Dyslexia is a neurodevelopmental disorder which primarily affects the development of reading accuracy, fluency and spelling skills (IDA, 2002; American Psychiatric Association, 2011). For some individuals with dyslexia, reading difficulties may be overcome leaving impairments only in spelling and aspects of phonological processing (Bruck, 1992; Ramus et al., 2003). For others, dyslexia persists well beyond the school years and may affect adult career and employment prospects (Maughan et al., 2009).
One obstacle to the identification of dyslexia is a lack of consensus concerning its defining symptoms, and debate concerning the core characteristics of dyslexia has continued for more than 40 years (Snowling, 2009; Vellutino, Fletcher, Snowling & Scanlon, 2004 for reviews). Increasingly, it is accepted that dyslexia is not an ‘all or none’ condition but rather a dimensional disorder underpinned by poor phonological skills (Pennington & Lefly, 2001; Hulme & Snowling, 2009). In addition, there is recognition that dyslexia tends to co-occur with other language and learning disorders (Pennington & Bishop, 2009). An individual is more likely to receive a ‘diagnosis’ of dyslexia if they have relatively severe reading and spelling difficulties or if their literacy difficulties are compounded by co-morbid difficulties with language or attention (Snowling, 2008). Indeed, problems of language and of attention are commonly considered to be dyslexia-associated traits (Rose, 2009).
In recent years, there has been a growth of interest in children at family risk (FR) of dyslexia across languages (e.g., in English, Scarborough, 1990; in Finnish, Lyytinen et al., 2006; in Chinese, McBride-Chang et al., 2011). An important aim of these studies is to identify the precursors of dyslexia before reading instruction begins. Equally important is to understand the inter-generational transfer of risk between parents and their offspring. Thus, van Bergen, de Jong, Plakas, Maasen and van der Leij (2012) showed that parental reading skills contributed unique variance to the prediction of children's reading fluency even when child-level predictors, such as phonological awareness, were controlled.
To date, the majority of studies of children at FR have screened parents only for literacy and related phonological skills (e.g., Snowling, Gallagher & Frith, 2003) or, in some cases, have relied on self-report of dyslexia (e.g., Hindson et al., 2005). Since current theories suggest dyslexia is the behavioural outcome of multiple risks acting together to increase the probability of poor reading (Pennington, 2006), it follows that family-risk studies need to incorporate procedures for estimating risk not only of dyslexia but also of commonly co-occurring disorders. Such disorders include Attention Deficit Hyperactivity Disorder, a disorder involving difficulties with the control of attention and behaviour associated with symptoms of inattention, hyperactivity/impulsivity or both (e.g., McGrath et al., 2011) and Language Impairment, a disorder characterized by delayed language development alongside normal nonverbal ability (Pennington & Bishop, 2009); both of these disorders, like dyslexia, are dimensional in nature and vary in severity.
Questionnaires and rating scales provide a time-saving way of estimating risk factors for dyslexia. Many dyslexia screening questionnaires include not only questions about literacy skills but also items which tap constructs, such as problems of attention, organization and word finding (e.g., Cooper & Miles, 2011; Smythe & Everatt, 2001; Vinegard, 1994). Potentially, these questionnaires offer additional information relevant to the quantification of risk, but few are validated. The present study investigated whether a self-report questionnaire could provide a valid measure of literacy, and further, whether this measure could be useful as part of a protocol for the identification of ‘dyslexia’ and co-occurring problems in adulthood. The primary reason for validating the protocol was for use in family studies investigating risk factors for disorders. We describe data collected from parents of pre-school children. The study sample as a whole was diverse in terms of educational level, socio-economic background and occupational status and included people with dyslexia and was therefore considered appropriate for the validation of a dyslexia-risk screening protocol.
A number of previous studies have assessed the validity of questionnaires and interviews for the self-report of reading difficulties (Gilger 1992; Gilger, Pennington & DeFries, 1991; Pennington & Lefly, 2001; Schulte-Körne, Deimel & Remschmidt, 1997). Evidence of the predictive validity of self-report comes from its relationship to measured reading and spelling skills as well as to the emergent reading skills of the offspring of respondents (Elbro, Nielsen & Petersen, 1994). Wolff and Lundberg (2003) found that a self-report measure of dyslexia correlated well with measures of word recognition in a group screening of Swedish adults. Lindgren and Laine (2007) also provided validation data from a Finnish version of the same measure. However, Gilger (1992) showed that the accuracy of retrospective self-report of academic attainments in school depends upon a number of factors including age, gender, level of achievement and history of learning disabilities. How to interpret these effects is unclear. We first sought to replicate and extend the findings to the self-report of dyslexia and then to gain better understanding of the weaknesses of self-report methods by considering the extent to which they under- and over-identify literacy difficulties.
