Specific language impairment (SLI) and reading development in early school years

Abstract

Specific language impairment (SLI) is a condition that affects children’s emerging language skills. Many different language skills can be affected in SLI, but not all individuals with SLI have the same set of difficulties. As a result, SLI is a highly heterogeneous condition. The ability to read and understand written text is a higher function of language, which has its base on the language skills that may be affected by SLI. Previous studies suggest that children with SLI are at risk of developing literacy difficulties at school age. To test this hypothesis a longitudinal follow-up study of SLI was performed in an urban city in Finland. Forty-three children diagnosed with SLI completed tests for lexical and rapid automatized naming (RAN) and reading ability (technical reading and reading comprehension). Children were tested in schools in years one to three, at the end of each school year. At the age of 10 (year 3) children with SLI had relatively weak technical reading abilities with somewhat stronger reading comprehension abilities and concomitant lexical retrieval impairments, but age-appropriate RAN abilities. These findings lend further support to the view that SLI imposes a considerable risk in learning to read. Even in an orthographically highly transparent language, such as Finnish, technical reading skills often seem to remain poor. In this study, the findings on the development of lexical retrieval abilities support the Matthew-effect model in reading development (Stanovich, 1986), while the results of the development of RAN abilities support the developmental lag theory (Stanovich et al., 1988). All in all, these children with SLI had great difficulties in learning to read, which places them at a serious risk of developing concomitant learning problems.

Keywords

Lexical impairment rapid automatized naming (RAN)reading SLI specific language impairment

I Introduction

Specific language impairment (SLI) is a developmental disorder that is diagnosed when a child’s language abilities are impaired in absence of obvious neurological, sensorimotor, nonverbal cognitive or social emotional deficits (Leonard, 1997). Children with SLI have impaired abilities in one or more aspects of language (grammatical morphology, complex syntax, semantics and/or pragmatics) (Cirrin and Gillam, 2008). The prevalence of SLI has been estimated between 1–8 percent of the population (Hannus et al., 2009; Tomblin et al., 1997), with evidence of rising prevalence (Hannus et al., 2009). Although this language impaired population is heterogeneous, their language and literacy difficulties often persist into adulthood (Simkin and Conti-Ramsden, 2006; Snowling et al., 2001; Webster et al., 2004: Whitehouse et al., 2009; Young et al., 2002).

Problems in reading acquisition may stem from many factors. The deficit model of reading (Francis et al., 1996) expostulates that there is a persistent lack of the necessary set of skills for reading, although the development in reading parallels that of the more accomplished readers. The developmental lag theory (Stanovich et al., 1988) claims that slower development in the beginning of reading is due to underlying delays in cognitive skills which, over time, will be conquered. The slower growth or ‘Matthew-effect’ model (Stanovich, 1986) claims that as children have lower levels of early reading ability, it then leads them to less exposure to texts and reading which, in turn, leads to lower levels of further achievement. Thus, the initial lag in reading ability will only widen compared to normally developing readers. So far most studies on dyslexia and SLI support the deficit model of reading (for example Bishop and Snowling, 2004; Botting et al., 2006; St. Clair et al., 2010).

Many cognitive skills underlie reading ability and the ability to understand what has been read. A strong connection exists between phonological abilities and reading and spelling, and problems in phonological abilities are thought to cause dyslexia (Snowling, 2000). Poor phonological abilities hinder the ability to translate orthography to phonology, which may result in inaccurate reading (Hulme and Snowling, 2009). Theoretically there is still controversy over to what extent phonological abilities form the core deficit in SLI (for reviews, see Bishop and Snowling, 2004; Vandewalle et al., 2012), but there is a significant overlap between dyslexia and SLI (Bishop and Snowling, 2004; Catts et al., 2005). In a longitudinal study by Vandewalle et al. (2012) children with SLI had continuous problems in phonological awareness tasks at the age of eight and were at high risk of developing literacy difficulties when this was combined with poor rapid automatized naming (RAN) abilities.

