Annexe 4 : Manuscript Zourou et al. in press

Zourou, F., Ecalle, J., Magnan, A., & Sanchez, M. (in press). The fragile nature of phonological awareness in children with Specific Language Impairment : Evidence form literacy development. Child Language Teaching and Therapy.

Abstract

Previous studies have suggested that children with specific language impairment (SLI) risk encountering subsequent literacy difficulties, due to difficulties in phonological awareness. This longitudinal study provides evidence in support of this view based on a group of 20 French-speaking children with SLI examined at the start of learning to read. It also investigates the specific difficulties in reading decoding, reading comprehension and spelling. The sample was tested on standardized measures of phonological awareness, nonverbal skills, morphosyntax, reading, and spelling. The assessment was performed twice separated by an interval of 30 months. At Time 2, the results indicated normal scores on phonological awareness. Despite this, the sample scored poorly on literacy tasks, and the majority exhibited particularly striking deficits on spelling. We argue that there is a reciprocal influence between the early PA deficit and the subsequent literacy development in children with SLI. The sample of the study presented normal PA skills after 2 or 3 years of schooling. Nonetheless, most of the participants experienced difficulties on the literacy tasks, thus proving that the acquired PA skills were fragile and not sufficiently developed to cope with tasks more complicated than typical PA tasks.

Keywords

Specific language impairment, phonological awareness, literacy acquisition, literacy profiles

Introduction

Specific Language Impairment (SLI) is a common developmental disorder, with a prevalence in Anglo-Saxon studies usually estimated at somewhere between 3% and 9% of the population (Tomblin et al. 1997). The disorder is identified when oral language lags behind other areas of development for no apparent reason (Leonard, 1997). In young children, SLI is usually manifested through a variety of language deficits and it often persists into adulthood in the form of reading and spelling difficulties. Various longitudinal studies, most of which have been conducted with English-speaking children, have repeatedly demonstrated that children with SLI are at risk of subsequent literacy difficulties due to their early phonological deficit (Bishop and Snowling, 2004). The disorder, however, may be extremely diverse in nature, and children with SLI vary in terms of both the specific linguistic deficits they exhibit and the severity of these deficits (e.g. differences in the severity of the phonological deficit). In the present study, we used a longitudinal design to investigate the effects of written language learning on phonological awareness (PA) in an attempt to identify the literacy profiles of children with SLI.

Specific Language Impairment

Children diagnosed as having SLI exhibit a significant deficit in most language skills and in production and/or comprehension. The diagnosis of SLI is usually based on exclusionary and discrepancy criteria. Children with SLI achieve age-appropriate scores on non-verbal intelligence, have normal hearing and do not suffer from emotional or psychiatric disorders or any evident neurological dysfunction (Leonard, 1997). The term SLI is applied to a diverse range of children (van Weerdenburg et al. 2006), such as very young children who do not initiate language acquisition normally, teenagers after several years of language remediation, children with minor language disorders (-1 standard deviation -SD- on language tests) or a more acute disorder (-2SD on language tests).

The causes of SLI are still the subject of intense debate and current cognitive theories are divided into those that attribute SLI to general processing limitations and those that consider the disorder as caused by a specifically linguistic deficit (see review in Joannise and Seidenberg, 2003). Ziegler et al (2005) argue that the children with SLI have problems excluding noise and this “will certainly have tremendous consequences for normal phonological development" (p. 14114). For Joanisse and Seidenberg the speech perception deficit of children with SLI explains the grammatical deficits. For the authors, the perceptual deficit affects the use of phonological information in working memory, which in turns leads to poorer than expected syntactic comprehension. As Leonard (2009) forecasts, the coming years will probably lead to a better understanding of the disorder. Genetic, behavioral, and treatment studies will probably both affect the employed diagnostic categories currently in use and change the basis for the different types of diagnoses that are produced.

