Reading comprehension skills of Dutch university students with dyslexia
Nelleke Willemijn Varkevisser University of Groningen
June 29th, 2018
Master’s Thesis Neurolinguistics Faculty of Arts
Supervisor: W. Tops Word count: 21.936
Abstract
Reading comprehension is an essential skill for university students. In universities in the Netherlands, a large number of academic texts has to be read in Dutch, but also in English. The present study investigated the reading comprehension skills of Dutch-English bilingual students with and without dyslexia. It examined whether differences could be perceived between the two groups and whether differences exist between reading comprehension in the L1 Dutch and the L2 English. Variables that could affect Dutch and English reading comprehension performance (language proficiency, vocabulary knowledge, print exposure, and prior knowledge) were also investigated. Moreover, the effects of text-to-speech software on reading fluency and reading comprehension were examined. The results indicated that the high-performing students with dyslexia still experienced difficulties with reading fluency compared to the students without dyslexia. Reading comprehension appeared to be similar in the two groups at first glance, but when examined in greater detail it became clear that the students with dyslexia had recall difficulties and that there was a difference between the performance on inferential and literal questions in the dyslexia group which could not be perceived in the control group. Use of the text-to-speech software did not appear to have any particular beneficial effects for students with dyslexia. When examining the difference between Dutch and English reading comprehension, it was clear that language proficiency especially was not equal for both languages, and that this was an important factor in determining reading comprehension performance in both groups. Other factors such as print exposure and prior knowledge also appeared to affect reading comprehension scores and these factors combined with language proficiency and vocabulary knowledge could possibly modulate the negative effects experienced by students with dyslexia.
Keywords: Dyslexia, bilingualism, reading comprehension, text-to-speech software, language
Introduction
On an international level, it has been observed that there is an increasing number of students with developmental dyslexia who enter higher education (Madriaga et al., 2010). This is not necessarily an indicator that the prevalence of dyslexia has increased over time, but rather that individuals with dyslexia may feel less inhibited by their learning disorders than in previous decades. In other words, they may feel more confident that they can achieve the same level of performance at higher education as their fellow students without learning disabilities. An increase of special facilities for students with learning disabilities is also likely to be a contributing factor. However, it is the question if even with such facilities, students with dyslexia can indeed perform at the same level as their peers without reading difficulties. Reading and reading comprehension are skills which are of primary importance in higher education. In the Netherlands, much of the course material at universities is comprised of both Dutch and English texts. This requires good reading and reading comprehension skills in both of these languages. The present study examined how well-developed the reading comprehension skills of high-performing Dutch students with and without dyslexia are, and if any differences can be perceived in the reading comprehension skills in Dutch and English between these two groups. As such, this study compared a group of students with developmental dyslexia and a group of students without dyslexia, in order to investigate if students with dyslexia have poorer reading comprehension skills in general, and if differences can be perceived between reading comprehension in the L1 Dutch and in the L2 English. Furthermore, possible factors which could modulate the reading and reading comprehension skills of the students were examined.
Developmental dyslexia
Developmental dyslexia is a specific learning disorder with impairment in reading (DSM-5; American Psychiatric Association, 2013). The DSM-5 further indicates that the term dyslexia can be used to describe a pattern of learning difficulties which may be concerned with accurate or fluent word recognition, poor decoding skills, and poor spelling abilities. In the Netherlands, the Dutch Dyslexia Foundation (Stichting Dyslexie Nederland (SDN), 2016) has adopted a similar definition of dyslexia. According to this definition dyslexia is a specific learning disorder, characterised by persistent difficulties in accurate and fluent word reading and/or word spelling. These problems do not occur as a consequence of environmental factors (e.g. poor literacy training) or a physical, neurological, or mental disorder (SDN, 2016). Because the present study concerns Dutch-speaking individuals with dyslexia, the SDN definition is used to define developmental dyslexia in Dutch adults and adolescents.
The difficulties of individuals with dyslexia may vary according to the orthographic depth of a certain language. Reading difficulties in consistent or transparent orthographies, such as Finnish, are often characterised by problems with reading fluency, as opposed to problems with reading accuracy (e.g. Ziegler & Goswami, 2005). In opaque orthographies, such as English, in which the mapping between
letters and sounds is less consistent, problems with accuracy as well as fluency may occur (Peterson & Pennington, 2012). Phonological awareness has been found to be a key predictor of reading skill in both transparent and opaque orthographies (Bekebrede, Van der Leij, Plakas, Share, & Morfidi, 2010; Peterson & Pennington, 2012). However, its influence has been found to be slightly weaker in transparent orthographies, as compared to more opaque orthographies (Ziegler et al., 2010). Phonological awareness can be described as the ability to access, process, and manipulate speech sounds. This is crucial for the development of reading skills, because it aids in the establishment of automatized letter-sound correspondences. The ability to phonologically decode words in turn underlies fluent and accurate word recognition, which is the basis for reading (Peterson & Pennington, 2012). Problems with phonological processing and phonological awareness have been described as a core characteristic of dyslexia in alphabetic languages (Vellutino, Fletcher, Snowling, & Scanlon, 2004). However, it is unlikely that a single phonological deficit causes dyslexia (Peterson & Pennington, 2012). Instead, phonological difficulties may interact with a variety of other risk factors and protective factors. As such, the development of dyslexia appears to be dependent on a complex interaction of various genetic, neurobiological, cognitive, behavioural, and environmental factors. None of these factors seem to be sufficient to cause dyslexia in and of their own (e.g. Blomert & Willems, 2010; Van der Leij et al., 2013; Van Bergen, Van der Leij, & De Jong, 2014). Current research has therefore taken a multi-factorial approach towards the factors that can influence the reading and spelling difficulties which are characteristic of developmental dyslexia.
Studies investigating adults with dyslexia have found that a phonological deficit persists into adulthood (e.g. Miller-Guron & Lundberg, 2000; Miller-Shaul, 2005; Snowling, Nation, Moxham, Gallagher, & Frith, 1997). This was also found to be the case for Dutch adults with dyslexia (Bekebrede et al., 2010). Rapid automatized naming, otherwise known as RAN or the ability to quickly name continuously presented familiar stimuli such as letters, digits, and colours, has been found to be a second core deficit in dyslexia (e.g. Georgiou & Parrila, 2013; Kirby, Georgiou, Martinussen, & Parrila, 2010). RAN deficits have been found in adults with dyslexia when compared to adults without dyslexia (Georgiou, Ghazyani, & Parrila, 2018). Similar to phonological deficits, these RAN deficits have also been found in Dutch adults with dyslexia (Callens, Tops, & Brysbaert, 2012). A third cognitive skill which is often linked with dyslexia and for which deficits have been found in adults with dyslexia is verbal short-term memory (Martinez Perez, Majerus, & Poncelet, 2013). However, the idea that verbal short-term memory can (partially) explain reading difficulties in adults with dyslexia is much more contested (cf. Staels & Van den Broeck, 2014).