Although phonological abilities have been linked to the accuracy of reading (Savage et al., 2005), their effect on reading development in transparent orthographies seems to be transient (Babayiğit and Stainthorp, 2011). In these orthographies (such as Turkish or Dutch) RAN abilities have been found to be the most consistent predictor of further reading development and to be linked with the fluency of reading (Babayiğit and Stainthorp, 2010; Verhagen et al., 2008). In turn, RAN abilities have been reported to depend on lexical retrieval and non-linguistic speed of processing, which both are problematic for children with SLI (Coady, 2013).

Oral language skills also affect reading development (Brizzolara et al., 2011; Lewis et al., 2011; Nation and Snowling, 2004). In studies on so-called ‘poor comprehenders’ (children, who have reading comprehension problems despite normal reading accuracy skills), it has been reported that weaknesses in semantic processing and comprehension monitoring lead to problems in reading comprehension, but also in narrative skill, receptive language and vocabulary (Hulme and Snowling, 2009). In a study by Nation et al. (2004) a substantial minority of these poor comprehenders met the eligibility criteria for SLI. Bishop and Snowling (2004) and Storch and Whitehurst (2002) claim that although theories of dyslexia have traditionally centred on phonological abilities, semantic and syntactic abilities play important roles in reading and reading comprehension. These findings place children with SLI at considerable risk for various difficulties in reading.

There is a growing body of evidence of children with SLI having continuous problems with different facets of language and reading (see above). These problems place them at additional risk of learning problems and have been shown to affect their peer relationships in early adulthood (Lindsay and Dockrell, 2012) as well as mental and social well-being in adulthood (Arkkila et al., 2008; Durkin and Conti-Ramsden, 2010). There is also evidence on SLI manifesting itself differently in different languages (Bedore and Leonard, 2005; Bortolini et al., 2002). In Finland, dyslexia has been studied extensively (e.g. Lyytinen at al., 2001) but there are still relatively few studies of SLI or its development in relation to reading. Finnish language is orthographically highly transparent, meaning that almost every phoneme is represented by a single letter. Thus, the reader needs to master fewer phoneme-to-letter conversions than, for example, in English; this simplifies the process of learning to read Finnish (Aro, 2004; Babayiğit and Stainthorp, 2011; Georgiou et al., 2008; Lyytinen et al., 2006). In theory, this should also help the child with SLI in learning to read.

In this study we examined the development of reading skills in children with SLI in an orthographically highly transparent language between the ages of seven and 10, to ascertain the extent of suggested reading difficulties in children with SLI. We wanted to study both technical reading and reading comprehension in SLI, to see if they followed a mutually similar developmental pathway. Special attention was placed on the role of lexical retrieval abilities and RAN abilities in reading development. In addition, the findings were considered in the light of the three reading models mentioned above, to see whether we would observe support to the deficit model of reading (Francis et al., 1996), the developmental lag theory (Stanovich et al., 1988) or the ‘Matthew-effect’ model (Stanovich, 1986) in the present population of children with SLI.

II Method

1 Participants

Participants were 43 (37 boys, 6 girls) Finnish-speaking (monolingual) children in an urban city in Finland (population 172,000). They were born in 1998 or 1999 and had been diagnosed with SLI (ICD-10 categories F80.1 or F80.2) in a secondary health care centre. Their families were contacted before school entry (children’s age at the time was 5–6 years) either through secondary health care centres and speech and language therapists (SLTs) working in the region, or through the kindergarten system. Of the 70 families thus reached, 54 were eligible to participate in this study, and 45 families opted to continue in a longitudinal study, most non-participants claiming their decision was influenced by the stress of school start. Furthermore, two children had to be eliminated from the study as one’s diagnosis changed significantly and the family of the other moved. In Finland, children start school in the fall of the year they turn seven years old, with nearly all of them having attended non-mandatory but free-of-charge preschool the previous year. This was also the case in this study.

2 Tests and procedure

The Finnish reading test ALLU (Lindeman, 1998) was used to measure technical reading and reading comprehension skills independently. The ALLU test was given by the teacher of each child at the end of each school year in a classroom testing situation. In the test for technical reading on year one the child has five minutes to find a corresponding word among four words for a picture. This test has 80 picture/word pairs. On year two and three the child has two minutes to find a corresponding sentence among four sentences for a picture. This test has 20 picture/sentence pairs. The words and sentences for a given picture are all different in meaning, but similar in structure and length. In the test for reading comprehension the child reads a text (approximately 90–120–150 words in years one to three) and answers 48 multiple choice questions in 60 minutes. The texts are factual in nature and handle everyday situations in children’s lives. For both types of tests, test scores are the amount of correct answers.