Phonological Awareness and Literacy Difficulties

The exact nature of SLI and the extent to which this impairment is able to explain the full range of language and cognitive difficulties encountered by these children continue to be areas of considerable debate. The research reported in the recent literature has provided a large body of evidence concerning two issues, the majority of children with SLI 1/ present poor phonological awareness (PA; the explicit awareness of sounds in words) and 2/ risk experiencing subsequent literacy difficulties. It has often been shown that most children with SLI have particularly pronounced PA deficits (Bortolini and Leonard, 2000). Moreover, in classification studies based on clinical observations, such as Rapin and Allen’s nosology (1988), children with both marked phonological and morphosyntactic difficulties have been identified as the largest subgroup. PA is the key to learning the relationships between spoken and written words, in other words, is crucial for the development of robust mappings between sounds and letters. There is now a great deal of evidence to support the idea of a reciprocal relationship between PA and literacy development (Castles and Coltheart, 2004). Indeed, many studies have demonstrated that PA is a central predictor of literacy.

A number of longitudinal studies, all conducted with English-speaking children with SLI, report that they run a high risk of experiencing literacy difficulties due to their early oral language and PA deficits (Bishop and Clarkson, 2003). In a study conducted by Catts et al (2002), the reading outcomes of 328 kindergartners with language impairments were evaluated at Grades 2 and 4. According to the results, approximately 50% of the sample exhibited significant reading difficulties after 2 or 4 years of schooling. Nonetheless, these results also show that some children with SLI seem to develop the same literacy skills as children with normal language development (Simkin and Conti-Ramsden, 2006). For example, children who have overcome their language deficits before the age of 5 and ½, seem to have better reading outcomes (Bishop and Adams, 1990). On the contrary, children who experience both language and PA deficits are the ones who run the greatest risk of failure (Catts et al. 2002). Despite the large body of research that has been conducted, few studies have specifically examined the types of written language difficulties experienced by children with SLI by simultaneously investigating reading decoding, comprehension and spelling skills. This is the aim of the current study.

The psycholinguistic approach

We adopted a psycholinguistic perspective which provides a possible explanation for why children with SLI often have associated literacy problems, ranging from poor reading skills to dyslexia (Stackhouse and Wells, 1997). At the heart of this model lies the assumption that children establish a speech processing system (SPS) by implicitly manipulating oral language. This SPS serves as the basis for the development of both speech and literacy. Any problems present in children's SPS will result not only in spoken language difficulties but also in problematic PA, which, in turn, will impact on their literacy skills. As argued above, children with SLI have a speech perception deficit which leads to difficulties in sound discrimination (Ziegler et al. 2005). This deficit in the input to the SPS affects phonological representations and the ability to manipulate phonemes. However, these early phonological representations are very important during the initial stages of literacy development when children have to rely on letter-to-sound and sound-to-letter mappings since their orthographic lexicon are only 'poorly stocked'. Difficulties in reading and spelling development are therefore unavoidable in children with SLI since these children are likely to develop inefficient and poorly structured sound-to-letter and letter-to-sound mappings.

Spelling skills of French children with SLI

Unlike reading skills the development of spelling skills in children with SLI who are learning alphabetic orthographies has attracted much less attention. Spelling in any alphabetic system is a more complex task than reading and is highly dependent on the establishment of fine-grained connections between orthography and phonology (for a review, see Caravolas et al. 2001). Learning to read and spell in French, an opaque language, confronts children with a serious mapping problem, which make the task more difficult than in other more transparent languages. Spelling, however, in French is more opaque (sound-to-letter mappings) than reading (letter-to-sound mappings). Learning to spell in an opaque orthography, with highly unpredictable sound-to-letter correspondences, is undoubtedly more difficult than learning to spell in a transparent orthography. A well-defined mapping between sounds and letters is very necessary for the development of appropriate spelling skills.

In the present longitudinal study, which extended over a period of 30 months, we tested the effects of literacy development on PA in 5- and 6-year-old French-speaking children with SLI. To our knowledge, this is the first longitudinal study of children with SLI in France. For the purposes of the present study, SLI was defined as language impairment on at least one test of expressive and/or receptive language (1 SD below the mean or below) in children who otherwise exhibited normal nonverbal abilities. We already argued about the existence of a reciprocal influence between PA and literacy development, thus it seems reasonable to expect that children with SLI will improve their PA skills in response to the explicit teaching of letter-sound mappings. However, due to the fragile nature of their phonological representations, we still expected to observe an effect of their early language deficit on literacy development. The current study examined this prediction in the light of two questions, 1/ will children with SLI perform normally on PA and oral language tasks after 2 and ½ years (Time 2) of explicit teaching of written language and 2/ will children who performed normally on PA tasks at Time 2 exhibit the same performance on literacy tasks. We examined the heterogeneity of children with SLI by examining the differences in their literacy profiles in terms of differences in decoding, comprehension and spelling skills. The study evaluated individual cases, an approach that is often considered to be a valuable means of providing information about the outcomes of speech and language intervention.

