Reading difficulties in a bilingual context: a diagnostic dilemma

(1)

Reading difficulties in a

bilingual context

A diagnostic dilemma

Liza Mossing Holsteijn

a thesis submitted to the Faculty of Humanities in partial fulfillment of the requirements of the degree

Master of Arts in Linguistics at University of Amsterdam 23 June, 2015 Supervisor: dr. Judith Rispens

(2)

ii

Acknowledgements

I would like to express my sincere gratitude to the people who helped me in the preparation, writing, and completion of this thesis. First of all, I would like to express my appreciation to dr. Judith Rispens for her expertise, her support, her effective feedback, and her willingness to assist me in dissolving the difficulties I encountered. Secondly, I would like to express my appreciation to dr. Jurgen Tijms for the opportunity to be working on this research project, and his assistance in organising and executing this research project. I would also like to express my gratitude to drs. Sebastián Aravena for familiarising me with his dynamic assessment of the acquisition of grapheme-phoneme correspondences in a novel script, and his willingness to lend me data from one of his prior studies.

Of equal importance was StudieMax, a public service that provides effective supplemental education for primary education pupils, where all the bilingual children who participated in the current study were recruited. Special thanks are due to Rasmus Polanen for his cooperation in this research project and the fruitful brainstorming sessions, not forgetting the numerous interns who helped in collecting the data of the bilingual children, and, moreover, the bilingual children themselves for participating in this research project.

Finally, I wish to thank my family, friends, and partner for their unceasing support and encouragement throughout this venture.

(3)

iii

Abstract

The research that was conducted in the present study was driven by a concern that the identification of a specific reading impairment, such as developmental dyslexia (hereafter referred to as DD), in bilingual children gives rise to a diagnostic dilemma. This diagnostic predicament particularly arises from the difficulty of differentiating between the effects of DD, on the one hand, and bilingual effects, on the other. Recently, it has been indicated that a dynamic assessment (DA) may be a possible remedy to this dilemma. Hence, the current study assesses the potential of a DA for identifying DD in Dutch bilingual children and its relation to a bilingual effect, that is, vocabulary level. A total number of 76 children participated in the current study. They were subdivided into two groups of native Dutch children who were either diagnosed with DD or who had average or above average reading skills, and two groups of Dutch bilingual children who either had below average or poor reading abilities or average or above average reading abilities. All the children in this study received a twenty-minute computerised training in eight novel grapheme-phoneme pairs. Subsequently, both a grapheme-phoneme identification task and a time-limited reading task within the artificial orthography were administered, along with a standardised task of expressive vocabulary. Results from the current study indicated that the DA offered no potential for identifying DD in Dutch bilingual children. The vocabulary level of Dutch bilingual children, however, did not relate to the DA, which speaks in favour of the DA. Thus, on the one hand the current study revealed that the DA offered no potential for detecting DD in Dutch bilingual children, but on the other, it was demonstrated that the DA offers excellent potential for disentangling bilingual effects, such as vocabulary level, from reading ability.

Keywords: developmental dyslexia, reading difficulties, bilingualism, dynamic assessment,

(4)

iv

1 Introduction

The research that was conducted in the present study was driven by a concern that the identification of a specific reading impairment in bilingual children, such as developmental dyslexia (hereafter referred to as DD), has been a diagnostic dilemma for researchers, practitioners, and educators, for instance. This diagnostic predicament particularly arises from the difficulty of differentiating between the effects of DD, on the one hand, and bilingual effects, on the other. For instance, poor performance on a reading test may be caused by more general poor language abilities in the second language (i.e. bilingual effects), rather than by a specific reading impairment per se. To illustrate, it has been observed that bilingual children typically have a smaller vocabulary size in either of their languages than do monolingual speakers of that same language (e.g. Bialystok, Luk, Peets, & Yang, 2010; Hoff et al., 2012). It has also been reported that the ability to read words closely relates to level of vocabulary (Ricketts, Nation, & Bishop, 2007). This means that bilingual children’s reduced vocabulary size in either of their languages may manifest itself in reading difficulties. In other words, a relatively low level of vocabulary may have a detrimental effect on the reading ability of bilingual children (note that references to reading intend technical reading). The diagnostic assessment of reading abilities in bilingual children, therefore, ought to be largely insensitive to these bilingual effects, such as vocabulary level. It has been revealed, however, that standard reading measures are, to a great extent, dependent on second language abilities, like vocabulary level (Elbro, Daugaard, & Gellert, 2012). Consequently, these standard reading measures may diagnose a bilingual child with DD, whereas in fact his/her poor reading abilities may stem from his/her limited vocabulary size, for instance, rather than from a specific reading disability, like DD. This may lead to an overestimation of DD in bilingual children. Hence, there is an urgent need for proper diagnostic assessment of reading abilities in bilingual children that reduces the influence of bilingual effects. The current study attempted to meet this demand.

In the first place, the current study will provide a sufficient amount of background knowledge, that is, information that is essential to understanding the position of reading difficulties in a bilingual context. First, the current study will aim to define DD by specifying its manifestations and by discussing major theories about the impairment that may underlie DD. Second, the current study will elaborate more on the diagnostic dilemma as regards the identification of a specific reading impairment (i.e. DD) in bilingual children, subsequent to a

(6)

2

detailed discussion on the diagnostic dilemma as regards the detection of SLI in a bilingual context, which will function as the starting point for the diagnostic dilemma as regards bilingual DD. Subsequently, the research questions and hypotheses that naturally proceed from the literature will be specified, followed by a section on the method that was employed in the current study. The ensuing chapters will demonstrate the findings of the current study, and will carefully consider the results successively.

(7)

3

2 Developmental dyslexia

DD is a specific learning disability that is neurobiological in origin (Lyon, Shaywitz, & Shaywitz, 2003; Snowling, 2013; Vellutino, Fletcher, Snowling, & Scanlon, 2004), and it has been reasoned that DD has a genetic basis (Pennington & Olson, 2005; Vellutino et al., 2004; Blomert, 2005; Scerri, Schulte-Körne, & Schulte-Körne, 2010). The prevalence rate of DD greatly varies between three and ten per cent, depending on the language and the assessment criteria that are used (Blomert, 2005). The symptoms of DD typically manifest themselves in severe reading difficulties, although it has to be taken into account that additional motor and/or sensory disorders may co-occur in a portion of individuals with DD (Ramus et al., 2003). The reading difficulties are typically characterised by persistent reading disfluencies and/or inaccuracies, meaning that adequate instruction and considerable remedial education are of little to no avail (Lyon et al., 2003; Shaywitz & Shaywitz, 2005). The findings of a number of early intervention studies evidently suggested that children who were prone to DD did not respond to intervention programs in spite of intensive instruction by well-educated instructors (Torgesen, 2000). This lack of response to profound instruction in early intervention programs appears to be an essential factor that differentiates between reading failure as a result from insufficient instruction, on the one hand, and reading failure resulting from DD, on the other hand. Hence, DD cannot be explained in terms of inadequate instruction. In addition, it has been observed that DD cannot be ascribed to low cognitive and academic abilities (Lyon et al., 2003; Shaywitz & Shaywitz, 2005). Generally, this means that there is a discrepancy between the reading age and the chronological age and/or between the reading ability and the educational attainment of an individual with DD (Lyon et al., 2003).