In general, the manifestation of dyslexia in adults is extremely variable (Mapou, 2008). Most adults with dyslexia have already had many years of reading instruction and print exposure, which means that it is likely that they have developed compensatory strategies in order to cope with the limitations posed by their learning disability. This makes it difficult to study adults with dyslexia, also because the definition of developmental dyslexia that is used for children with this learning disability may not apply
anymore due to this compensation (Mapou, 2008). There is evidence that problems with reading accuracy appear to decrease for more transparent languages as reading experience increases (e.g. Kuijpers et al., 2003). Dutch is a relatively transparent language (e.g. Ellis et al., 2004). As such, it has been found that Dutch adults with dyslexia tend to have fewer problems with accuracy than Dutch children with dyslexia, but reading fluency remains an issue for adults (Callens et al., 2012). For less transparent languages, such as English, problems with decoding accuracy as well as fluency issues are characteristic for dyslexia in both children and adults (Kuijpers et al., 2003).
Dyslexia in higher education
In the past few decades, there has been an increase of adolescents and adults with developmental dyslexia who enter higher education (e.g. Pavey, Meehan, & Waugh, 2010). Although there do not appear to be any recent studies about the prevalence of dyslexia in Dutch higher education, a large-scale study in 2001 determined from a sample of 478.000 Dutch students that approximately 2-3% of the students enrolled in higher education in the Netherlands have dyslexia (Broeninck & Gorter, 2001). The number of Dutch students with dyslexia may have grown slightly more in the years between the study of Broeninck and Gorter (2001) and the present study. However, it is important to note that such an increase is not likely to indicate a general increase in the prevalence of dyslexia in the Netherlands. Instead, the optimisation of primary and secondary education for individuals with dyslexia may allow them to overcome their difficulties, and as a consequence this could result in the increase of individuals with dyslexia in higher education (Tops, Callens, Lammertyn, Van Hees, & Brysbaert, 2012). Furthermore, by offering special arrangements (e.g. exam facilities, education facilities) to students with learning disabilities like dyslexia, these students are enabled to attain the same level of high performance as students without a learning disability. Adolescents and adults with dyslexia may have developed sufficient compensatory strategies and therefore they may feel less hampered by their learning disability, which in turn enables them to attain the same level of performance as individuals without dyslexia (Callens & Tops, 2015).
Studying individuals with dyslexia in higher education is challenging, precisely because they have often developed compensatory strategies to overcome the effects of dyslexia (e.g. Deacon, Cook, & Parrila, 2012; Lefly & Pennington, 1991). This is a result of their continued interaction with the educational system and the demands that this puts on their academic skills. However, although these students are often described as compensated or high-functioning dyslexics (Kemp, Parilla, & Kirby, 2009; Lefly & Pennington, 1991), they are rarely able to achieve complete compensation for their reading and/or spelling difficulties. Timed as well as untimed word-reading difficulties are especially common to persist in this group of individuals (Deacon et al., 2012). Nevertheless, many of them are successful in their academic careers. Studying this particular group of individuals with dyslexia may provide important insights in the strategies that high-functioning adults employ to counter reading and
spelling difficulties. This in turn could shed light on the problems of students who are not as successful in compensating for their difficulties, and indicate ways in which they could possibly counterbalance their reading and/or spelling difficulties, consequently optimising their academic careers.
Reading comprehension in adult dyslexia
Reading represents one of the most important skills in university (Onwuegbuzie & Collins, 2002). Reading is not a simple process of decoding the letters and words and the sentences that they form; it also involves reading comprehension. This is thought to represent the efficient integration of previously acquired knowledge with the information that is provided in a text (Onwuegbuzie & Collins, 2002). In other words, reading comprehension is the ability to process a text and integrate its meaning with knowledge that is already present. It has been shown that developmental dyslexia can affect higher level reading comprehension and writing skills (Tops, Callens, Cauwenberghe, Adriaens, & Brysbaert, 2013). The quantity of reading that is required for university courses in all disciplines is usually very high (Du Boulay, 1999). Furthermore, academic texts in many university courses are lengthy, syntactically complex, and contain abstract technical language and new terminology. As a consequence, they are very time-consuming to read for many students with reading difficulties or dyslexia (Pedersen, Fusaroli, Lauridsen, & Parrila, 2016).
Several studies have investigated reading comprehension skills in adults and adolescents with dyslexia. Simmons and Singleton (2000) compared the English reading comprehension abilities of ten native English adults with dyslexia and ten control participants without dyslexia. All participants were current university students or had just graduated. In order to test their reading comprehension, the participants were required to read a text of 655 words, which was followed by five literal and five inferential multiple choice questions. There was no time limit for the task and the participants were allowed to refer back to the text while answering the questions. The results revealed that the adults with dyslexia performed poorer than the adults without dyslexia on the inferential questions, but not on the literal questions. Simmons and Singleton (2000) concluded that the difficulties with reading comprehension could not be attributed directly to an inability in decoding individual words in the text. Instead, they suggested that an impaired working memory could contribute to the reading comprehension problems that were evident in the students with dyslexia. These students might devote more resources to decoding the individual words. Since the working memory capacity is finite, this means that they cannot devote as many resources to comprehending the text as the individuals without dyslexia (Simmons & Singleton, 2000). Since the participants were allowed to refer back to the text while answering the questions, it is not surprising that their performance on the literal questions did not differ significantly, as the answers to these questions could be found directly in the text. However, it is interesting that even though there was no time limit and they were allowed to use the text, the participants with dyslexia still scored significantly lower on the inferential questions. Although the decoding skills
of students in higher education may have been sufficiently boosted by intervention and practice (Simmons & Singleton, 2000), difficulties in reading comprehension may persist in adulthood and the effects of dyslexia do not seem to be eliminated. In a similar vein, Pedersen et al. (2016) examined reading comprehension skills in combination with oral reading quality. Their participants were native Danish-speaking students with dyslexia and a comparison group without reading difficulties. The participants were required to read aloud a complex text of 223 words, and subsequently their reading comprehension was measured by means of aided text retellings. Pedersen et al. (2016) found that the students with dyslexia made more errors in their retellings, and that these students had more trouble making inferences, rephrasing, and giving additional information about the narrative. Furthermore, the retellings of the students with dyslexia were based more around literal facts from the text, often retold exactly as they had been written in the text. The study of Pedersen et al. (2016) for Danish therefore appears to corroborate the findings of Simmons and Singleton (2000) for English. The students with dyslexia in the Pedersen et al. (2016) study were also more frequently unable to answer the questions which measured reading comprehension, which could be an indicator that they were more focused on decoding than on comprehension while reading the text. Like Simmons and Singleton (2000), Pedersen et al. (2016) suggested that having to engage in both decoding and comprehension at the same time may be too demanding for readers with dyslexia, and as such, these individuals may choose to focus their attention on one of these components at a time.