Method

Participants

Participants were 20 monolingual French-speaking children divided in two subgroups, the 5-year-old sample (attended kindergarten at Time 1) and the 6-year-old sample (attended 1st grade at Time 1). All of them had been diagnosed as having SLI by a multidisciplinary team of specialists in two reference centers for learning disabilities in Lyon, France. The children recruited for this study had to satisfy the following criteria: Diagnosis of SLI; French as primary language; no history of mental retardation or emotional disorder; no hearing (>70dB) or visual difficulties; performance IQ over 85 on the Wechsler scales of intelligence (Wechsler, 2001), verbal IQ below 85 or significant discrepancy from the performance IQ score of more than 15 on the Wechsler scales, together with a score of 1.25 SD below the mean on at least two additional language measures. The N-EEL²⁴ (Chevrier-Muller and Plaza, 2001) test battery is widely used by specialists in France and provides a fairly complete picture of a child’s language skills (phonology, morphology, and vocabulary, both in production and comprehension).

All the participants had been receiving speech-language therapy (SLT) administered by speech therapists for 1 (5-year-old sample) or 2 years (6-year-old sample) before the beginning of the study and continued to receive this support during our study. Among them, 16 children were diagnosed with expressive-only language impairment and 4 with combined expressive-receptive language impairment. Table 1 presents the entire sample at Time 1. At Time 1, a group of age-matched controls was recruited. At Time 2, we used an extensive battery of standardized formal tests which are in widespread use in France and no additional control group was therefore required.

Insert table 1

Tasks and Materials

Nonverbal skills were assessed using the Picture Completion subtest of the BREV neuropsychological test (Billard et al., 2002). The BREV battery has been standardized on 873 children, among which 173 with learning disorders. Each item includes a target image, which the child has to complete by choosing the appropriate picture from among 4 smaller ones all presented on the same page. Each correct answer scored 2 points, however, if the correct answer was preceded by a wrong answer only 1 point was awarded.

Oral language was assessed using the Test of Grammatical Closure (TCGR-C, Deltour, 2002), which evaluates induced language. The TCGR-C test has been standardized on 500 children (ages 3;0-9;0). It comprises a series of pictures presented in pairs. A target sentence is presented for the first picture and the participant has to complete this on the basis of the second picture. One point was awarded for each correct answer, thus resulting in a maximum score of 52.

To measure PA skills at T1, we administered a forced-choice phoneme deletion (PhD). The examiner named 18 pictures (e.g. /boeuf/ [beef]) and then the child had to choose the target word, i.e. a word sharing the same initial phoneme (e.g. the word /banc/), from a set of three pictures representing potential answers (Sanchez et al. 2007). Between the other two pictures, one shared the same final phoneme (e.g. /neuf/) and the other was a semantically equivalent word (e.g. /viande/ [meat]). No oral answer was required for this task. One point was awarded for each correct answer. At T2, we administered two standardized tests of the ODEDYS battery (Jacquier-Roux, Valdois and Zorman, 2005). The ODEDYS battery has been standardized on 536 children (ages 7;0-11;0). In the phoneme deletion (PhD) task, the participants were told to delete the initial sound of a spoken word and produce the remaining sound sequence (e.g. /ouvert/, /vert/). In the phoneme blending (PhB) task, the children were required to use the initial sounds of two spoken words (e.g. /bonne/ and /année/) to produce a syllable (/ba/). One point was awarded for each correctly produced syllable. All three tasks involved phoneme manipulation. The age of our participants at Time 1 forced us to use a task, which measured implicit PA skills. At Time 2, we were no longer constrained by the participants' age and were able to measure explicit PA skills.