The question arises as to what the cause may be of the persistent reading difficulties that have been observed in individuals with DD. Over the years, a number of theories as regards the underlying impairment of DD have been proposed, of which the major theories will be reviewed here. These theories explain the reading impairment in DD on three levels, namely on biological/neurological level, on cognitive level, and on behavioural level, which may interact with each other. Prior to discussing these theories, however, it is necessary to place them in the context of reading. By way of illustration, it has been assumed that, on neurological level, the cerebellum, and the magnocellular cells, among others, play a vital role in the ability to read successfully. They are responsible for the automatisation of the reading process and the identification of the visual word form respectively, which make them

(8)

4

necessary conditions for becoming a successful reader. In addition, particular cognitive systems have also been deemed essential for becoming a proficient reader. The phonological system in particular has been assumed to play a crucial role in the ability to read. More specifically, phonological awareness, which involves the segmentation and manipulation of speech sounds, has been suggested to be critical for reading ability. For example, the ability to segment and identify the various speech sounds in a word is necessary for understanding the graphemic representation of a speech sound. In other words, phonological awareness is an essential prerequisite for the establishment of letter-speech sound correspondences. In turn, the establishment of grapheme-phoneme associations has been regarded as the fundamental mechanism that underlies successful reading, and is typically manifested on behavioural level. That is, proper grapheme-phoneme binding manifests itself on behavioural level by sufficient reading fluency. Taking this into account, it seems that certain neurological, cognitive, and behavioural mechanisms play a major role in reading ability. These mechanisms, however, have often been reported to be impaired in individuals with DD, consequently resulting in their persistent reading difficulties.

2.1 Biological/neurological theories of developmental dyslexia

To begin with, the cerebellar theory of DD explains the specific learning disability in neurological terms. The theory postulates a “mildly dysfunctional” cerebellum in individuals with DD, which has been assumed to underlie the reading difficulties in DD (Ramus et al., 2003). By way of illustration, it has been suggested that the cerebellum plays an important role in motor control, and, consequently, in articulation, which provides an explanation for the motor problems that may occur in individuals with DD (Ramus et al., 2003). Moreover, it has been proposed that the cerebellum also plays a role in the automatisation of autonomous processes, such as driving, typing, and, more importantly, reading (Ramus et al., 2003). Consequently, a mildly dysfunctional cerebellum in individuals with DD may interfere with the automatisation of grapheme-phoneme mapping, resulting in reading difficulties. Convincing evidence for this theory comes from the findings of brain imaging studies, which demonstrate differences in activation as regards the cerebellum of individuals with DD (Nicolson, Fawcett, & Dean, 2001). Figure 1 depicts the brain regions in individuals with DD that show considerably less activation compared with typically developing controls.

(9)

5

Figure 1. The right hemisphere of the cerebellum (a) and the cerebellar vermis (b) in individuals with DD,

which show considerably less activation compared with typically developing controls (Nicolson, Fawcett, & Dean, 2001).

A major pitfall of the cerebellar theory, however, is that it fails to account for the absence of motor dysfunctions in individuals with DD, and, furthermore, it does not account for the sensory problems that have been observed in a portion of individuals with DD (Ramus et al., 2003).

A second major neurological theory of DD is the magnocellular theory, which does account for both the presence of sensory problems, and the absence of motor disorders in individuals with DD. The magnocellular system is responsible for identifying visual motion and detecting changes in auditory cues (Ramus et al., 2003). The magnocellular theory assumes an impaired development of the magnocellular neurones in the brain of individuals with DD, either sensory or motor. That is, this theory postulates a general magnocellular defect, which, consequently, affects sensory (i.e. visual and auditory) and motor modalities (Ramus et al., 2003). As regards the reading difficulties that have been observed in individuals with DD, the magnocellular theory suggests a causal connection between a dysfunction in the visual magnocellular system and these reading difficulties (Stein & Talcott, 1999). By way of explanation, it has been postulated that the magnocellular system is important for the visual function in humans, and, hence, for identifying the visual form of a word. Therefore, it has been reasoned that an impaired development of the visual magnocellular system affects the reading ability in individuals with DD. This neurological claim has been supported by post-mortem examinations of the brains of individuals with DD (Stein, 2001; Livingstone, Rosen, Drislane, & Galaburda, 1991; Galaburda & Livingstone, 1993). It has been observed that the magnocellular layers of the lateral geniculate nucleus (LGN) of the thalamus in individuals with DD is disrupted, and it has been found that the magnocellular neurones in the brains of individuals with DD are approximately thirty per cent smaller in area than in the brains of typically developing controls (Figure 2) (Galaburda & Livingstone, 1993).

(10)

6

Figure 2. The magnocellular cells in the LGN of the brain in individuals with DD (encircled in red) appear to be

smaller and more variable as regards size and shape in comparison with the cells in the brains of controls. Also, they appear to be more disrupted (Galaburda & Livingstone, 1993).

Although the magnocellular theory provides an account for the sensory symptoms, i.e. visual and auditory impairments, that may co-occur in individuals with DD, a thing in which the cerebellar theory does not succeed is that it fails to explain the absence of these auditory and visual disorders in a portion of individuals with DD (Ramus et al., 2003). Hence, the unique ability of the magnocellular theory to account for all symptoms of DD is devalued by its inability to account for the absence of these symptoms in particular groups of individuals with DD.