The use of text-to-speech software is one of the options in the domain of assistive technology that students with dyslexia could use to mitigate their decoding difficulties, which would allow them to turn their focus more towards comprehension of the text. The implementation of text-to-speech software means that rather than having to read the text visually, the reader can access it aurally (Smythe, 2010). For those who have difficulties with decoding this could be beneficial, as several studies have noted (for an overview see e.g. Holmes & Silvestri, 2012). There are several different types of text-to-speech software, but for the purpose of the present study only the classic version is addressed here. This concerns stand-alone programs which function with a toolbar with different controls, such as stop and play, alternative voices, and reading speed changes (Smythe, 2010). ClaroRead (2005) is one example of such a program. Although no studies appear to have focussed on the beneficial effects of ClaroRead (2005) in specific, there are studies suggesting that text-to-speech programs reduce frustration of inaccurate decoding and consequently allow for better reading comprehension (Forgrave, 2002). For example, Callens et al. (2012) reported that the students with dyslexia in their study performed on the same level as the control group on the text comprehension test in which the text was also read out by a computer. Similarly, Montali and Lewandowski (1996) found that the adolescents in secondary school (eighth and ninth grade) who were characterised as poor readers reached the comprehension levels of students with average reading skills in the condition in which the text was simultaneously presented visually and aurally (Forgrave, 2002). As such, for students with dyslexia in higher education, a bimodal presentation of texts could have a potential benefit in terms of reading comprehension.
There are several other factors which could also be of influence in reading comprehension in dyslexia. Ransby and Swanson (2003) examined English reading comprehension accuracy in order to determine the role of word recognition in reading comprehension in adults with dyslexia. They found that phonological processing, naming speed, vocabulary, general knowledge, and listening comprehension were all factors which contributed to reading comprehension accuracy. As such, the authors suggested that no single cognitive process dominates other processes in the prediction of reading comprehension skill. Instead, the coordination of several cognitive processes is the best predictor of reading comprehension difficulties. It is also likely that the amount of reading experience that individuals have had with (academic) texts influences their reading comprehension. For example, Onwuegbuzie and Collins (2002) found that English graduate students without reading difficulties had higher scores on the reading comprehension measures than undergraduate students without reading problems. Although the authors did not examine students with dyslexia, it is very likely that students
with reading difficulties who have had more exposure to academic texts are better able to comprehend
these texts than students with dyslexia who have not yet had a great deal of reading experience with academic texts.
Dutch-English bilingualism in the Netherlands
By the time Dutch adolescents and adults enter higher education, they have typically already had many years of reading instruction in both Dutch and English. They have been formally exposed to texts in their first language from grade 1 (at the age of six) onwards, and formal English (reading) instruction normally starts when children are around ten years old (grade 5) in the Netherlands (Van Setten et al., 2017),although some primary schools may start even earlier (Van Berkel, 2012). Furthermore, there is a high level of bilingualism between English and Dutch in the Netherlands, which means that English should be regarded as a second language for many Dutch individuals, rather than as a foreign language (McArthur, 1996). As has been mentioned before, the orthography of English is less transparent than the orthography of Dutch. While the English orthography consists of 40 phonemes which can be written in 561 different ways, Dutch orthography in comparison has 40 phonemes which can be written in only 76 different ways (Dewey, 1971). Because the orthography of English is much more opaque than the orthography of Dutch, it is possible that individuals with developmental dyslexia will have more difficulties with reading and reading comprehension in English than in Dutch, despite the fact that in the Netherlands most people come into contact with (written) English on a daily basis from a very young age onwards.
Bilingualism and adult dyslexia
Depending on the course programme and the available course material, academic texts at Dutch universities and universities of applied sciences may have to be read in either English or Dutch. Having to read and understand academic texts in a non-native language may pose an additional challenge for students with reading difficulties or dyslexia. When assessing dyslexia in a second language (L2), it is important to separate reading difficulties from poor language proficiency in the L2 (Elbro, Daugaard, & Gellert, 2011), that is, to assess whether individuals with dyslexia also have reading difficulties in their L2, and to examine if these difficulties are in addition to the difficulties that are posed by learning a second language in general (Elbro et al., 2011). In other words, it is important to know if reading difficulties in the L2 are due to the learning disability itself, or if they are also caused in part by a poorer L2 proficiency. This issue is also addressed by Hedman (2012), who examined Spanish-Swedish adolescents (mean age: 14;9) with and without dyslexia. After examining phonological processing and decoding for these individuals in both languages, Hedman (2012) proposed a bilingual dyslexia
continuum. By doing so, the author aimed to establish the importance of considering the results of tests
in both a bilingual’s languages; only then can the scores of bilinguals with dyslexia be evaluated accurately and fairly. Because reading comprehension is partly dependent on language comprehension (Elbro et al., 2011), it follows that if L2 language proficiency is poor, then reading comprehension in that language is also likely to be poor. Another natural assumption is that reading skills of most unimpaired readers are better in the first language than in a second language (Miller-Guron & Lundberg, 2000). Furthermore, difficulties with foreign language learning appear to be founded upon native language deficits (Miller-Guron & Lundberg, 2000). As a consequence, individuals with dyslexia, who have specific difficulties with reading and/or spelling in their first language, may be inhibited in their ability to become fully proficient in their reading abilities in a second language.
There are few studies which have examined dyslexia in adults in a second language. Notable exceptions are the studies conducted by Lindgrén and Laine (2007; 2010; 2011), who examined Swedish-Finnish students with dyslexia in a series of studies. In 2010, Lindgrén and Laine compared the performance of multilingual Finnish students with dyslexia with matched control participants. The test battery included two text-reading tests, one of which measured oral reading speed and accuracy, and one which tapped reading comprehension skills. The reading comprehension test had a variety of different questions, including retelling the story, multiple choice questions, and true-false judgements. Like in the study by Simmons and Singleton (2000), the participants were allowed to refer back to the text, but only when they were instructed to do so by the examiner. The results indicated that during the reading aloud task in Swedish, the group with dyslexia read the text significantly slower than the control group. For the Finnish text and the combined score of the two texts, the difference approached significance. The participants with dyslexia did make significantly more errors in both the reading of the Finnish and the Swedish texts compared to the control participants. Interestingly, no significant differences emerged in the reading comprehension task for Finnish or Swedish. This is in contrast to the
findings of Simmons and Singleton (2000) and Pedersen et al. (2016) described earlier. Lindgrén and Laine (2010) do not appear to have investigated if there were significant differences within the group of participants with dyslexia on these tasks, i.e. if the scores for Finnish reading comprehension differed significantly from those of Swedish reading comprehension. The authors note that a possible reason for the deviant results in comparison to other studies could be that reading comprehension problems often surface in self-reports (which was also the case in their study) and that it is possible that reading comprehension could be affected under more demanding circumstances than those present in their study. In their study, Lindgrén and Laine (2010) used fiction texts, and there was no time restriction for the texts. They note that these two factors, i.e. time pressure and type of text, can increase the demands on reading comprehension. This may in turn bring about differences between a group of individuals with dyslexia and a group of unimpaired readers. Lindgrén and Laine (2010) also make a suggestion which is not related to reading comprehension, but to reading fluency in transparent languages. They suggest that in transparent languages, speed should not only be considered to be the primary marker of dyslexia, at least not for high-performing university students with dyslexia. Their results for the text-reading task clearly indicated that accuracy measures were more important than speed in differentiating the two groups of participants. Since the present study is concerned with Dutch university students with dyslexia, and Dutch is also a relatively transparent language (Ellis et al., 2004), this is an important finding which should be taken into consideration.