In summary, in the present study, we set out to validate a protocol to assess dyslexia and dyslexia-associated traits in adults. We anticipated that self-reported dyslexia should differentiate people with and without dyslexia-associated traits, namely problems of word finding, organization, attention and hyperactivity. Also, we expected that people who rated their reading difficulties/dyslexia as more severe should gain lower scores on measures of literacy. Finally, given that reading difficulties tend to be associated with low levels of print exposure both in children and in adults (Cunningham & Stanovich, 1997; Maughan et al., 2009, Snowling, Muter & Carroll, 2007), we expected that ratings of reading difficulty would predict a lower frequency of literacy-related activities in everyday life.
- Top of page
- Appendix A
The main aim of the present study was to develop and evaluate a protocol for identifying dyslexia and related difficulties in adults. A specific objective was to design a self-report scale that could be used in family-risk studies to provide a means of estimating the multiple risks associated with dyslexia that a parent confers on their child. Hence, it was important to include items that assessed domains beyond literacy per se and included measures of attention control and expressive language, specifically word finding. The present protocol included a new ARQ asking for ratings of reading and spelling proficiency, frequency of reading and writing on an everyday basis and self-report of dyslexia, as well as completion of a validated screening tool for attention deficit hyperactivity disorder. We assessed the factor structure of the protocol and the validity of the constructs it measured in relation to behavioural measures of reading and spelling.
A four-factor measurement model (Reading, Word Finding, Attention and Hyperactivity) provided a reasonably good fit to the data from the two questionnaires. The Reading and Word Finding factors correlated strongly, as did the Word Finding and Attention Factors, with the remaining intercorrelations being moderate. The Reading, Word Finding and Attention factors showed reasonable reliability, the Hyperactivity Factor (defined by two items) was less reliable.
The validity of the new reading questionnaire (ARQ) appears to be good. Importantly, the Reading Factor showed strong concurrent relationships with measured literacy skills (particularly decoding fluency and spelling). Furthermore, adults who self-reported as ‘dyslexic’ gained lower scores on the Reading Scale than those who did not. In addition, the Rated Severity of individuals' reading difficulties correlated with their scores on the Reading Scale and, consistent with previous research, negatively with measures of reading and spelling (Gilger 1992; Gilger et al., 1991; Schulte-Körne et al., 1997; Wolff & Lundberg, 2003). However, contrary to prediction, Severity of difficulties was only weakly related to Frequency of Reading and Writing in everyday life.
Together these findings indicate that the ARQ provides a valid continuous measure of literacy skills, and, if used together with the ADHD screener, the protocol can identify dyslexia-associated traits including difficulties with expressive language (word finding) and attention. The protocol may therefore be useful as a way of estimating some of the risk factors (or endophenotypes) involved in the etiology of dyslexia in ‘at-risk’ children. Furthermore, the current findings fit well with causal hypotheses of dyslexia. First, it has been suggested that the learning to read capitalizes on the neural circuitry involved in object naming (Price & McCrory, 2005, Lervåg & Hulme, 2009). Consistent with this view, there was a high correlation between Reading and Word finding Scales. Second, the correlation of Attention and Word Finding suggests that a frontal brain network is associated with both word retrieval and attention control, as might be predicted from the high degree of association between dyslexia and symptoms of inattention (Carroll, Maughan, Goodman & Meltzer, 2005).
However, it is important to signal a note of caution about using a self-report measure for the purposes of classifying individuals. While there were robust correlations between the Reading Scale scores and behavioural measures of literacy, this study identified a substantial number of people with low levels of literacy who did not rate themselves as having poor reading and spelling skills or who did not self-report as dyslexic. The reasons for this under-reporting are likely to be diverse as also noted by Gilger (1992). Indeed, over and above objective measures of literacy skill, there was a stronger tendency for fathers, for older parents and for those with higher educational qualifications to self-report as dyslexic. It is difficult without more data to understand an individual's propensity to self-report dyslexic symptoms. Speculatively, and based on anecdotal evidence, it may be that it is socially more acceptable for men to admit dyslexia, and they may often be encouraged by their partners to do so. Theoretically, however, there is growing evidence that people rate their own emotional state in relation to their social group (e.g., happiness; Boyce, Brown & Moore, 2010). Applying this hypothesis to the current findings, we propose that adults with higher educational qualifications who have relatively mild dyslexic difficulties may see themselves as more handicapped in the workplace than their peers and hence be more likely to self-report as dyslexic than a similarly affected person in a manual job.
In summary, a protocol such as the one described here provides a useful tool for screening for dyslexia and attention difficulties in adults. The finding that people who self-report as dyslexic rate themselves as having more difficulty with word finding, attention and hyperactivity than those who do not, underlines the fact that parents confer multiple continuous risks for learning disorders on their children. It is important for research following children at family risk of dyslexia to be aware of these effects.