To measure reading attainment, we administered two subtests of the K-ABC battery (Kaufman and Kaufman, 1983). The K-ABC test has been standardized on more than 2000 children (ages 2;6-12;6). The reading-decoding (RD) subtest of the K-ABC measures the participant’s ability to pronounce accurately French printed words. The stimuli consisted of 38 individual printed words and one point was awarded for each correct answer. The reading-understanding (RU) subtest comprised 24 items, each of which consisted of a printed command (1 to 20 words). The children had to act out the printed command to demonstrate that they had understood what they had read (e.g. /mange/ [eat]). One point was awarded for each correct answer. We also administered the spelling (S) subtest of the BREV neuropsychological battery (Billard et al., 2002), which comprises a total of 10 words and pseudowords. In this task, the children were asked to spell single words, single pseudowords and a series of words presented in a sentence context. One point was awarded for each correctly spelled word whereas ½ point was awarded for each word containing a single misspelled letter.

Procedure

The participants were evaluated twice separated by an interval of 30 months (Time 1 and Time 2). In each case they were assessed individually by a single examiner during the course of a single session (~1h) which included breaks to prevent fatigue. All the measures included a small number of demonstration items for which the examiner provided feedback regarding the correctness of the participant’s response. All items were administered to all participants.

Results

Phonological awareness in children with SLI: longitudinal results

Our first question related to PA and how they evolve over time in children with SLI. To assess deviance from normal development we calculated z-scores based on the mean and SD of a norm group (either the control group used at Time 1 or the norms of the standardized measures used at Time 2). The criterion adopted for deviance throughout the present study was n=1.65 SD which, in a normal distribution, corresponds to the 5th percentile and is therefore a reasonable threshold for deviance. Z-scores below -1.65 were considered to be statistically significant and to indicate that the participant's performance was abnormal in the task in question. Table 2 indicates the scores on PA. The results we obtained at Time 1 revealed that both the 5- and the 6-year-old samples achieved low performances on the PhD task. In terms of the percentage, 65% (13 out of 20) of the SLI children obtained significantly low scores in the PhD task at T1, with the majority of these belonging to the 6-year-old sample (9 out of 13). At Time 2, however, both the 5-and the 6-year-old samples performed well in the PhD task. In the case of the PhB task, the scores were slightly lower than those of normally developing children although still in the normal range. Only 2 children (S12 and S20) in the 6-year-old sample obtained significantly low scores at Time 2 and then only for the PhB task.

Insert table 2

Oral language development of children with SLI

Scores on the Test of Grammatical Closure (TCGR-C) were transformed to age of development (AD) equivalents. We then calculated the discrepancy between Chronological Age (CA) and AD for each child. The results are presented in Table 3. One way to characterize the oral language skills of the sample is to calculate the percentage of children who could be classified as having an oral language deficit. For the purposes of this investigation, an oral language deficit was defined as a discrepancy of at least 18 months between CA and AD. Our results indicated that 25% of the sample (5 out of 20) exhibited a major oral language deficit even after 2 (participant S2 of the 5-year-old sample) or 3 years (participants S12, S14, S15, S20 of the 6-year-old sample) of schooling. Furthermore, another 15% (3 out of 20) of the sample appeared to experience significant difficulties, with a discrepancy of 12 to 18 months between CA and AD in oral language (participants S1, S3, S7 of the 5-year-old sample).

Insert table 3

Literacy outcomes of children with SLI

At Time 2, we measured spelling, reading decoding, and reading understanding skills in both samples. We calculated z-scores in order to measure deviance from normal development in the same way as for the phonological awareness tasks (Table 2). The results showed that the spelling scores of 80% of the sample (16 out of 20) were significantly deviant and that no significant group differences were observed. Concerning the two reading measures, the sample performed significantly better in the reading-understanding (RU) task than in the reading-decoding (RD) task (TTEST, p=0.04, 2-tailed). More precisely, 35% of the sample (7 out of 20) performed significantly low level in the RD task and 20 % of the sample (4 out of 20) in the RU task. No significant differences between the two age groups were observed for the reading tasks. The four children exhibiting poor performances on the RU task were those with the receptive type SLI (S2, S3, S15, S20).