In their 2011 study, Lindgrén and Laine aimed to examine reading and writing skills of students with and without dyslexia in two domestic languages (Swedish and Finnish) and a foreign language (English). The test battery included an oral reading text for all three languages. The authors found that the dyslexia group read the English text significantly slower than the control group. For Swedish the difference in speed approached significance, but there were no significant differences for Finnish text reading. In terms of accuracy, the dyslexia group produced significantly more errors than the control group when reading the texts out loud, regardless of the language. Consequently, Lindgrén and Laine (2011) concluded that language proficiency and orthographic depth can modulate the reading performance of high-performing multilinguals with dyslexia. Their reading difficulties appeared to surface most clearly in English, the less proficient foreign, orthographically opaque language. Thus, although Lindgrén and Laine (2010; 2011) did not find any significant differences in reading
comprehension skills between adults with dyslexia and adults without reading impairment, their findings
did indicate that high-performing adults with dyslexia are unlikely to fully compensate their reading impairment. They remain slower readers compared to adults without dyslexia and may also have difficulties with reading accuracy, despite the shallow orthography of the languages in question. Lindgrén and Laine (2010; 2011) also provided evidence that factors such as task demands are important to take into consideration when assessing the reading skills of high-performing adults with dyslexia.
Other (environmental) factors, such as reading motivation, print exposure to the L2, and exposure to the L2 in general, are also likely to affect reading performance in a certain language for both readers
with and without dyslexia. In 2007, Lindgrén and Laine published the results of a Finland-Swedish adaptation of a screening test for dyslexia for (young) adults. Once again their participants were Swedish-Finnish university students with and without dyslexia. From their results they concluded that there were associations between task performances and language proficiency. Consequently, Lindgrén and Laine (2007) pointed to the importance of taking into account the language background of participants when interpreting results of bilingual or multilingual individuals, even if they are highly proficient in both of their languages. This conclusion is in line with the findings of their later (2011) study, in which they investigated Swedish, Finnish, and English proficiency. Miller-Guron and Lundberg (2000) also aimed to discover if individuals with dyslexia indeed have more difficulties with their L2 than their first language. They conducted a study on the basis of several contradicting reports from English teachers in Sweden. These teachers reported exceptional cases of English (L2) facility in Swedish students with dyslexia. Miller-Guron and Lundberg (2000) hypothesised that there are a number of socio-cultural and emotional factors that can influence early L2 learning, and as a result cause a preference for reading English. Such factors could in turn cancel out some of the negative effects of dyslexia for the L2. The participants of Miller-Guron and Lundberg (2000) were three groups of native Swedish students, whose reading performance was examined. The three groups consisted of students with dyslexia who preferred to read Swedish, students with dyslexia who preferred to read English, and unimpaired readers whose preference was to read in Swedish. The authors investigated the factors that might modulate a reading disadvantage in the L2 (e.g. exposure to English, familiarity with English before print exposure, motivation to read English, etc.) by means of interviews. Firstly, they found that the impaired readers who preferred English had approximately equivalent reading comprehension scores in English and Swedish, whereas the impaired readers who preferred to read in Swedish scored lower on the English comprehension than on the Swedish comprehension. For other English tasks the group that preferred English also outperformed the group that preferred to read in Swedish, indicating a clear difference in L2 reading skills between the two groups of students with dyslexia. Secondly, from the interviews Miller-Guron and Lundberg (2000) concluded that positive early experience with reading in English could have caused an early preference to read English texts as opposed to Swedish texts. This study thus provides evidence that not all individuals with dyslexia feel inhibited in their abilities to become fully proficient in their L2 reading abilities, and that there are modulating factors that can influence reading proficiency in a second language.
The present study
While the above makes it clear that few studies have investigated bilingual dyslexia in adults, even fewer studies have focused specifically on reading comprehension in bilingual dyslexia. However, this is a valuable area of research, because reading comprehension is such an important skill in higher education. Research into the reading and reading comprehension skills of bilingual students with dyslexia may also
provide ideas on how these students can optimise their studying and with it their academic careers. By investigating the reading comprehension skills of Dutch-English students with dyslexia and comparing these to the reading comprehension skills of matched students without dyslexia, it is possible to examine if the students with dyslexia have more difficulties with reading comprehension in English than in Dutch compared to control participants. As such, the primary aim of the present study was to investigate if there were differences in reading comprehension skills between Dutch-speaking students with and without dyslexia, and to examine if reading comprehension skills of students with dyslexia would be more or less affected in their L2 (English) as opposed to their L1 (Dutch). Secondly, possible factors that might influence reading comprehension performance were examined. These included the use of text-to-speech software, print exposure in both languages, prior knowledge about the topics of the reading tests, and language proficiency and vocabulary knowledge in Dutch and English. As such, the following research questions were formulated and investigated in the present study:
1. Is there a difference in reading comprehension performance between the dyslexia group and the matched control group?
a. Is there a difference in performance according to the type of question (inferential or literal) that is asked?
b. Does the use of text-to-speech software positively affect the performance of participants with dyslexia?
2. Is there a difference in Dutch (L1) and English (L2) reading comprehension performance within the dyslexia group and within the control group?
3. To what extent do the factors print exposure, prior knowledge, general language proficiency, and vocabulary knowledge affect the reading comprehension performance in the dyslexia group and the control group?