2.2 Cognitive theory of developmental dyslexia

Contrary to the cerebellar and the magnocellular theory of DD, which postulate neurological claims concerning the origin of the reading impairment that has been observed in DD, the phonological theory attempts to explain DD on a cognitive level. The phonological theory of DD posits the idea that a deficit in the phonological system underlies the reading impairment in individuals with DD (Ramus et al., 2003). More specifically, this theory postulates that individuals with DD have a specific impairment in the processing of speech sounds. Although the exact nature of DD is still under considerable debate, the idea that DD stems from a phonological deficit is a prevailing view. On a neurobiological level, this idea has been supported by findings of functional magnetic resonance imaging (fMRI) studies. During the execution of phonological tasks, such as the repetition of non-words, significant differences have been found as regards the activation of particular brain regions in individuals with DD compared with non-impaired controls (Shaywitz et al., 2002). For instance, it has been

(11)

7

demonstrated that the left hemisphere posterior brain region in individuals with DD is considerably less activated in comparison to the brain activation patterns in non-impaired controls, that is, during the execution of phonological tasks (Shaywitz et al., 2002). On behavioural level, further evidence for the phonological theory of DD comes from a poor performance of individuals with DD on tasks that are related to phonological abilities. Phonological abilities of individuals with DD have typically been measured by tasks that are concerned with rapid automised naming (RAN), phonological short-term memory (pSTM), and phonological awareness. RAN relates to the rapid naming of visual stimuli, such as colours, pictures, numbers, and letters, which addresses the retrieval of phonological information from the mental lexicon (Norton & Wolf, 2012). By way of illustration, Figure 3 shows a stimulus card of a RAN subtest that was developed by Denckla and Rudel (1976), which is concerned with the rapid naming of letters. Hereby, the subjects had to name the visual stimuli, i.e. letters, as rapidly as possible.

Figure 3. A RAN stimulus card developed by Denckla and Rudel (1976).

Research has demonstrated that individuals with DD perform substantially poorly on RAN tasks in comparison to typically developing controls, i.e. they are typically slower in retrieving phonological information during RAN tests (Denckla & Rudel, 1976). What is more, it has been suggested that RAN tests can differentiate between individuals with DD and individuals without DD (Denckla & Rudel, 1976). Hence, the ability to rapidly retrieve phonological information from the lexicon in individuals with DD has been indicated to be impaired. Furthermore, it has been assumed that individuals with DD have a poor pSTM, or, rather, phonological loop. The phonological loop is one of the components of the four-component working memory model that was proposed by Baddeley (2000), which is shown in Figure 4.

(12)

8

Figure 4. Baddeley’s (2000) four-component model of working memory.

Initially, it has been proposed that the phonological loop has two major features. Namely, the phonological loop has been assumed to comprise a “phonological store,” which only temporarily holds phonological information, and a “rehearsal process,” which can retain the decaying phonological information in the phonological store by means of verbal/subvocal rehearsal (Baddeley, Gathercole, & Papagno, 1998). Recently, it has been suggested that the phonological loop also plays an important role in language learning, or, rather, vocabulary learning (Baddeley, 2010; Baddeley et al., 1998). The functioning of the phonological loop can be measured by tasks that address temporary storing and recalling auditory stimuli, such as the repetition of non-words. A non-word repetition (NWR) task assesses the capacity of the phonological loop by examining the ability to repeat pseudowords, such as luubuf /lybʏf/ (Rispens & Baker, 2012). Although research that directly assesses NWR in individuals diagnosed with DD is scarce, it has been demonstrated that individuals that are reading impaired have significantly more difficulty in repeating such non-words than their typically developing peers, which implies a reduced capacity regarding the phonological loop in reading impaired individuals (de Bree, Rispens, & Gerrits, 2007). Moreover, it has been demonstrated that individuals with DD have a lack of phonological awareness (Fawcett & Nicolson, 1995). Phonological awareness encompasses the segmentation and manipulation of speech sounds, which can be assessed by tasks that involve sound categorisation tasks, such as rhyming, alliteration, and phoneme deletion, among others. The results of studies that are concerned with phonological awareness in individuals with DD demonstrate that the performance of these individuals is inferior to that of typically developing controls (Fawcett & Nicolson, 1995). In other words, research has been suggested that individuals with DD

(13)

9

encounter persistent problems with respect to phonological awareness tasks, i.e. their phonological awareness has been assumed to be impaired.

These disrupted phonological abilities of individuals with DD, their poor phonological awareness in particular, have been assumed to impede the establishment of proper grapheme-phoneme correspondences, which has been considered to be an essential prerequisite for the ability to read an alphabetic system (Peterson & Pennington, 2012). Consequently, this may lead to the reading difficulties that are widely observed in individuals with DD. In this fashion, the phonological theory of DD seems to provide a direct relation between the underlying cognitive deficit and the reading problems that are typically manifested in DD. Recently, however, it has been suggested that the poor acquisition of letter-speech sound correspondences in individuals with DD cannot directly be ascribed to their disrupted phonological abilities (Blomert & Willems, 2010). By way of illustration, McNorgan, Randazzo-Wagner, and Booth (2013) found that grapheme-phoneme mapping did not depend on phonological abilities in individuals with DD. Rather, grapheme-phoneme mapping is related to phonological awareness merely in typically developing controls. This suggests that an isolated disrupted integration of orthographic and phonological information may underlie the reading difficulties that have been observed in individuals with DD. Hence, it has been reasoned that the reading difficulties that are manifested in DD may be explained by a grapheme-phoneme binding deficit on behavioural level.

2.3 Behavioural theory of developmental dyslexia

Successful grapheme-phoneme binding typically involves the integration of different modalities, namely a visual modality that is concerned with the processing of orthographic information, and an auditory modality that is engaged with the processing of phonological information. Hence, grapheme-phoneme binding has been considered to be a cross-modal process. By way of illustration, the cortical network that is typically implicated in grapheme-phoneme binding generally comprises the fusiform gyrus (FG), the posterior superior temporal gyrus (pSTG), the posterior superior temporal sulcus (pSTS), and heteromodal regions, such as the supramarginal and angular gyri (SMG and AG, respectively), among others (McNorgan et al., 2013; Booth et al., 2002). It has been demonstrated that a portion of the FG is particularly engaged with the processing of orthographic information (McNorgan et al., 2013; Booth et al., 2002; Dehaene & Cohen, 2011). McCandliss, Cohen, and Dehaene (2003) proposed to name this particular region the Visual Word Form Area (VWFA). It has

(14)