Links between phonological awareness, language and literacy outcomes

Another question that interested us in the present study was to study the links between the PA skills that children possessed at the beginning of the study (Time 1) and their language and literacy outcomes 2-and-half years later (Time 2). To answer this question, we calculated correlations between the PA skills (PhD task) at Time 1 and the language (TCGR-C), spelling (S), reading-understanding (RU), and reading-decoding (RD) skills at Time 2. The results we obtained suggest that PA (Time 1) correlated significantly with both spelling (r=.40, p =.05) and reading-understanding (r=.51, p=.05) at Time 2. However, reading-decoding (r=.33, p=.10) and language skills (TCGR-C) (r=.21, p=.30) did not correlate significantly with PA. In other words, performance in the phoneme deletion task at Time 1 correlated significantly only with performance on the spelling and the reading-understanding tasks at Time 2.

Toward a typology of beginning-readers with SLI

We used a K-means cluster analysis to examine the literacy profiles of the sample. The main objective of such an analysis is to identify homogeneous subgroups by maximizing the inter-variation between groups and minimizing the intra-variation inside each group. We used three parameters 1/ the language parameter, defined by the oral language (TCGR-C task) and phonological awareness skills (mean on the phoneme deletion and phoneme blending tasks of Time 2), 2/ the reading parameter, defined by the reading-decoding (RD) and the reading-understanding (RU) skills, and 3/ the knowledge of written words parameter (KWW) defined by the reading-decoding (RD) and spelling (S) skills. We formed 4 subgroups on the basis of these three parameters. The results are presented in table 4. Group A comprised the children who exhibited deviant scores on oral language and spelling, group B comprised those who achieved poor scores on spelling only, subgroup C comprised those with poor spelling performances but relatively good performances in the reading-understanding test, group D comprised the children who exhibited quasi-normal skills on all three parameters but with a relatively poor spelling performance. One important observation deriving from the results we obtained is that the discrepancy on the z-scores for spelling was significant for all 4 subgroups (range from z = -3.22 to z= -1.89).

Insert table 4

Figure 1 presents the results of the K-means cluster analysis for the 4 subgroups run on the three parameters presented above. The central range around 0 is set at 1.65 SD or 1 year and 6 months of discrepancy between chronological age and age of development, and represents the ‘normal’ range. In other words, scores in this range are not considered as significantly deviant.

Insert Figure 1

Discussion

The major questions that interested us in this study were a) to identify whether schooling and literacy acquisition help children with SLI improve their deficient PA and b) to find out whether children who achieve normal scores on PA tasks after some of years of schooling perform equally well on literacy tasks.

As far as the first question is concerned, we evaluated the PA skills of children with SLI in a longitudinal design. The sample performed poorly at Time 1, whereas at Time 2 the majority of children performed in normal levels. In other words, after 2 (for the 5-year-old-sample) or 3 (for the 6-year-old-sample) years of schooling and participation in remediation programs, the scores on PA tasks achieved by the SLI children were no longer deviant compared to those obtained by NLD children of the same age. This result seems to corroborate the principal hypothesis of this study. Literacy development contributes at some level to SLI children's sensitivity to the phonemic aspects of their language through the development of mappings between letter and sounds. Although the evidence provided by the present study was gathered from a somewhat small group of participants who performed a limited number of tasks, this result seems important to us, in particular because it is the first study that has used a longitudinal design in French-speaking children with SLI. As far as the second question raised here is concerned, one might expect that children who obtain normal scores on PA tasks (Time 2), and who are therefore able to manipulate phonemes explicitly, will also obtain scores close to normal on literacy tasks. However, this is not the case. Although the majority of the children seemed to acquire a certain level of PA over time (Time 2), their skills remained largely task-specific since they were unable to generalize the use of this knowledge to tasks such as reading and, to a greater extent, spelling which require a high level of explicit phoneme manipulation. In the light of the above, we argue that literacy development helps children with SLI improve their PA skills. However, the fragile nature of their early phonological representations prevents children with SLI from transferring their acquired knowledge to more complicated tasks such as reading and spelling. This longitudinal study involving French-speaking monolingual children with SLI seems to confirm that a history of SLI is one important factor causing literacy difficulties in children (Bishop and Clarkson, 2003).