Research has shown that high-performing students with dyslexia often do not manage to fully compensate for their learning disability (e.g. Deacon et al., 2012). Adults without dyslexia have been found to outperform adults with dyslexia in terms of reading comprehension skills (e.g. Pedersen et al., 2016; Simmons & Singleton, 2000). Thus, for the present study it was hypothesised that the Dutch control group would also outperform the dyslexia group. Furthermore, studies have shown that the performance of individuals with dyslexia can be equal to that of control participants when they are asked literal questions, but that they show a poorer performance when inferential questions are asked (Pedersen et al., 2016; Simmons & Singleton, 2000). This was the basis for the hypothesis that the students with dyslexia would have an accuracy rate that was equal to that of the students without dyslexia when it came to the literal questions in one of the reading comprehension tests in the present study, but that the accuracy rate would be poorer than that of the control group for the inferential questions. The use of text-to-speech software could possibly mitigate decoding difficulties and consequently have a positive
effect on reading comprehension (Forgrave, 2002), which is why it was hypothesised that the students with dyslexia who used reading software would show a better performance than the students with dyslexia who did not use this software.
It was further hypothesised that L2 reading comprehension would be poorer than L1 reading comprehension in the dyslexia group, but not necessarily in the control group. Earlier studies showed that the transparency of the orthography plays an important role in reading performance (e.g. Lindgrén & Laine, 2011). Because English has a much less transparent orthography than Dutch, with a phoneme-grapheme connection that is far more arbitrary, it is likely that this will affect the reading comprehension skills of the students with dyslexia. The students with dyslexia will likely have poorer reading comprehension skills in English. For the control group without reading difficulties, orthographic depth should not be a major determinant for reading comprehension skills. However, it was stressed by other researchers (e.g. Elbro et al., 2011; Hedman, 2012) that language proficiency is an important factor that should be kept in mind when assessing reading and reading comprehension skills. If the proficiency levels in the two languages are not similar, this could cause the difference between reading skills in the L2 and L1, both in the dyslexia group as well as the control group. However, if proficiency levels are approximately similar, then it follows that the opaque orthography of English could result in poorer reading comprehension abilities in the students with dyslexia. Nevertheless, findings of other studies (e.g. Miller-Guron & Lundberg, 2000) also indicated that there are factors which could possibly mitigate this disadvantage. As such, the influence of vocabulary knowledge and print exposure in Dutch and English and the effect of prior knowledge about the topics of the reading tests will be investigated in the present study as well.
Method
The present study aimed to investigate the reading comprehension skills of high-performing Dutch-English students, in order to examine how the reading comprehension skills of students with dyslexia would compare to the reading comprehension skills of students without reading difficulties. Possible factors which could influence reading comprehension skills in English and Dutch were also taken into account. The test battery consisted of a comprehensive questionnaire, language proficiency tests, and tests measuring reading skills and reading comprehension skills.
Participants
Table 1 shows an overview of all the participants and relevant background details, such as age, sex, and education. A total of 36 students participated in the present study. All the participants were native speakers of Dutch enrolled in higher education in the Netherlands. Twenty-three of these students were officially diagnosed with developmental dyslexia. The remaining 13 students did not have dyslexia and
did not report any other reading or spelling difficulties. The control participants without dyslexia were matched on age, sex, and field of study with the participants with dyslexia. All the participants were undergraduate students from various disciplines, who had been recruited from the University of Groningen and the University of Applied Sciences Groningen. Testing and data collection were carried out in the interest of a PhD project at the University of Groningen. The students were given a small financial compensation after participation.
The mean age of the participants with dyslexia was 21.2 (SD = 2.8), and there were 13 females and ten males in this group. Twenty-one students with dyslexia were enrolled at university, whereas two students were following higher education at the university of applied sciences. The mean age of the participants without dyslexia was 20.3 (SD = 1.9). There were nine females and four males in the control group. Only one of the students in this group was enrolled at the university of applied sciences, the others were students at the university of Groningen. Four participants with dyslexia as well as two participants without dyslexia reported other neurological conditions or other difficulties aside from dyslexia. One control participant had epilepsy but was on medication and had not had any seizures in the last ten years. One participant without dyslexia and one participant with dyslexia were diagnosed with ADHD. Two other participants with dyslexia were diagnosed with ADD, and the remaining participant with dyslexia had colour blindness and mild hearing difficulties. The other participants did not report having any neurological conditions or uncorrected visual or hearing difficulties. Most of the students were right-handed, there was one handed control participant and there were three left-handed participants in the dyslexia group. Lastly, there were many participants who reported having family members with reading and/or spelling difficulties or a diagnosis of dyslexia: seven participants in the control group and 15 participants in the dyslexia group had a familial risk for developing dyslexia.
Table 1. Details of the participants with and without dyslexia (N = 36)
Dyslexia (n = 23) Control (n = 13)
Mean age (years) 21.2 (SD = 2.8) 20.3 (SD = 1.9)
Sex (M / F) 10 / 13 4 / 9
Education (U / AS) 21 / 2 12 / 1
Handedness (L / R) 3 / 20 1 / 12
Comorbid disorders ADHD (n = 1), ADD (n = 2),
colour blindness and mild hearing problems (n = 1)
ADHD (n = 1), epilepsy (n = 1)
Familial risk (Yes / No) 15 / 8 7 / 6
Self-reported proficiency Dutch 6.7 (SD = 0.7) 6.7 (SD = 0.9)
Self-reported proficiency English 6.6 (SD = 0.9) 7.0 (SD = 0.9)
Note. Sex = number of male participants (M) / number of female participants (F); Education = number of participants enrolled at university (U) / number of participants enrolled at the university of applied sciences (AS); Handedness = left-handed participants (L) / right-handed participants (R); Familial risk = participants with a familial risk for dyslexia (Yes) / participants without a familial risk for dyslexia (No); Self-reported proficiency = mean grade received for final exams at secondary school, ranging from 1-10.
All participants could be considered to be bilingual or multilingual in the sense that they had had formal instruction in the L2 English during primary and secondary school and came into contact with both English and Dutch in daily life from an early age onwards. The mean self-reported proficiency was the grade that the students had received on their final exams at secondary school, prior to entering higher education. The mean grades were similar between the groups for both Dutch and English. The mean self-reported Dutch proficiency was 6.7 (SD = 0.7) for the students with dyslexia as well as for the students without dyslexia (SD = 0.9). The participants with dyslexia reported a mean grade of 6.6 (SD = 0.9) for their English proficiency, whereas this was 7.0 (SD = 0.9) for the students without dyslexia. A Mann-Whitney U test indicated that there were no significant differences in the self-reported Dutch proficiency (U = 163.5, p = .649) or in the self-reported English proficiency (U = 187.0, p = .226) between the groups with and without dyslexia. Two Wilcoxon Signed Ranks tests within the dyslexia group and within the control group indicated that there were no significant differences between the English proficiency grade and the Dutch proficiency grade within the groups either (dyslexia: Z = -0.24,
p = .813, control: Z = -1.10, p = .271). As such, it could be concluded that the students with and without
dyslexia considered themselves to be approximately equally proficient in English and Dutch.