In the ALLUsubtests raw scores are transformed to numerical results (skill levels 1–9) which, in turn, enable the comparison of skills across years. Comparing children’s reading skills across the years using the raw scores was not possible, as the reading tasks used in the test vary between years and the distribution of scores in a given year differs from that of another year. Thus, direct comparisons between a score of 17 in technical reading on year one and score of 25 on year three cannot be made. When the raw scores are transformed to skill levels, 1 to 3 mean weaker than average skill, 4 to 6 correspond to average skills, and 7 to 9 stand for above average skills.

At the end of each school year, children’s lexical/picture naming abilities were tested with the Boston naming test (Finnish version, Laine et al., 1997; test includes 60 pictures of objects, which the child is asked to name) and the rapid automatized naming (RAN) with the Test for rapid serial naming (Finnish version, Ahonen et al., 2003). This RAN test has 6 subtests (objects, colours, letters, numerals, mixed letters and numerals and mixed letters, numerals and colours). A RAN subtest includes 50 pictures in 5 rows of 10 pictures, which the child is asked to name in order as rapidly as possible. Test scores are the amount of time used (in seconds) and the mistakes made by the child. Tests were given in a face-to-face situation by the corresponding author of this article.

3 Statistical methods

The statistical data were analysed with SPSS for windows. Due to the sample size and the fact that the results of the language tests did not adhere to normal distribution, nonparametric methods of analysis were used. Transforming the scores to achieve normal distribution and thereby expanding our possibilities in choosing the statistical methods was considered. Transforming the data to Gaussian distribution is often considered in order to benefit from the powerfulness of parametric testing (Spina, 2007). Specific methods are, for example, square root or logarithmic transformations (Wallenstein et al., 1980). This proved not to be an option, as the distributions between our test results in a single year were highly variable and, in some parameters, skewed. With the nonparametric methods the changes between years in the results in the language and reading tests were analysed using Friedman’s test (Tukey’s test post-hoc). Correlations between the language tests at the end of the first and second year of school and reading skill levels at the end of the third year of school were calculated. The possible correlation of change in the language tests to the change in reading skill levels was analysed by calculating Spearman correlation co-efficient. Statistical significance in these analyses was defined at 95% confidence levels (p < 0.05).

In the sample, there were missing results of the reading tests for every year studied (Table 1). Most of the teachers claimed these were due to children’s inability to read well enough to be able to participate in these tests. In the light of this information, but us also being aware that some missing results could be caused by other reasons (for example illness on the day of testing), and in order to enable the statistical analyses of reading development, we decided to replace the missing results with the median number of the obtained results, a method recommended by Acuña and Rodriguez (2004) for parameters with a skewed distribution. Other approaches for dealing with missing data are, for example, case deletion or parameter imputation (see Acuña and Rodriguez, 2004).

Table 1.

Number of missing results in reading test in years 1 to 3.

Missing results on ALLU reading test (n = 43)	Year 1	Year 2	Year 3
Technical reading	4	11	17
Reading understanding	20	12	14

III Results

In this study, in technical reading skills in year one the majority of children with SLI were at average skill levels, but in years two and three the majority of them scored weak skill levels. In concurrence with that, technical reading development showed a statistically significant decline between the first and third year of school (Friedman’s test, F = 9.346, Df = 3, p < 0.001) (Figure 1).

Figure 1.

Development of Technical Reading skill level during the 3-year follow-up period.

The development of reading comprehension progressed somewhat differently. Through the years approximately half of the children had weak skills in reading comprehension and half had average skill levels. A statistically significant improvement in skill levels was apparent between years two and three (Friedman’s test, F = 7.61, Df = 2, p < 0.001) (Figure 2).

Figure 2.

Development of Reading Comprehension skill level during the 3-year follow-up period.