10

been found that a portion of the FG is specifically activated during the execution of tasks that unimodally involve the visual word form (Booth et al., 2002). More convincing evidence comes from studies that are concerned with pure alexia. Pure alexia has been attributed to a lesion in the VWFA, which, accordingly, should lead to impaired recognition of the visual word form. Indeed, pure alexia manifests itself in impaired word recognition, i.e. the loss of the ability to correctly identify letter strings (McCandliss et al., 2003), which strongly suggests that the FG is involved in the visual word form. In addition, it has been demonstrated that the pSTG particularly shows activation during tasks that are unimodally engaged with the spoken word form (Booth et al., 2002), which suggests that the pSTG is particularly involved in the processing of phonological information. Furthermore, it has been reported that the pSTS is involved in the integration of both orthographic and phonological information (McNorgan et al., 2013). This has been supported by the findings of van Atteveldt, Roebroeck, and Goebel (2009), who found that this region is specifically activated during bimodal tasks, i.e. tasks that address both orthographic and phonological processing. Moreover, it has been found that the SMG and AG are typically activated during the execution of tasks that involve the conversion between orthographic and phonological representations, e.g. rhyming tasks that contain the visual word forms (Booth et al., 2002). Note that the conversion from visual forms into auditory forms is fundamental to reading (Snowling, 1980). This conversion from orthography into phonology, however, appears to be disrupted in individuals with DD (Fox, 1994; Snowling, 1980). Snowling (1980) suggested that the development of grapheme-phoneme conversion in individuals with DD stagnates in comparison with their typically developing peers, whose conversion ability increases with age. The fact that Fox (1994) found that individuals with DD are most prone to making errors in tasks that require grapheme-phoneme conversion substantiates this assumption. By way of illustration, Fox (1994) employed a “visual presentation – auditory recognition (V –A)” task, which means that the subject is firstly presented with a visual word form, followed directly by an auditory stimulus, after which the subject is required to decide whether or not the visual and auditory stimulus are identical. In order to be able to recognise the auditory stimulus as being identical to the visual stimulus, the visual word form needs to be converted into the auditory form. That is, the graphemes in the visual form ought to be converted into phonemes. The findings by Fox (1994) demonstrate that individuals with DD particularly encounter significant difficulties in V – A tasks. Hence, it has been postulated that individuals with DD have a specific deficit in grapheme-phoneme conversion, i.e. the associations between graphemes and their corresponding phonemes are disrupted. More convincing evidence comes

(15)

11

from fMRI studies, which indicate differences with respect to the activation patterns of brain regions in individuals with DD that are involved in grapheme-phoneme binding compared to the activation patterns in typical readers. For instance, individuals with DD demonstrate weaker activity in the FG as regards the processing of graphemes, and in the pSTG that is concerned with the processing of phonemes (Blau et al., 2010). The findings of electrophysiological studies of the brain mechanisms in individuals with DD during symbol-sound mapping tasks also imply a deficit in processing and integrating visual and auditory information (Widmann et al., 2012). For example, it has been found that the event-related potential (ERP) components in individuals with DD that are typically concerned with the congruence between stimuli, which are visual and auditory stimuli in this case, show a delay in onset, have a smaller amplitude in comparison with the ERPs in typical readers, or are even absent (Widmann et al., 2012).

What is more, a grapheme-phoneme binding deficit is discernable on behavioural level in individuals with DD. Aravena, Snellings, Tijms, and van der Molen (to appear) developed a dynamic assessment (hereafter referred to as DA) that assessed the acquisition of grapheme-phoneme correspondences in Dutch native children with DD (N = 72, M age = 9.26 years) and their TD peers (N = 46, M age = 9.37 years) with the aim of deploying their DA as a diagnostic assessment tool of DD. Their DA consisted of three parts, namely a computerised training session in which eight novel grapheme-phoneme associations were taught to the participants, which was followed by two assessment tasks (i.e. a grapheme-phoneme identification task and a time-limited reading task within the novel script). The training session of Aravena et al. (to appear) was a learning task in which pairing novel letter-speech sound correspondences played a central role. More specifically, the computerised training session directly addressed the participants’ ability to acquire novel grapheme-phoneme correspondences within an artificial script. The Hebrew script was adopted as the artificial script to capture the characteristics of letters as they naturally exist. The novelty of the grapheme-phoneme associations facilitated a complete control over differences in exposure to the concerned letter-speech sound correspondences on condition that the participants were not previously exposed to the Hebrew script. Hence, this excluded possible external influences on the participants’ performance to a great extent. Furthermore, the training phase had a total duration of twenty minutes. It has been observed that difficulties in grapheme-phoneme binding already manifest themselves after a training session of thirty minutes, and, moreover, it has been indicated that a training phase of only twenty minutes was able to provide ample exposure to the experimental stimuli, that is, to the novel grapheme-phoneme associations

(16)

12

(Aravena et al., to appear). Hence, a training phase of twenty minutes was adopted. The grapheme-phoneme identification task within the artificial script was administered subsequent to the training session, followed by the time-limited reading task. Within these tasks, the participants had to attach the speech sounds to their corresponding letters and were asked to read 22 words that were written in the artificial script respectively. Both the speed and the accuracy of the participants’ ability to associate novel graphemes with their corresponding phonemes and their ability to read words within the artificial script were examined in these tasks. Aravena et al. (to appear) incorporated a speed measure in the identification task to obtain a high level of sensitivity in this task, even if the participants’ performance on the accuracy measure would reach ceiling levels. The results of the letter-speech sound binding task show that individuals with DD perform significantly poorer than their peers without DD with respect to accuracy and speed. Additionally, the number of words that are read correctly by individuals with DD in the word reading task is considerably smaller than the number of words that are read successfully by individuals without DD. Moreover, it was demonstrated that the grapheme-phoneme identification task correctly classified no less than 76.6 per cent of the children who were diagnosed with DD. These findings provide empirical evidence of a core grapheme-phoneme binding deficit in individuals with DD, manifesting itself on behavioural level. In addition, these results imply that the DA that was developed by Aravena et al. (to appear) is able to differentiate between individuals with and without DD, offering excellent potential for employing this DA as diagnostic assessment tool.

(17)

13

3 Bilingualism and language impairment

As the number of children who are being exposed to two languages from an early age increases, bilingualism is surging globally. More specifically, it has been estimated that half to two-thirds of the world’s population is bilingual (Baker & Prys Jones, 1998). Although the term bilingualism is widely used, it appears that no straightforward definition of bilingualism has been established yet. The majority of people may typically define a bilingual person as an individual who is able to speak two languages. Upon closer examination, however, it seems that the term involves a number of issues. The main issue that has been raised by the term involves the question as to whether individuals are required to be as competent in both languages to be considered bilingual (Baker & Prys Jones, 1998). By way of illustration, there may exist a difference between language proficiency and language use within a bilingual individual, which is often referred to as the difference between degree and function respectively. For example, an individual may be highly proficient in two languages, but in practice he/she may tend to use merely one language. On the contrary, it may be the case that a person may speak two languages regularly in practice, but he/she only has sufficient language abilities in one language. Often one language tends to be more advanced than the other, which is commonly referred to as the dominant language. In addition to these between-language differences, a person’s proficiency may also vary within a between-language (i.e. within-language differences) (Baker & Prys Jones, 1998). For instance, an individual may use his/her first language (i.e. L1) predominantly for reading and he/she may use his/her second language (i.e. L2) primarily for speaking. Hence, he/she may be more fluent in reading in his/her L1 than in speaking in his/her L1, and, consequently, he/she may be more fluent in speaking in his/her L2 than in reading in his/her L2. In other words, a person may use the four language modalities (i.e. speaking, listening, reading, and writing) within a language for different purposes, which may lead to differences in proficiency within a language. Furthermore, a person may, for instance, have an excellent comprehension (i.e. reading and listening) of his/her L2, but his/her L2 production (i.e. speaking and writing) is highly insufficient. Thus, besides having a dominant language, bilinguals tend to use their languages for different purposes (Baker & Prys Jones, 1998). These findings clearly suggest that the majority of bilingual individuals may not be equally proficient in both languages, which is often thought to be the case, i.e. various degrees of bilingualism may exist. Hence, it seems to be nearly impossible to establish a concise, and, simultaneously, an all-embracing definition of