Material and procedure
Since the present study was a small study within an ongoing PhD project, all participants had to complete a test battery, including tests which will not be discussed here. The duration of the test battery was between 2.5 and 3.5 hours. The tests of interest for the present study included an online test assessing Dutch vocabulary, a short online English vocabulary test, a Dutch and an English C-Test, each consisting of three texts assessing general language proficiency, and finally two Dutch and two English reading comprehension (RC) tests. One RC test for each language was short and about a general topic, whereas the other RC test was slightly longer and concerned an academic text. Dutch and English tests were administered in separate blocks. Between these two blocks there was a break of approximately 15-30 minutes. The order of the blocks (i.e. Dutch first or English first) was alternated between participants.
A comprehensive online questionnaire was also administered prior to the test session in order to discover relevant background details from the participants. The test battery itself was administered in a quiet room with an examiner at the University of Groningen. At the beginning of each session, general information about the test procedure was briefly repeated, after which the participants were required to sign a written consent for the data to be used for research purposes. The subtests were administered in the same order for each participant, although the block order was alternated. A brief overview of all the tasks, the relevant scoring measures, and their purpose for the present study is presented in Table 2. All tests will also be discussed in detail below.
Table 2. Details of the relevant tasks in the test battery
Tasks Task information Scoring Measure
Dutch vocabulary task - 75 items
- No time constraint
- Accuracy (%) L1 vocabulary knowledge
English LexTALE task - 60 items
- No time constraint
- Accuracy (%) - Time
L2 vocabulary knowledge
Dutch C-Test - 3 texts with 60 gaps
- 5 minutes per text
- Accuracy (words) - Time
L1 proficiency
English C-Test - 3 texts with 60 gaps
- 5 minutes per text
- Accuracy (words) - Time
L2 proficiency
Dutch “Sun” task - Short text - Summary or 20 true/false judgements - 4 minutes
- Correct ideas or;
- Accurate T/F judgements - Reading time
- Answering time
L1 reading comprehension
English “Otters” task - Short text - Summary or 20 true/false judgements - 4 minutes
- Correct ideas or;
- Accurate T/F judgements - Reading time
- Answering time
L2 reading comprehension
Dutch “Myths” task - Academic text - 3 open questions - 7 minutes - Correct answers - Reading time - Answering time L1 academic reading comprehension
English “Emotions” task - Academic text
- 3 open questions - 7 minutes - Correct answers - Reading time - Answering time L2 academic reading comprehension Background questionnaire
Preceding the test session, all participants were required to complete the comprehensive background questionnaire. A link to the online questionnaire was sent to the participants several days prior to their test session, and the participants filled out the questionnaire in their own time. Most participants did not require more than 10-15 minutes to complete the full questionnaire, as was indicated by the timestamps in the questionnaire. The participants with dyslexia required more time because they had an extra block of questions which the participants without dyslexia did not have to fill out.
The background questionnaire was in Dutch and consisted of six blocks of questions, with mostly multiple choice questions, questions with Likert scales, and open questions. The first block of questions was related to general information about the participants (i.e. age, education, mother tongue, handedness, etc.). The second block of questions was about dyslexia and other possible related disorders. The questions about dyslexia were only available to the participants who had given a positive answer on a previous question in the first block, which asked about the presence of dyslexia or reading and/or spelling difficulties. The second block further consisted of questions about a diagnosis, family members with comparable problems, and remedial teaching or other forms of extra literacy training. The third block of questions was about studying and computer usage, and among others included questions about the preferred kind of exam and about the computer as a possible reading aid. The fourth block was about (print) exposure and Dutch language proficiency, whereas the fifth block questioned about (print) exposure and language proficiency in English. The final block was included to determine how much
prior knowledge about the topics of the reading comprehension tests the participants had. It consisted of four questions formulated as: how much do you know about…, which the participants could answer by means of a 5-point Likert scale. A translated version of the questionnaire is included in Appendix I.
C-Tests
A C-Test was administered for each language. The C-Test is a variant of the cloze test (Keijzer, 2013). It is an integrative written language test, which can provide a measure of general language proficiency for a language, in this case Dutch and English. For the present study, three texts from the C-Tests developed by Keijzer (2007) were selected for Dutch, and three were selected for English. For each language, the texts were arranged in order of difficulty, with the easiest text being administered first. The texts for Dutch and English were comparable in their levels of difficulty. In each text, the first sentence remained untouched, but the other sentences contained gaps. The participants were never required to complete the gaps by filling in whole words, they only had to fill in parts of the words that were missing. There were 20 blanks in each text that needed to be filled in. An example of a sentence with blanks in one of the Dutch texts is the following: Hij raa… (raakte) depressief, ge… (genas), maar
we… (werd) nooit meer de ou… (oude). For one of the English texts an example is: In Canada, we’re fort… (fortunate) to have pol… (polling) stations wi… (within) a short wa… (walk) or dr… (drive).
For both languages, the participants were given a maximum of five minutes to read and fill in the gaps for each of the three texts. Thus, the maximum amount of time to complete this task was 15 minutes, although the students did not always require this amount of time, especially in the case of the Dutch C-Test. A note was made of the time that the participants required to complete each of the individual texts. If the five minutes were up, the examiner asked the participants to go on to the next text. The time score was the required amount of time in seconds for each text and the required time to complete all three texts (total time score). Since there were 201 gaps in each of the texts, the maximum accuracy score for each of the texts was 20. This accuracy score was the number of correct words. The total accuracy score had a maximum of 60 correct words. Spelling errors were tolerated and words which were not correct according to the answer model but did fit the context of the text equally well were also counted as correct.
Vocabulary tests
The Dutch vocabulary test was an online adaptation of the vocabulary test designed by Vander Beken, Woumans, and Brysbaert (2017). This vocabulary test contains 75 items of various difficulty, and can be used as a general measure of L1 proficiency. Participants were shown an existing Dutch target word (e.g. rups “caterpillar”, patstelling “stalemate”) and they had to decide from four answer alternatives
1 There was an inaccuracy in the first English text, which meant that the participants only had to fill in 19 blanks. All participants were given
which of the four alternatives was the best match for the target word. An example is the target word luit (lute), with the four answer alternatives bouwmateriaal (building material), dier (animal),
keukenapparaat (kitchen appliance), and muziekinstrument (musical instrument). Since the participants
had to pick the word which was the best match for the target word, they had to select the last word in this example, because a lute is a musical instrument.
The Dutch vocabulary task had to be completed on a computer, with instructions also being presented on the screen. The number of correct answers and the percentage of correct answers were used for scoring purposes. There was no time constraint for this task, but participants required approximately 15 minutes to complete the test.