As a group, children with SLI in this study made statistically significant gains in their lexical/picture naming abilities (Figure 3, Boston naming test) during their first three years at school (from age 7 to age 10 years). However, the overall pace of this lexical development did not achieve the levels of their peers. Table 2 shows that the mean results in the group with SLI at the age of seven year was 34.5 (age equivalent mean 38.9), and at the age of 10 the children with SLI had fallen further behind in their lexical development compared to their peers (SLI mean 39.2 and age-equivalent mean 52.7).

Figure 3.

Development of Boston naming test scores during the 3-year follow-up period.

Table 2.

The means, standard deviations (SD) and medians (MD) of language test scores for children with SLI.

Test/subtest	Year 1			Year 2			Year 3
Test/subtest	Mean	SD	MD	Mean	SD	MD	Mean	SD	MD
Technical reading (ALLU)	4.42	2.163	4.0	2.98	1.819	3.0	2.44	2.153	1.0
Reading comprehension (ALLU)	2.49	2.028	1.0	2.98	3.0	1.0	2.95	2.38	2.0
Boston naming test (BNT)	34.51	5.771	35.00	37.07	5.663	37.00	39.19	5.636	39.5
Rapid automatized naming (RAN):
colour	59.26	17.735	56.00	54.05	14.496	51.50	48.33	14.246	45.00
object	68.33	18.258	64.50	58.20	15.103	56.00	51.90	11.036	49.50
numeral	46.76	20.266	42.00	34.81	8.827	32.00	30.71	9.913	26.50
letter	47.29	16.650	41.00	35.95	12.360	33.00	29.55	9.459	28.00
letter +numeral	61.13	24.214	54.50	43.29	12.136	39.50	38.36	13.142	35.50
letter+numeral+colour	68.80	27.615	63.00	52.52	20.910	46.50	44.83	16.257	41.00

RAN abilities developed differently from other naming abilities. In all six subtests (colours, objects, numerals, letters, letters+numerals, letters+numerals+colours) the results developed in a constant fashion compared with each other (Table 2), so for this article only the results from the letter-naming subtest will be presented in a figure.

Figure 4 shows a decline in the time used to name 50 letters (Friedman’s test, p < 0.001). As a group, in year one (at the age of seven) children with SLI performed this task 1 SD slower than the norm. By the end of year two they had caught up in speed with the age norms. The accuracy of rapid automatized naming did not differ from the age norms in any measured year.

Figure 4.

Development of Rapid automatized naming test scores (letter subtest) during the 3-year follow-up period.

The connections between reading development and different naming abilities were not uniform. Lexical/picture naming abilities measured in year one correlated positively with reading measures in year three; technical reading development, Spearman r = 0.670 (p < 0.01**) and reading comprehension, Spearman r = 0.516 (p < 0.01**). On the other hand, RAN abilities measured in years one or two had no significant connections with either of the reading measures in year three. However, the change in RAN ability between years one and two correlated negatively with the change in technical reading abilities between years one and two (r = −0.475, p < 0.01). This means that as children’s ability to name letters speeded up, at the same time their technical reading abilities strengthened.

IV Discussion

Our results offer further support to the view that children with SLI are at risk of developing reading disorders (Simkin and Conti-Ramsden, 2006; Snowling et al., 2001; Webster et al., 2004; Whitehouse et al., 2009; Young et al., 2002). At the age of 10, only 20 percent of the children in our study achieved average or better than average results in technical reading. In reading comprehension the results were slightly better as 32.5 percent of the children achieved average or better than average results. On the strength of these findings, the children with SLI in this study are exposed, as a group, to considerable difficulties in continuing their learning and schooling. In Finland, many new subjects are introduced to the curriculum in the third and fourth year of school. In many of these the emphasis is on assimilating new information through printed text. This in turn, and depending on the teacher, relies heavily on children’s ability to read fluently and to understand what they have read. If special attention and support is not given to these children, many of them may find school work difficult and unrewarding. This may have far-reaching effects on their motivation and self-esteem, as noted by Arkkila et al. (2008).