The correlation analysis showed that the children who experienced the greatest difficulties in the PA task at Time 1 were also those who experienced the greatest difficulties in spelling and reading comprehension at Time 2. This does not mean that PA is the only factor responsible for the literacy difficulties experienced by children with SLI. It does, however, confirm that PA is an important factor predicting subsequent literacy failure. Spelling development seemed to be more severely affected, thus proving that the participants had not only failed to develop appropriate sound-to-letter mappings but that they were also particularly affected by the specificities of the French language in which the level of inconsistency is greater for spelling than for reading. Another argument concerning the discrepancy between reading and spelling performances is that reading may still progress to some extent provided that the child has intact visual processing skills and sufficient verbal comprehension to compensate. Spelling development, on the other hand, is more dependent on speech processing skills, with the result that spelling problems may be more evident than reading problems in children with speech and literacy difficulties. With regard to those children in the sample who exhibited normal performances in the task designed to measure reading comprehension but significantly low performances in the reading decoding task, we argue that such children use compensatory mechanisms based on semantic information and context which may enable them to understand a sentence even if they do not correctly decode the words it contains. These compensatory mechanisms were theoretically available to all our participants, even though they could only be employed by some of them or by a specific sub-group of them. It is therefore clear that this question needs to be examined in greater detail among a larger population.

Earlier psychological, neuropsychological and clinical studies in children with SLI have drawn particular attention to the issue of the extremely diverse forms taken by the disorder (van Weerdenburg et al. 2006). The results we obtained highlight the heterogeneity of children with SLI. For instance, as the cluster analysis revealed, almost all of the evaluated variables cut across the typology, with the exception of spelling for which the majority of the sample exhibited relatively poor performances. Due to the relatively limited sample of participants in this study, the cluster analysis identified only small subsets of children (3 to 6 children per subgroup) and no inter-individual variables were therefore considered during the discussion of the results. This question undoubtedly deserves to be re-examined using a larger sample of participants that might make it possible to obtain more precise results. However, the cluster analysis used in this study highlights some interesting potential groupings characterizing populations of children with SLI.

Further research is needed in order to clarify the importance of other factors, such as vocabulary development and acquisition of grammatico-syntactic rules. Recently, attention has also focused on poor short and long-term memory resources as possible causes of the difficulties encountered by children with SLI (Archibald and Gathercole, 2007). Poor working memory skills could account not only for the differences in the literacy profiles but also for the differences in the persistence of the PA deficits. By conducting further longitudinal studies in children with SLI at the start of learning to read, we will eventually arrive at a more precise understanding of the nature of the early phonological awareness deficit and its impact on literacy development.

The present study possesses a certain clinical value as well, in particular for clinicians working with French-speaking children with SLI. It provides longitudinal data on French-speaking children with SLI and shows that schooling and remediation programs can be effective. It also provides a certain amount of evidence suggesting that low scores on PA tasks at 5 or 6 years-olds are correlated with low scores in spelling and reading comprehension two and a half years later. Finally, it examines the specific literacy profiles of children with SLI on three different literacy tasks – reading-decoding, reading-understanding and spelling – that show that spelling skills are specifically affected in children with SLI who are learning French. This result sends an important clinical message about the need for clinicians to propose direct intervention programs designed to encourage reading and spelling development when conducting SLT remediation programs.

References

Archibald, L.M.D. and Gathercole, S.E. (2007) The complexities of complex memory span: Storage and processing deficits in specific language impairment. Journal of Memory and Language, 57 (2), 177-194.

Billard, Catherine et al (2002) The BREV neuropsychological test: Part I. Results from 500 normally developing children. Developmental Medicine and Child Neurology, 44, 391-398.

Bishop, D.V.M. and Adams, C. (1990). A prospective study of the relationship between specific language impairment, phonological disorders and reading retardation. Journal of Child psychology and Psychiatry, 31, 1027-1050.

Bishop, D.V.M. and Snowling, M.J. (2004) Developmental dyslexia and specific language impairment: Same or Different. Psychological Bulletin, 130 (6), 858-886.

Bishop, D.V.M and Clarkson, B. (2003). Written language as a window into residual language deficits: A study of children with persistent and residual speech and language impairments. Cortex, 39, 215-37.