The LexTALE (Lexical Test for Advanced Learners of English) test is a vocabulary test which can be administered online and which can be used to examine the vocabulary knowledge of participants with and advanced level of English. The test was designed by Lemhöfer and Broersma (2012). It is a short lexical decision task of 60 trials, in which participants see a word and have to judge if it is an existing English word by pressing either yes or no. There are 40 existing words of various difficulty (e.g. muddy,
plaintively) and 20 non-words (e.g. mensible) in the test. Lemhöfer and Broersma (2012) conducted a
large-scale study with (among others) Dutch advanced learners of English and determined that the LexTALE can give a valid indication of English vocabulary knowledge. The present study used the LexTALE task as a general measure of L2 proficiency.
The participants were given an on-screen instruction, after which they could start the LexTALE test themselves. Their accuracy scores appeared on the screen immediately after completion of the task. This score was the number of correct judgements that had been made, expressed as a percentage. A note of the required time was also made by the examiner. Although there was no time constraint for this task, on average, the test took approximately five minutes to complete, including reading the instructions.
Reading comprehension tests “Sun” and “Otters”
There were four reading comprehension tests in total. Two tests were in Dutch, and two tests were in English. In both cases, there was one shorter text with a topic in the domain of natural sciences, and one academic text which described the results of a psychological research experiment. The short tests will be discussed first.
The Dutch “Sun” text listing several facts about the sun was originally taken from the study of Roediger and Karpicke (2006) and translated into Dutch by Vander Beken and Brysbaert (2017). Because the translated version by Vander Beken and Brysbaert (2017) contained some Flemish sentence constructions which are not common for Dutch in the Netherlands, this version was slightly adapted for the present study. The short text for English, “Otters”, described several general facts about sea otters and was also used by Roediger and Karpicke (2006) and Vander Beken and Brysbaert (2017) in their studies. The texts were of approximately the same length (Dutch: 249 words; English: 279 words) and had a comparable difficulty level. The texts were presented on paper in the Times New Roman font,
with a letter size of 12 and a line-spacing of 1.5. The test that accompanied these texts was either a truth-value judgement task or a free recall task (summary); this was alternated between participants. For the free recall task the participants were required to write a summary, with the instruction to be as detailed as possible in their writing. This test format was similar to that of Roediger and Karpicke (2006), who divided the texts into 30 “ideas” that could be reproduced in the summary. The truth-value judgement task consisted of 20 questions and was based on the 46 true/false questions developed by Vander Beken and Brysbaert (2017). In the Dutch truth-value judgement task there were 13 literal questions and seven inferential questions, the English counterpart included 16 literal questions and four inferential questions. Participants were given a total of four minutes to read and study the texts. They were asked to raise their hand when they had read the entire text for the first time, so a note could be made of the required reading time (reading time score) as well as the number of words read per minute. Since these were short texts, all of the participants had finished reading the texts at least once before the time was up. As such, they could use the remaining time to reread the text or parts of it. When the four minutes were up, the text was removed and the participants either had to complete the truth-value judgement task or the free recall task. Before receiving the text, they were told which one of the two tasks they would have to complete. The participants were not allowed to refer back to the text when answering the questions or writing the summary. A note was made of the time that participants required to answer the questions or to write the summary (answering time score) although there was no time constraint. For the truth-value judgement task there was a maximum score of 20. For the summary the total number of correct ideas was evaluated, with a maximum score of 30 correct ideas. Regardless of test type, the entire test took approximately ten minutes in total to complete. This was the case for both the Dutch “Sun” test and the English “Otters” test.
Academic reading comprehension tests “Myths” and “Emotions”
The longer academic text in Dutch (“Myths”) was about myths about subconscious cognitive processing and discussed experiments that implemented subliminal messages. The English academic text (“Emotions”) was about emotion perception and discussed experiments that implemented semantic priming. Both texts were excerpts taken from a Dutch psychology book by Brysbaert (2014); the academic text for the English test was translated into English. Once again, the texts for each language had a comparable difficulty level and approximately the same length (Dutch: 543 words; English: 488 words). These texts were presented on a computer screen, in Times New Roman font, with letter size 12 and with a line-spacing of 1.5. Three open questions were formulated based on the texts. For each language, there was one reproductive question, and two inferential questions that required more applied reasoning; that is, the answers to these questions could not be found directly in the text. Half of the participants read the Dutch text with the aid of text-to-speech software called ClaroRead (2005), the other half read the English text with the reading software. Whether the participants read the Dutch text or the English equivalent with the reading software was alternated between participants. ClaroRead
(2005) is a programme which allows the text to be read aloud by a speaker (Dutch: Claire; English:
Daniel). The participants were allowed to adjust the reading speed to their own preference. In order to
set this up appropriately, those who used the software were given an example text (a cooking recipe) first, so they could experience how the software worked and how to adjust the speed to their own preference. They were given a few minutes to try this, after which they had to read the academic text. The participants who did not read the text with the software were not given the example text first.
All participants were given seven minutes in total to read the Dutch “Myths” and the English “Emotions” texts. As with the shorter texts, they were asked to raise their hand when they had read the text in its entirety for the first time, so a note could be made of the required reading time (reading time score) and the number of words read per minute. This was done in order to investigate to what extent students with and without dyslexia had adapted the reading speed when using the software. After reading the full text once, they were allowed to read parts of the text or the entire text again until the time was up. Those who read the text with the reading software were required to use this software when rereading parts of the text as well. When the seven minutes were up, the document with the text was closed and the participants were given the three open questions which tested their comprehension of the text. As such, they were not allowed to refer back to the text when answering the questions. A note was also made of the time that was required to answer these questions (answering time score), although there was no time constraint for this part of the task. A total of five points could be earned for the reading comprehension questions, one point for the reproductive question and two points for each of the inferential questions. For the Dutch as well as the English text, the entire task took approximately 15 minutes to complete.
Analyses
The analyses for the present study consisted of the following statistical tests: Mann-Whitney U tests to investigate if there was a significant difference in the performance on all the tasks between the control group and the dyslexia group; Wilcoxon Signed Ranks tests to examine within-group differences between the English and Dutch reading comprehension tasks; and correlational analyses (Spearman’s
rho) to investigate which factors influenced the performance of the participants on the reading
comprehension tasks. The effect size r was also calculated.
There were three factors that were investigated with correlational analyses. The reading comprehension scores were first correlated with the prior knowledge about each of the topics of the four tests. The prior knowledge was a number ranging from 1-5, indicating how much the participants knew about the topics of the “Sun”, “Otters”, “Myths”, and “Emotions” texts. Secondly, the reading comprehension scores were correlated with the language proficiency scores: the C-Test scores, vocabulary scores, and the self-reported proficiency. Lastly, correlations between print exposure and reading comprehension were investigated. In order to do this, five questions were selected from the
background questionnaire to form a general idea of print exposure for Dutch and English (see Appendix I). Each of the questions had five options ranging from a very low amount of print exposure to a very high amount of print exposure. These options were scored from 1-5, which meant that the maximum print exposure score for both Dutch and English was 25. This total print exposure score was then correlated with the reading comprehension scores. An alpha-value of .05 was considered to be the significance level for all statistical tests that were conducted.