Our results on the development of RAN skills in relation to the development of reading skills do not support the previous results of Babayiğit and Stainthorp (2010) and Verhagen et al. (2008), who suggested that the RAN abilities are the best predictors of further reading development. In our study, all RAN subtests correlated with technical reading skills in year one, but only the letter-naming subtest correlated with technical reading skills in year two and with reading comprehension in year one, and no RAN subtest correlated with reading skills in year three. In our study, children’s RAN abilities strengthened up to the age of nine, and by then they were as fast and accurate in this task as their peers. On the other hand, at the ages of eight and nine, the best correlations with reading results were found for the letter-naming subtest. This result is in concurrence with Lervåg and Hulme (2009) and Ziegler et al. (2010), whose studies show evidence of alphanumerical naming being the best predictor of reading development after the first school year. Based on these studies and on our results, in clinical use it would be prudent to administer only the letter-naming subtest when the emphasis is on studying reading development and the participant has already been exposed to letter names. In the case of younger participants, the object and colour matrixes may, of course, give evidence on the RAN abilities in general, but those results should not be taken as direct evidence on the development of further reading abilities.

Our results from the Boston naming test suggest that children with SLI made slight progress, but at the same time their ability to name objects fell further behind in comparison with their peers. This means that at the ages of eight to 10 other children were reading fluently and learning new words almost daily, while children with SLI made very little progress with their object-naming skills and reading skills. Of course, although our results are parallel in time, it does not automatically follow from that that they are causally connected to one another. Still, our results on the development of object-naming skills offer further support to Bishop and Snowling (2004) and Storch and Whitehurst (2002), in drawing attention to children’s semantic skills in connection with reading and reading comprehension. It is also easy to postulate that as children learn to read fluently they are exposed to more words and text and, as a result, also extend their vocabulary and naming skills.

All in all, our results offer mixed support to the models of reading mentioned in the beginning of our article. Our results in reading development lend support to Stanovich’s (1986) ‘Matthew-effect’ model, especially our results on development of technical reading skills. Our results on reading comprehension are somewhat more ambiguous and could also support Francis’s (1996) deficit model of reading. Therefore, further studies with longer follow-up times are still needed. What seems to be clear is that children with SLI do not conquer their possible reading difficulties as postulated by Stanovich et al. (1988) in developmental lag theory.

Our study has some limitations. The number of missing results in the reading tests remains a problem despite our efforts to take it into account in the statistical analysis. The number of missing results in the reading comprehension tests decreased over the three years, which is in concurrence with the expected development of reading skills. The same is not the case with the increase of missing results in technical reading test. We believe this might be due to the nature of the reading task at year three. At that point, the text used is longer and structurally more intricate than on either of the previous years. This may cause severe problems for a slow and unsure reader, to such extent that the child refuses the task or the teacher decides to spare him or her from the effort and subsequent failure.

Another limitation is that, at the time of this study, no standardized test measuring phonological abilities (PA) in this age group was available in Finnish, so this parameter was not studied. As phonological abilities have been shown to play an important role in reading development (Savage et al., 2005) and to be problematic in children with SLI (Vandewalle et al., 2012) it would be prudent to incorporate some measure of these abilities with measurements of reading and naming. This still remains an area of interest in studying an orthographically transparent language such as Finnish and reading development in SLI.

It must be noted that our results are from a relatively small group of individuals (n = 43). Given the small size of the Finnish-speaking population (5.4 million) and the problems in diagnosing SLI reliably (Hannus et al., 2013), we are happy to have been able to gather this many participants in our study. Statistically, a more reliable form of study would have been to use matched pairs as a comparison group to the children with SLI.

Footnotes

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This research was supported by a special governmental subsidy for health sciences (ERVA).

References

Acuña

Rodriguez

(2004) The treatment of missing values and its effect in the classifier accuracy. Classification, Clustering, and Data Mining Applications: Studies in Classification, Data Analysis, and Knowledge Organisation 4: 639–47.

Ahonen

Tuovinen

Leppäsaari

(2003) Nopean sarjallisen nimeämisen testi [Rapid Serial Naming Test]. 2nd revised edition. Lievestuore, ER-Paino: Haukkaranta school publications, Niilo Mäki Instituutti.