Bortolini, U. and Leonard, L.B. (2000) Phonology and children with specific language impairment: Status of structural constraints in two languages. Journal of Communication Disorders, 33, 131-150.

Caravolas, M., Hulme, C. and Snowling, M.J. (2001) The foundations of spelling ability: evidence from a 3-y longitudinal study. Journal of Memory and Language, 45, 751-774.

Castles, A. and Coltheart, M. (2004). Is there a causal link from phonological awareness to success in learning to read? Cognition, 91, 77-111.

Catts, H.W., Fey, M.E., Tomblin, J.B. and Zhang, J.B. (2002) A longitudinal investigation of reading outcomes in children with language impairments. Journal of Speech, Language and Hearing Research, 45, 1142-1157.

Chevrier-Muller, Catherine and Plaza, Monique (2001) N-EEL : les Nouvelles Epreuves pour l’Examen du langage [New tasks for language evaluation]. Paris, Centre de Psychologie Appliquée.

Deltour, J.J. (2002) Test de Closure Grammaticale [Test of Grammatical Closure]. Paris, EAP.

Jacquier-Roux, Monique et al (2002) Outil de Depistage des Dyslexies, ODEDY’S [Tool for dyslexia screening]. Grenoble, Laboratoire Cogni-Sciences IUFM.

Joanisse, M.F. and Seidenberg, M.S. (2003) Phonology and syntax in specific language impairment: Evidence from a connectionist model. Brain and language, 86, 40-56.

Kaufman, A.S. and Kaufman, N.L. (1983) Kaufman Assessment Battery for Children (K-ABC). American Guidance Service. [Adaptation française: Batterie pour l’examen psychologique de l’enfant. 1993]. Paris, ECPA.

Leonard, L.B. (1997) Children with specific language impairment. Cambridge (MA), MIT Press.

Leonard, L.B. (2009) Some reflections on the study of children with specific language impairment. Child Language Teaching and Therapy, 25 (2), 169-171.

Rapin, I. and Allen, D.A. (1988) Syndromes in developmental dysphasia and adult aphasia, in F. Plum (ed.), Language, communication and the brain. New York, Raven Press.

Sanchez, M., Magnan, A. & Ecalle, J. (2007). Habiletés phonologiques chez des enfants dysphasiques de GS et CP: étude comparative avec des enfants au développement langagier normal. Psychologie Francaise, 52, 41-45.

Simkin, Z. and Conti-Ramsden, G. (2006) Evidence of reading difficulty in subgroups of children with specific language impairment. Child Language Teaching and Therapy, 22, 315-331.

Stackhouse, Joy and Wells, Bill (eds) (1997) Children’s speech and literacy difficulties: A psycholinguistic framework. London, Whurr Publishers.

Tomblin, J.B., Records, N.L., Buckwalter, P.R., Zhang, X., Smith, E. and O’Brien, M. (1997) Prevalence of specific language impairment in kindergarten children. Journal of Speech, Language and Hearing Research, 40 (5), 1245-1260.

van Weerdenburg, M., Verhoeven, L. and van Balkom, H. (2006) Towards a typology of specific language impairment. Journal of Child Psychology and Psychiatry, 47 (2), 176-189.

Wechsler, D. (2001) Wechsler Intelligence Scale for Children – Third Edition. London, The Psychological Corporation Limited.

Ziegler, J.C., Pech-Georgel, C., George, F., Alario, F-X. and Lorenzi, C. (2005) Deficits in speech perception predict language learning impairment, PNAS, 102 (39), 14110-12115.

¹PA= mean z-scores on the phoneme deletion and phoneme blending task

RU = Reading-understanding; Phonology = Phonological skills; TCGR-C = Grammatical Closure test; RD = Reading-decoding;

Authors' note

We should like to thank neurologist Sibylle Gonzalez, psychologists Christelle Glissoux and Annie Ritz from ‘Lyon-Est’ Hospital and neuropsychologist Vania Herbillion from ‘Debrousse’ Hospital for allowing us access to the population and providing us with support throughout the study. Warm thanks are also extended to the families and to all the children who participated in the studies.