Results
Language proficiency
One of the aims of the present study was to investigate the effects of language proficiency on reading comprehension in Dutch and English. The general language proficiency of the participants was measured by means of C-Tests in both languages. The participants also gave a self-reported proficiency level, which was the mean grade (1-10) that they had received at secondary school for their final exams for Dutch and English, before entering higher education. The mean scores obtained in the Dutch and English C-Tests and the self-reported proficiency are presented in Table 3. As was already mentioned in the participants section, there were no significant differences between the groups for the Dutch or English self-reported proficiency, nor were there within-group differences between the self-reported proficiency for the two languages.
Table 3. Mean self-reported proficiency and mean scores of the C-Tests for Dutch and English
Dutch English Dyslexia (n = 23) Control (n = 13) Dyslexia (n = 23) Control (n = 13)
Mean SD Mean SD p U r Mean SD Mean SD p U r
Self-rep. proficiency 6.7 0.7 6.7 0.9 .649 67.5 0.08 6.6 0.9 7.0 0.9 .226 187.0 0.22
C-Test 1 score 17.1 1.5 17.8 1.8 .123 197.0 0.27 7.7* 2.8 10.4 3.5 .034 198.5 0.36 C-Test 1 score % 85.7 7.6 88.9 8.9 .123 197.0 0.27 38.5* 14.2 52.1 17.4 .034 198.5 0.36 C-Test 1 time (secs) 150.0 52.5 142.0 49.2 .795 141.5 -0.04 257.6 59.4 243.3 59.9 .548 120.0 -0.11 C-Test 2 score 14.4 2.2 15.9 2.7 .087 202.0 0.29 13.0 3.4 13.6 1.9 .719 148.5 0.06 C-Test 2 score % 71.7 10.9 79.2 13.4 .087 202.0 0.29 65.2 16.8 67.9 9.6 .719 148.5 0.06 C-Test 2 time (secs) 211.0 59.0 194.9 71.6 .454 126.0 -0.13 250.6 50.5 229.6 61.1 .327 109.5 -0.17 C-Test 3 score 13.0 3.2 14.7 3.6 .107 199.0 0.27 9.3 3.9 11.8 4.2 .151 179.5 0.25 C-Test 3 score % 65.2 16.0 73.5 18.2 .107 199.0 0.27 46.5 19.5 59.2 20.8 .151 179.5 0.25 C-Test 3 time (secs) 252.0 52.7 252.4 61.9 .871 155.0 0.03 244.0 55.9 232.5 61.5 .619 123.5 -0.09
Total C-Test score 44.5 5.2 48.4 6.1 .060 207.0 0.32 30.0 8.8 35.0 8.7 .161 178.5 0.24
Total C-Test score % 74.1 8.7 80.7 10.2 .060 207.0 0.32 50.0 14.6 58.4 14.6 .161 178.5 0.24
Total C-Test time (secs) 617.4 136.4 589.3 151.2 .721 138.5 -0.06 752.2 149.8 697.0 160.4 .381 112.0 -0.16 Note. Self-rep. proficiency = self-reported proficiency, i.e. the grade (1-10) that the participants had received for their final exams at secondary
school; secs = seconds; For each text of the Dutch and English C-Tests (1, 2, and 3) there was a maximum accuracy score of 20 (100%) and a maximum time score of 300 seconds; There was a maximum total accuracy score of 60 (100%) and a maximum total time score of 900 seconds for both the Dutch and English C-Test.
As indicated by the results of Mann-Whitney U tests, there was no significant difference between the two groups for the time that was required to complete the full Dutch C-Test (U = 138.5, p = .721). The difference between the total accuracy scores of the two groups seemed to approach significance (U = 207.0, p = .060, r = 0.32). The students with dyslexia had a mean accuracy score of 44.5 (SD = 5.2), or 74.1 % correct, whereas the control group had a mean score of 48.4 (SD = 6.1), or 80.7 % correct. For the English C-Test, there was no significant differences between the two groups in the total accuracy score either (U = 178.5, p = .161), even though there was a significant difference for the accuracy score on the first text of the English C-Test (U = 198.5, p < .034, r = 0.36). There was no significant difference between the two groups for the amount of time that they required to complete the full C-Test (U = 112.0,
p = .381).
For both groups there were within-group differences between the scores for the Dutch C-Test and the English C-Test (see Table 4). For the group with dyslexia, the Wilcoxon Signed Ranks test indicated significant differences in both the total accuracy score or accuracy percentage of the Dutch and English tests (Z = -4.20, p < .001, r = -0.62) and the time that the students with dyslexia required to complete the tasks (Z = -3.36, p < .001, r = -0.50). This was also the case for the group of students without dyslexia (accuracy: Z = -3.06, p < .002, r = -0.63, time: Z = -2.51, p < .012, r = -0.51). Both groups had a higher accuracy score for the Dutch test, and required more time for the English C-Test; the effect sizes were large in all cases.
Table 4. Mean scores of the Dutch and English C-Tests and self-reported proficiency
Dyslexia (n = 23) Control (n = 13)
Dutch English Dutch English
Mean SD Mean SD p Z r Mean SD Mean SD p Z r
Self-rep. proficiency 6.7 0.7 6.7 0.9 .813 -0.24 -0.03 6.6 0.9 7.0 0.9 .271 -1.10 -0.22
Total C-Test score 44.5* 5.2 48.4 6.1 .001 -4.20 -0.62 30.0* 8.8 35.0 8.7 .002 -3.06 -0.63
Total C-Test score % 74.1* 8.7 80.7 10.2 .001 -4.20 -0.62 50.0* 14.6 58.4 14.6 .002 -3.06 -0.63
Total C-Test time (secs) 617.4* 136.4 589.3 151.2 .001 -3.36 -0.50 752.2* 149.8 697.0 160.4 .012 -2.51 -0.51
Note. Self-rep. proficiency = self-reported proficiency, i.e. the grade (1-10) that the participants had received for their final exams at secondary
school; secs = seconds; There was a maximum total accuracy score of 60 (100%) and a maximum total time score of 900 seconds for both the Dutch and English C-Test.
* Significant difference compared to the mean score of the English test at the p < .05 level.
Vocabulary knowledge
Dutch vocabulary knowledge was measured with a 75-item multiple choice test administered on a computer. General English vocabulary knowledge was measured with the online LexTALE task. The mean scores for both tasks are presented in Table 5. The mean score for the Dutch vocabulary task was 44.0 (SD = 7.4), or 58.5 % for the students with dyslexia. The students without dyslexia had a slightly higher score of 48.2 (SD = 9.8), or 64.5 % correct. The difference between the two groups was not significant according to the results of a Mann-Whitney U test: U = 191.5, p = .169.