Arkkila

Räsänen

Vilkman

Roine

(2008) Specific language impairment in childhood is associated with impaired mental and social well-being in adulthood. Logopedics, Phoniatrics, Vocology 33: 179–89.

Aro

(2004) Learning to read: The effect of ortography. Jyväskylä Studies in Education Psychology and Social Research 237. Jyväskylä: University of Jyväskylä.

Babayiğit

Stainthorp

(2010) Component processes of early reading, spelling, and narrative writing skills in Turkish: A longitudinal study. Reading and Writing 23: 539–68.

Babayiğit

Stainthorp

(2011) Modelling the relationships between cognitive–linguistic skills and literacy skills: New insights from a transparent orthography. Journal of Educational Psychology 103: 169–89.

Bishop

Snowling

(2004) Developmental dyslexia and specific language impairment: Same or different? Psychological Bulletin 130: 858–86.

Bedore

Leonard

(2005) Verb inflections and noun phrase morphology in the spontaneous speech of Spanish-speaking children with specific language impairment. Applied Psycholinguistics 26: 195–225.

Bortolini

Caselli

Deevy

Leonard

(2002) Specific language impairment in Italian: The first steps in the search for a clinical marker. International Journal of Language and Communication Disorders 37: 77–93.

10.

Botting

Simkin

Conti-Ramsden

(2006) Associated reading skills in children with a history of Specific Language Impairment (SLI). Special issue, Reading and Writing 19: 77–98.

11.

Brizzolara

Gasperini

Pfanner

Cristofani

Casalini

Chilosi

(2011) Long-term reading and spelling outcome in Italian adolescents with a history of specific language impairment. Cortex 47: 955–73.

12.

Catts

Adlof

Hogan

Weismer

(2005) Are specific language impairment and dyslexia distinct disorders? Journal of Speech, Language, and Hearing Research 48: 1378–96.

13.

Cirrin

Gillam

(2008) Language intervention practices for school-aged children with spoken language disorders: A systematic review. Language, Speech, and Hearing Services in Schools 39: S110–37.

14.

Coady

(2013) Rapid naming by children with and without specific language impairment. Journal of Speech, Language and Hearing Research 56: 604–17.

15.

Durkin

Conti-Ramsden

(2010) Young people with specific language impairment: A review of social and emotional functioning in adolescence. Child Language Teaching and Therapy 26 105–21.

16.

Francis

Shaywitz

Stuebing

Shaywitz

Fletcher

(1996) Developmental lag versus deficit models of reading disability: A longitudinal, individual growth curves analysis. Journal of Educational Psychology 88: 3–17.

17.

Georgiou

Parilla

Papadopoulos

(2008) Predictors of word decoding and reading fluency across languages varying in orthographic consistency. Journal of Educational Psychology 100: 566–80.

18.

Hannus

Kauppila

Launonen

(2009) Increasing prevalence of specific language impairment (SLI) in primary healthcare of a Finnish town, 1989–99. International Journal of Language and Communication Disorders 44: 79–97.

19.

Hannus

Kauppila

Pitkäniemi

Launonen

. (2013) Use of language tests when identifying specific language impairment (SLI) in primary health care. Folia Phoniatrica et Logopaedica 65: 40–46.

20.

Hulme

Snowling

(2009) Developmental disorders of language, learning and cognition. Chichester: Wiley.

21.

Laine

Koivuselkä-Sallinen

Hänninen

Niemi

(1997) Bostonin nimentätesti: Suomenkielinen versio [Boston naming test: Finnish version]. Helsinki, Psykologien Kustannus.

22.

Leonard

(1997) Children with specific language impairment. Cambridge: MIT Press.

23.

Lervåg

Hulme

(2009) Rapid automatized naming (RAN) taps a mechanism that places constraints on the development of early reading fluency. Psychological Science 20: 1040–48.

24.

Lewis

Avrich

Freebairn

. (2011) Literacy outcomes of children with early childhood speech sound disorders: Impact of endophenotypes. Journal of Speech, Language and Hearing Research 54: 1628–43.

25.

Lindeman

(1998) ALLU: Ala-asteen Lukutesti [ALLU: Finnish reading test]. Oppimistutkimuksen keskus, University of Turku.

