Abstract
A disproportionally high number of children with specific language impairment (SLI) develop dyslexia. Yet it is hard to predict which individual child is at risk. This article presents a longitudinal study of phonological and early literacy development of 18 Dutch-speaking children with SLI, compared to 18 typically developing controls over a period of 2 years. These matched pairs of children were followed from 1 year before until 1 year after the start of formal literacy instruction. Children with SLI had problems with phonological awareness (PA) and verbal short-term memory (vSTM) in kindergarten. PA and vSTM in kindergarten were not good predictors for dyslexia in children with SLI, but rapid automatized naming (RAN) measured in kindergarten was strongly correlated with reading and spelling in both groups. We can infer that only SLI children who additionally failed on RAN in kindergarten developed reading and spelling problems at the end of Grade 1.
Notes
1It is relevant to analyze the seven children with SLI that did not go to G1. Means for nonverbal IQ, oral language, and all phonological tasks were lower for this group compared to the 18 SLI children who went on to G1, but only the differences for nonverbal intelligence, rhyme production, nonword repetition, and the vSTM composite score were significant. The decision to restart KG3 or move to a special school was jointly taken by the parents and teachers at the end of KG3 and depended on different factors, like the presence of emotional problems, doubts about cognitive skills, severe language, phonological and articulation problems, and personal factors. Hence, we can conclude that the 18 children with SLI who attended G1 are a more homogeneous group than the initial group of 26 SLI children. This could be the reason why matching correspondence for the restricted group of 18 pairs of children was better than for the 26 child pairs.
2The results of the group comparisons of the 18 pairs of children for the kindergarten data presented in and are in line with results of the group comparisons for the original group of 2 ×] 26 SLI and control children. Only on maternal educational level results were different: The original group of 26 SLI children did differ significantly from their 26 controls (p = .03).
a n = 18.
bMixed Model Analysis with age, nonverbal IQ, and parental educational level as covariates.
cTransformed scores with population average M = 100 and SD = 15.
a n = 18.
3The results of the phonological data presented in are similar to the group comparisons of the original 26 pairs of children. Only on four measures results of the MMA with covariates turned out to be different: The original group of 26 SLI children did differ significantly from their 26 controls on rhyme production (p = .007), end rhyme identity (p = .03), PArhyme composite score (p = .006), and the digit span task (p = .005).
an = 18.
bMixed Model Analysis with age, nonverbal IQ, and parental educational level as covariates.
cThese values were corrected for articulation accuracy.
a n = 18.
bMixed Model Analysis with age, nonverbal IQ, and parental educational level as covariates.
† p < .10.
*p < .05.
**p < .01.
***p < .001.
*p < .05.
**p < .01.
***p < .001.
a n = 9.
b n = 14.
cMixed model analysis with age, nonverbal IQ, and parental educational level as covariates.
dThese values were corrected for articulation accuracy.
† p < .10.
* p < .05.
** p < .01.
*** p < .001.