(18)

14

bilingualism (refer to Baker and Prys Jones (1998) for an in-depth discussion of the definition of bilingualism). In the current study, however, the term bilingualism is used to ‘simply’ describe the use of two languages within an individual. In addition to the difficulty concerning defining bilingualism that is outlined above, it should be noted that bilingualism may be separated into two types on the basis of children’s exposure patterns, namely simultaneous bilingualism and sequential bilingualism. In the former type of bilingualism, children typically learn both languages through early exposure (i.e. before the age of three years) to these languages, often from birth onwards (Paradis, 2010). The latter form of bilingualism, i.e. sequential bilingualism, involves children whose L1 is already moderately established (although not entirely acquired yet) before L2 learning commences (Paradis, 2010). Although researchers commonly differentiate between these two types of bilingual language development, in the current study the term bilingualism is used as an equivocal term for either form of bilingualism.

3.1 Bilingual SLI

Currently, the considerable increase in the number of children that are being raised bilingually gives rise to a diagnostic dilemma for researchers, practitioners, and educators, among others, as regards disentangling the effects of bilingualism from the effects of a language impairment in bilingual children. Namely, the linguistic manifestations of a specific language impairment (SLI), which presents noteworthy similarities to DD (Bishop & Snowling, 2004), may closely resemble the manifestations of particular patterns in L2 acquisition in children. In recent years, a small number of studies has revealed that the expressive language use (i.e. spontaneous speech) in the L1 of children with SLI presents a striking resemblance to the spontaneous speech in the L2 of child L2 learners. Håkansson and Nettebladt (1996), as pioneers in this field of research, have drawn parallels between the acquisition of Swedish word order in children with SLI and child L2 learners of Swedish. Whereas child L1 learners of Swedish correctly use verb-second (V2) word order, both children with SLI and child L2 learners of Swedish transitorily adopt inaccurate verb-third (V3) utterances (Håkansson, 2001). More recently, it has been demonstrated that there are also remarkable similarities between the linguistic manifestations in French-speaking children with SLI and child L2 learners of French. Paradis and Crago (2000) compared the morphosyntax (e.g. tense morphology, agreement morphology, and temporal adverbials) of French-speaking children with SLI to the morphosyntax of child L2 learners of French. Their results indicate that,

(19)

15

amongst other findings, both children with SLI and child L2 learners are particularly prone to errors as regards tense morphology. In addition, the use of object clitics by children with SLI and child L2 learners have presented similar evidence (Paradis & Crago, 2003); both groups are especially prone to omission of object clitics in object pronominalisation contexts. Furhermore, Paradis (2005) examined the expressive language particularities of English monolingual children with SLI, on the one hand, and children L2 learners of English, on the other. Her findings indicate that there are interesting parallels between the error patterns in grammatical morphology (e.g. third person singular, past tense, and irregular past tense) of children with SLI and the children who acquire English as their L2. Moreover, as abovementioned parallels between the error patterns of children with SLI and child L2 learners are merely restricted to language production, Grüter (2003) investigated whether these similarities may also extend to language comprehension. Indeed, the similarity of error profiles between children with SLI and child L2 learners seems to extend to the domain of language comprehension as well.

Taken aforementioned into account, bilingual SLI has been a diagnostic dilemma and, hence, the identification of true impaired language abilities in bilingual children appears to be a formidable challenge for diagnosticians. This diagnostic dilemma of bilingual SLI incorporates over-identification, on the one hand, and under-identification, on the other hand (Bedore & Peña, 2008). Genesee, Paradis, and Crago (2004) described these two sorts of misclassifications as ‘mistaken identity,’ and ‘missed identity,’ respectively. The term ‘mistaken identity,’ or over-identification, signifies the unnecessary, or, rather, unjustified placement of children in special education programs (Genesee et al., 2004). The latter term, ‘missed identity,’ or under-identification, denotes the overlooking of the problems of a child who is, in fact, seriously language impaired (Genesee et al., 2004). One of the main reasons for this diagnostic dilemma is that, still, there appears to be a poor understanding of bilingual language development (i.e. L2 acquisition) (Genesee et al., 2004). As a result, there is limited normative data about bilingual language acquisition (Bedore & Peña, 2008). A delay in language development in bilingual children may therefore unjustly be attributed to the consequences of L2 learning (‘missed identity’), and it may also wrongly be ascribed to impaired language abilities (‘mistaken identity’). Furthermore, it should be noted that SLI typically should manifest itself in both languages (de Jong, 2008). Unfortunately, however, a child’s first language (L1) often cannot be assessed due to the lack of diagnosticians who are native speakers of the L1 of a child. Moreover, the assessment tools that are initially directed

(20)

16

at monolinguals often are not applicable to bilinguals because the data on clinical markers in bilingual SLI remains yet to be insufficient (Bedore & Peña, 2008).

The diagnostic predicament that is outlined above has instigated a new field of research, namely the study of bilingual children with SLI (BiSLI), which, above all, endeavours to disentangle the effects of bilingualism from the effects of SLI. Studies that are concerned with BiSLI have been comparing the language competences of typically developing (TD) child L2 learners to those of monolingual children with SLI, for instance. However, as the results of abovementioned studies by Håkansson and Nettebladt (1993, 1996), and Paradis and Crago (2000, 2003) for instance, suggest, the linguistic behaviour of child L2 learners and children with SLI presents striking similarities, which invalidates the comparison of bilingual children to their monolingual peers with SLI for identifying SLI in bilingual children. As such, comparing the language competences of bilingual children to the norms for SLI in monolingual children for the identification of SLI in bilingual children is not valid because it would over-identify a bilingual child as being language-impaired, that is, it would falsely diagnose SLI in bilingual children. Researchers have also been comparing the language abilities in TD child L2 learners to those in child L2 learners with SLI, which proves to be more productive than the comparison of bilingual children with monolingual children with SLI. In general, these studies indicate that there exist consistent differences between TD bilingual children and bilingual children with SLI as regards their linguistic competences. More specifically, Restrepo (1998) found that measures such as a parental report of a child’s speech and language abilities, the mean length of utterance in a child’s spontaneous speech, the error rate per utterance, and an overview of the family history of language problems may effectively discriminate Spanish-English bilingual children with SLI from TD bilingual children. Jacobson and Schwartz (2005) presented further evidence for the identification of SLI in bilingual children by means of comparing the language abilities of bilingual children with SLI to those of TD bilingual children. The results of their study indicate that Spanish-English bilingual children with SLI consistently differ from TD Spanish-Spanish-English bilingual children regarding their error pattern in English past tense morphology. The bilingual children with SLI had a considerable low accuracy rate compared with that of the TD bilingual children. The bilingual children with SLI also differed from their TD peers with respect to error type. For instance, the bilingual children with SLI merely produced overregularisation errors (i.e. wrongly adding –ed to the past tense form of an irregular verb) (1 %), whereas this type of error was highly common for the TD bilingual children (26 %). This suggests that Spanish-English bilingual children with SLI may be discriminated from their TD peers on the