26.

Lindsay

Dockrell

(2012) Longitudinal patterns of behavioral, emotional, and social difficulties and self-concepts in adolescents with a history of specific language impairment. Language, Speech, and Hearing Services in Schools 43: 445–60.

27.

Lyytinen

Ahonen

Eklund

. (2001) Developmental pathways of children with and without familial risk for dyslexia during the first years of life. Developmental Neuropsychology 20: 535–54.

28.

Lyytinen

Erskine

Tolvanen

Torppa

Poikkeus

A-M

Lyytinen

(2006) Trajectories of reading development: A follow-up from birth to school age of children with and without risk for dyslexia. Merrill-Palmer Quarterly 52: 514–46.

29.

Nation

Snowling

(2004) Beyond phonological skills: broader language skills contribute to the development of reading. Journal of Research in Reading 27: 342–56.

30.

Nation

Clarke

Marshall

Durand

(2004) Hidden language impairments in children: Parallels between poor reading comprehension and specific language impairment? Journal of Speech, Language, and Hearing Research 47: 199–211.

31.

Savage

Frederickson

(2005) Evidence of a highly specific relationship between rapid automatic naming of digits and text-reading speed. Brain and Language 93: 152–59.

32.

Simkin

Conti-Ramsden

(2006) Evidence of reading difficulty in subgroups of children with specific language impairment. Child Language Teaching and Therapy 22: 315–31.

33.

Snowling

(2000) Dyslexia. 2nd edition. Oxford: Blackwell.

34.

Snowling

Bishop

Stothardt

(2001) Is preschool language impairment a risk factor for dyslexia in adolescence? Journal of Child Psychology and Psychiatry 41: 587–600.

35.

Spina

(2007) Statistics in pharmacology. British Journal of Pharmacology 152: 291–93.

36.

Stanovich

(1986) Matthew effects in reading: Some consequences of individual differences in the acquisition of literacy. Reading Research Quarterly 21: 360–406.

37.

Stanovich

Nathan

Zolman

(1988) The developmental lag hypothesis in reading: Longitudinal and matched reading-level comparisons. Child Development 59: 71–86.

38.

St. Clair

Durkin

Conti-Ramsden

Pickles

(2010) Growth of reading skills in children with a history of specific language impairment: The role of autistic symptomatology and language-related abilities. British Journal of Developmental Psychology 28: 109–31.

39.

Storch

Whitehurst

(2002) Oral language and code-related precursors to reading: Evidence from a longitudinal structural model. Developmental Psychology 38: 934–47.

40.

Tomblin

Records

Buckwalter

Zhang

Smith

O’Brien

(1997) Prevalence of specific language impairment in kindergarten children. Journal of Speech, Language and Hearing Research 40: 1245–60.

41.

Vandewalle

Boets

Ghesquière

Zink

(2012) Development of phonological processing skills in children with specific language impairment with and without literacy delay: A 3-year longitudinal study. Journal of Speech, Language, and Hearing Research 55: 1053–67.

42.

Verhagen

Aarnoutse

Van Leeuwe

(2008) Phonological awareness and naming speed in the prediction of Dutch children’s word recognition. Scientific Studies of Reading 12: 301–24.

43.

Wallenstein

Zucker

Fleiss

(1980) Some statistical methods useful in circulation research. Circulation Research 47: 1–9.

44.

Webster

Majnemer

Platt

Shevell

(2004) The predictive value of a preschool diagnosis of developmental language impairment. Neurology 63: 2327–31.

45.

Whitehouse

Line

Watt

Bishop

(2009) Qualitative aspects of developmental language impairment relate to language and literacy outcome in adulthood. International Journal of Language and Communication Disorders 44: 489–510.

46.

Young

Beitchman

Johnson

. (2002) Young adult academic outcomes in longitudinal sample of early identified language impaired and control children. Journal of Child Psychology and Psychiatry 43: 635–45.

47.

Ziegler

Pech-Georgel

Dufau

Grainger

(2010) Rapid processing of letters, digits and symbols: what purely visual-attentional deficit in developmental dyslexia? Developmental Science 13: F8–14.