specially trained teaching assistants to children who were selected for having weak oral language skills at school entry. The children who received the P + R programme did significantly better on tests of phoneme awareness, letter-sound knowledge, basic reading and spelling skills than children who received the OL programme, whereas those who received the OL programme did significantly better on tests of vocabulary and grammar, and there was a trend for more improvement in narrative skills. Moreover, the relative gains for both groups were maintained some 5 months after the intervention had ceased. A very important question, that is still virtually unaddressed, is the extent to which such interventions can have long-term effects, and there is a pressing need for large scale randomized trials of long duration to evaluate this issue.