(21)

17

basis of English past tense morphology as a grammatical measure. Furthermore, it has been demonstrated that SLI may accurately be identified in Spanish-English bilingual children by means of a NWR task (Girbau & Schwartz, 2008). Thus, it seems that measures such as a parental report, past tense morphology, and an NWR task may accurately identify SLI in (Spanish-English) bilingual children when they are compared to TD (Spanish-English) bilingual children. As such, certain measures appear to hold potential for the screening and assessment of SLI in bilingual children, on condition that their TD bilingual peers are used as the standard of comparison. These normative comparisons, however, are particularly concerned with high-density bilingual groups, such as Spanish-English bilinguals, and, consequently, these normative comparisons may be less viable for other bilingual groups. Furthermore, static grammatical measures, such as described above, which analyse the current language abilities of bilingual children, may not be the most convenient method for identifying SLI in bilingual children because TD bilingual children (i.e. the standard of comparison) demonstrate shifts in their level of language proficiency during their language development (Kohnert, 2010). As a result, the standard of comparison may be inconsistent. Moreover, children’s language abilities need to be evaluated in both their L1 and L2, which may be problematic considering that a native speaker of the L1 and the L2 of each child is then required to assess a child’s language competences, and the diagnostic assessment needs to conform to both languages. Therefore, a third method has been employed by researchers, which concentrates on within-child comparisons. That is, this approach examines the change of performance within a child by means of so-called limited training tasks or, rather, fast mapping (FM) tasks and DA tasks. In the former tasks, children are taught new linguistic information, e.g. invented grammatical rules and novel words, often through a period of exposure to the novel stimuli. Subsequent to this period of exposure, they are tested on their ability to produce or comprehend forms of the novel linguistic information that they have previously been exposed to in different contexts. The quality of the child’s performance is measured by means of the rate of children’s language-learning process, which is calculated on the basis of pretest and posttest measures, or the strategies that they use, for instance. By way of illustration, Hwa-Froelich and Matsuo (2005) employed an FM word-learning task for the ultimate purpose of examining the performance of TD Vietnamese-English bilingual on FM tasks, among other tasks. The TD Vietnamese-English children in their study were taught four novel (monosyllabic) words through a ‘hide-and-seek’ framework. That is, in this task the children were asked to find an unfamiliar object (using a novel word), for instance, which a dinosaur puppet had hid earlier. In reverse, the children were asked to tell the puppet to hide

(22)

18

an unfamiliar object, for example, which required the children to use a novel word expressively. The latter tasks, i.e. DA tasks, have the same principle as the FM tasks. It should be noted, however, that DA tasks typically incorporate the communication of clear instructions towards the participants and truly attempt to teach the novel stimuli, whereas FM tasks simply provides an activity in which the children are exposed to the novel stimuli. Furthermore, in DA tasks the participants’ level of performance is often determined by the number of novel stimuli that they have acquired and the rate at which they have acquired these, for instance. For example, Hwa-Froelich and Matsuo (2005) also employed DA word- and rule-learning tasks that were concerned with the children’s ability to acquire four invented bisyllabic words and an invented monosyllabic suffix rule. Their main findings indicate that TD Vietnamese-English bilingual children are perfectly capable of fast mapping a novel grammatical rule or even invented words. Bilingual children with SLI, however, appear to perform more poorly on FM and DA tasks (e.g Roseberry & Connell, 1991). Roseberry and Connell (1991) administered a DA task aiming at the identification of bilingual children with SLI within a wider group of bilingual children. The DA task that they employed in their study involved the teaching of an invented morpheme to a group of Spanish-English bilingual children with SLI and a group of TD Spanish-English bilingual children. Roseberry & Connell (1991) found that the bilingual children with SLI acquired the novel morpheme at a considerable slower rate in comparison with the TD bilingual children, i.e. the bilingual children with SLI performed more poorly on a DA task than their TD bilingual peers. Hence, it seems that bilingual children with SLI may be identified by their poor ability to acquire novel linguistic information. In other words, the effects of BiSLI may be discriminated from the effects of bilingualism on the basis of DA and FM tasks that assess the language learning ability of children. What is more, such tasks eliminate the need to evaluate the children’s language competences in both their L1 and L2 because these tasks solely focus on their language learning ability, or lack thereof. These DA and FM tasks, then, seem to offer excellent potential for diagnosing SLI in bilingual children. More importantly, such tasks also seem to be highly effective for the identification of a specific reading disability, such as DD, in bilingual children, which is illustrated below.

3.1 Bilingual developmental dyslexia

Abovementioned diagnostic predicament clearly exemplifies the problems that are involved in assessing bilingual language impairment. These diagnostic difficulties also seem to be

(23)

19

manifested in the identification of a specific reading impairment, such as DD, in bilingual children. It may be considerably difficult to differentiate between the effects of DD, on the one hand, and the effects of a bilingual setting, on the other hand, because poor performance on a reading test may be caused by more general poor language abilities in the second language, rather than by a specific reading difficulty per se. By way of illustration, reading abilities, such as word recognition, for instance, heavily depend on more general language abilities, such as vocabulary level. It has been reported that vocabulary level directly influences the reading ability of bilingual children (Ricketts et al., 2007). Note that it has been found that bilingual children generally have a smaller receptive vocabulary size in one of their languages than do monolingual speakers of that particular language. Bialystok et al. (2010), for instance, investigated the receptive vocabulary differences between bilingual children and monolingual children. They used the Peabody Picture Vocabulary Test (PPVT) to analyse the receptive vocabulary of 1,738 children, of which 772 children were monolingual and the remaining 966 children were bilingual. This task is a widely used measure for the assessment of the receptive vocabulary size of children. The PPVT requires children to indicate one of the four pictures that corresponds to the word that was uttered by the experimenter. The findings that were obtained by Bialystok et al. (2010) revealed lower receptive vocabulary scores for bilingual children than for their monolingual peers. Thus, bilingual children tend to know fewer words in one of their languages than their monolingual peers do in that specific language. Furthermore, it has been demonstrated that the smaller receptive vocabulary size of bilingual children persists into adulthood (Luk & Bialystok, 2012).To illustrate, an analysis of a large sample of adults between the ages of 17 and 89 years revealed that the monolingual adults outperformed the bilingual adults, i.e. monolingual adults obtained significantly higher scores on the PPVT than did the bilingual adults (Luk & Bialystok, 2012). Thus, considering the findings of these two studies by Bialystok et al. (2010), and Luk and Bialystok (2012), bilinguals generally tend to have a smaller receptive vocabulary size than do monolinguals. Likewise, bilingual children usually have a smaller expressive, or productive, vocabulary size in one of their languages than do their monolingual peers in that language. Hoff et al. (2012) compared the language development, including measures on expressive vocabulary, of bilingual children to that of monolingual children. Their results demonstrated that the monolingual children attained notably higher scores on the expressive vocabulary assessment task that Hoff et al. (2012) administered than did their bilingual peers. It is not extremely remarkable that bilinguals tend to know fewer words in one of their languages than do monolingual speakers of that particular language. Recall that bilinguals often use their

(24)

20

languages for different purposes. Therefore, particular words may be used in a context in which they merely use one of their languages. Consequently, these particular words may rarely, if ever, occur in their use of the other language, resulting in limited vocabulary sizes. It should be noted, however, that the bilingual children only obtained lower vocabulary scores in one of their languages. Namely, a measure of total vocabulary (i.e. the vocabulary size of both languages) suggest that the performance of bilingual children and monolingual children is comparable, or, even, the bilinguals’ total vocabulary is, on average, larger than that of their monolingual peers (Pearson, Fernández, & Oller, 1993). Although it seems that the total vocabulary size of bilingual children is not limited in comparison with that of monolingual children, bilingual children maintain to have a smaller vocabulary size in one of their languages than do their monolingual peers in that language. Recall that vocabulary level directly influences the reading ability of bilingual children. Thus, a limited vocabulary size may, consequently, impede the ability to read, facilitating reading difficulties that do not necessarily stem from a specific reading impairment, such as DD, but may be attributed to it. In this way, however, a bilingual child’s reading problems may unjustly be ascribed to the effects of a specific reading impairment, as DD (‘mistaken identity’). In reverse, the problems of a serious reading-impaired child may be overlooked (‘missed identity’). To illustrate, it has been found that bilingual individuals typically outperform their monolingual peer as regards reading ability. A study by Bialystok, Luk, and Kwan (2005), for example, investigated the decoding abilities in 40 English monolingual children (M age = 81.1 months) and in groups of 33 Spanish – English (M age = 82.7 months), 30 (voweled) Hebrew – English (M age = 80.0 months), and 29 Cantonese – English bilingual children (M age = 78.8 months). Note that Spanish, (voweled) Hebrew, and English all three have alphabetic writing systems, whereas the Cantonese writing system is considered to be a semanto-phonetic system, which bears no resemblance with those of Spanish, (voweled) Hebrew, and English. The decoding abilities of the participants were examined by means of non-words. Prior to the non-word decoding task, a character named Tony was introduced to the children. It was explained to the children that Tony came from another planet. In order for his spaceship to take off Tony was required to recite a number of ‘secret words’, but he forgot how to pronounce these words. Hence, the children were asked to assist Tony by pronouncing the secret words for him. One of their main findings was that Spanish – English and Hebrew – English bilingual children consistently obtained higher scores on the non-word decoding task with the English items than did their English monolingual peers with the same items. The Cantonese – English children also slightly outperformed the monolingual children on the decoding task, but

(25)

21

obtained lower scores than did the other two groups of bilingual children (i.e. the Spanish – English and Hebrew – English children). In sum, then, it has been reported that bilingual children, on average, perform better on a (non-word) decoding task in one language than do their monolingual peers in the same language. These findings suggest that bilingualism may facilitate the decoding ability of bilingual children, provided that the two languages hold similar writing systems. In other words, there may be a bilingual advantage in reading ability. This finding may be explained by the fact that language skills from one language may be positively transferred to the other language. To be more specific, a number of studies has revealed that phonological skills, for instance, may positively transfer across the two languages of bilinguals (Geva & Siegel, 2000; Wade-Woolley & Geva, 2000). Recall that well-developed phonological abilities are considered to be essential prerequisites for reading in TD (bilingual) children. Accordingly, a positive transfer of phonological abilities across the two languages in bilinguals may enhance bilinguals’ reading abilities. Moreover, the advantage in the reading abilities of bilingual children may be facilitated by their familiarity with decoding graphemes into phonemes in his/her L1. This, in turn, may facilitate a general understanding of the act of reading, and may provide bilingual children with a more solid basis for learning to read than their monolingual peers (Bialystok et al., 2005). As a result, bilingualism may facilitate reading ability on condition that the languages hold similar orthographic systems. In this way, however, a bilingual child’s specific reading impairment may unjustly be overlooked. This may often lead to late referral of children for reading intervention programs, which may obviously have a serious detrimental effect on a child.

In addition, a number of misleading beliefs about a specific reading disability (i.e. DD) in bilingual children can be added to the diagnostic dilemma that is outlined above, which, likewise the diagnostic dilemma, impede the identification of DD in bilingual children. These beliefs are mainly concerned with L2 oral language proficiency and its relation to reading (dis-)ability in bilingual children. It has often been thought that the reading difficulties that bilingual children may encounter in their L2 are a direct consequence of their insufficient L2 oral language proficiency (Geva, 2000). Taking this into consideration, bilingual children who may benefit from a reading intervention program because they have DD will, regrettably, remain unnoticed. As a result of the aforementioned belief, the diagnostic assessment of DD in bilingual children suffers from the widespread idea that efficient diagnosis of DD in bilingual children is problematic until their L2 oral language abilities are fully developed. Professionals, therefore, tend to avoid the diagnostic assessment of the reading difficulties of

(26)

22

bilingual children, for years even (Geva, 2000). Unfortunately, delayed assessment may also lead to the under-identification of bilingual children with a specific reading disability. To avoid the practice of under- and over- identification of DD in bilingual children, it is necessary that appropriate assessment tools as regards the identification of DD in bilingual children will be developed rapidly. Unfortunately, however, DD in bilingual children has only recently been studied in some depth. Fortunately, Elbro et al. (2012) acknowledged the large demand for research into bilingual DD, and, hence, attempted to fill this research gap. They developed a DA task that evaluates native speakers and L2 learners’ ability to acquire novel grapheme-phoneme associations. Note that it has been suggested that DD may typically be characterised by a grapheme-phoneme binding deficit. Thus, the DA task that Elbro et al. (2012) developed directly addressed the ability of grapheme-phoneme binding in native speakers and L2 learners. More specifically, the participants comprised two groups of adult native speakers of Danish, which were subdivided into 34 native speakers of Danish who were diagnosed with DD (M age = 36.4 years) and 31 non-dyslexic native speakers of Danish (M age = 26.4 years), and two groups of adult Danish L2 learners, which consisted of 53 Danish L2 learners with a suspicion of DD (M age = 34.7 years) and 35 non-dyslexic Danish L2 learners (M age = 33.1 years). Similar to the more recent study by Aravena et al. (to appear), Elbro et al. (2012) administered a DA consisting of a training session with the purpose of teaching three novel grapheme-phoneme associations to the participants. The novel letter-speech sound associations were composed of the graphemes ╔, ◊, and ◘, which corresponded to the phonemes /s/, /m/, and / α/ respectively. The training in these three novel grapheme-phoneme correspondences was presented as a paired-associate (PA) learning task, which typically involves the pairing of two items (i.e. letters and speech sounds in this case), that provided corrective feedback. To illustrate, the three novel graphemes were presented to the participant in a random order. Then, the participant was asked to produce the corresponding speech sounds. If an erroneous phoneme was produced, the correct speech sound was produced by the experimenter, who then initiated a new trial of graphemes (a trial consisted of three novel graphemes). This training session was ceased as soon as the maximum of ten trials was reached, or when the participant produced the accurate phonemes in three successive trials. Subsequently, the DA consisted of a session in which the adults were taught to read two-letter non-words that were written in the novel orthography. For instance, the experimenter combined two cards that both contained a single novel grapheme, and, hence, the participant had to produce the correct non-word that was thus created. Again, corrective feedback was provided. The task was discontinued when the participant produced

(27)

23

the correct words in two successive trials (one trial consisted of four two-letter non-words), or when the maximum of five trials was reached. Lastly, a word reading test with corrective feedback was administered, whereby the participants were asked to read twelve non-words that were composed of combinations of the three novel graphemes. The word reading test was stopped after three successive erroneous responses. A comparison between the performance of the non-dyslexic Danish L2 learners and the Danish L2 learners with a suspicion of DD revealed that the Danish L2 learners with a suspicion of DD typically performed more poorly on the dynamic reading test than did the non-dyslexic Danish L2 learners. These results suggest that a DA task may be a promising tool for the diagnostic assessment of DD in L2 learners. Moreover, Elbro et al. (2012) investigated the correlation between the participants’ performance on the DA, and their years of schooling and their scores on a receptive vocabulary test (i.e. their L2 language knowledge). A strong correlation between the L2 learners’ performance on the DA and their receptive vocabulary level may result in the over-identification of DD in L2 learners. Hence, for this DA test to be insensitive to the language proficiency of L2 learners, the correlation between the DA task and the receptive vocabulary test has to be relatively weak. Indeed, their findings revealed that the DA was less dependent on L2 receptive vocabulary than standard tests of reading (Elbro et al., 2012). In addition, it was also demonstrated that the DA was not dependent of the amount of schooling that the participants had received. Thus, the DA that was developed by Elbro et al. (2012) accurately identify L2 learners with a suspicion of DD and, more importantly, it is, to some extent, insensitive to the effects of bilingualism (i.e. poor L2 language knowledge). Although research concerning the usage of DA tasks for the identification of DD in L2 learners is yet scarce, the study by Elbro et al. (2012) reveals that these tasks may offer excellent potential for diagnosing DD in bilinguals.

(28)

24

4 Current study

The objectives of the present study were twofold. First, the promising results from a previous study by Aravena et al. (to appear) were a motivation for examining the potential of their DA for identifying, or, rather, diagnosing, a specific reading problem, i.e. DD, in bilingual children. In other words, the current study aims to extend the DA that was developed by Aravena et al. (to appear) to a bilingual context. Second, the current study was conducted with the aim of investigating the influence of bilingual effects on this DA. More specifically, the possible relation of vocabulary level to the results of the DA was examined. The questions that correspond to these aims are as follows: i) What is the potential of a dynamic assessment of the acquisition of grapheme-phoneme correspondences for identifying DD in Dutch bilingual children?, and ii) What is the relation of vocabulary level to this DA?

Prior studies of the reading development of bilingual children revealed that bilingual children, on average, perform better on a decoding task in either one of their languages than do their monolingual peers in that same language (e.g. Bialystok et al., 2005). Hence, it has been suggested that bilingualism may facilitate the decoding ability of bilingual children. In other words, it has been observed that there may be a bilingual advantage in reading development, which may be explained by the fact that language skills from one language, such as decoding abilities, may be positively transferred to the other language. With respect to the DA of the acquisition of grapheme-phoneme associations, then, it may be hypothesised that their bilingual advantage in decoding ability may enhance their performance on DA tasks, resulting in a strong performance on these tasks. In turn, bilingual children with DD may remain unnoticed. However, it has also been reported that the acquisition of grapheme-phoneme correspondences is the core deficit of DD (e.g. Aravena et al., to appear). Note that the acquisition of grapheme-phoneme associations is crucial to the ability to decode graphemes into their corresponding phonemes, and, thus, to the ability to read. Predictably, then, a grapheme-phoneme binding deficit in individuals with DD may disrupt their decoding ability. With respect to bilingual children with DD, their disrupted decoding abilities as a consequence of a grapheme-phoneme binding deficit may be transferred to a second language, which, consequently, may negatively affect the decoding abilities in that language. Taking this into account, it may be predicted that bilingual children with DD, whose disrupted decoding abilities negatively affect their decoding ability in a second language, perform more poorly on a DA than do their bilingual peers without DD, whose unimpaired decoding

(29)

25

abilities may positively transfer to a second language. As a result, this may lead to a correct identification of DD in bilingual children on the basis of a DA. Therefore, it was hypothesised in the current study that the DA of the acquisition of grapheme-phoneme pairs has full potential for identifying DD in Dutch bilingual children. That is, the DA that was employed in the current study may correctly detect a specific reading impairment, such as DD, in Dutch bilingual children. Furthermore, on the basis of the findings by Elbro et al., (2012) it was hypothesised that the vocabulary level of Dutch bilingual children does not correlate with the DA. In other words, it was predicted in the current study that the DA and the vocabulary level of the bilingual participants are not related to each other.

Reading difficulties in a bilingual context: a diagnostic dilemma