Thursday, October 27, 2011

Precursors to numeracy in kindergartners with specific language impairment

Kleemans, T., Segers, E., & Verhoeven, L. (2011). Precursors to numeracy in kindergartners with specific language impairment. Research in Developmental Disabilities, 32, 2901-8.


Kleemans et al. examined whether children with SLI significantly differed on measures of numeracy in comparison to typically developing children and which precursors were more closely related to numerical processing skills in children with SLI. A group of 111 typically developing children and 61 with SLI completed measures of working memory, naming speed, phonological awareness and grammatical ability, as well as measures of logical operations, numeral representations and numeral estimations. The researchers found that phonological awareness and grammatical ability predicted logical operations and numerical representations. Furthermore, there was an interaction between group and naming speed, with naming speed being significantly correlated with logical operations and numerical representations in the SLI group, but the typically developing group.


These findings are consistent with the Triple Code Model proposed by Dehaene et al. (2003). This model predicts that depending on the tasks, there are three distinct systems of representation that may be recruited: the quantity system (nonverbal), a verbal system (where numerals are represented lexically, phonologically and syntactically) and visual system. The verbal system is associated with activating the left angular gyrus (located in the parietal lobe), which has been found to be associated with arithmetic fact retrieval, but also other language mediated processes such as verbal STM and reading. The findings of a relationship between language abilities and the verbally based numerical tasks in the Kleemans et al. study are consistent with the verbal numeracy system proposed in the Triple Code Model.


Although the paper is consistent with previous findings, the hierarchical regression analysis reported was confusing. Furthermore, multiple t-tests were conducted increasing the chance of finding a type I error. The correlation analyses would have been stronger if partial correlations controlling for age were reported. As well, the numeracy tasks that measured numerical operations and numerical representations were very similar to one another and relied heavily on language processes to perform. Given the fact that the numeracy tasks were not independent of language skills, the results are not surprising.


These findings may have clinical and diagnostic implications. Although further research is necessary, the rapid naming may be related to number deficits in children with SLI. However, given the statistical methods used, the results should be interpreted with caution.


Blogger: Stephanie Bugden is our guest blogger and a PhD candidate working in the lab of Dr. Daniel Ansari, Canada Research Chair in Developmental Psychology at the University of Western Ontario. Stephanie’s research examines the developmental trajectories of cognitive impairments in children with math learning disabilities. Stephanie currently holds the record for the shortest time between finishing our lab meeting and writing her post for this blog!

Tuesday, October 4, 2011

Domain-specific treatment effects in children with language and/or working memory impairments: A pilot study

Wener, S.E., & Archibald, L.M.D. (2011). Domain-specific treatment effects in children with language and/or working memory impairments: A pilot study. Child Language Teaching and Therapy, 27, 313-330.

Archibald & Joanisse (2009) identified three groups of children: (1) Those with a Specific Language Impairment (SLI), an impairment in language development but not working memory; (2) those with a Specific Working Memory Impairment, an impairment in working memory but not language; and (3) those with both language and working memory impairments (L&WMI). Given these separable groups, it would be reasonable to hypothesize that these children may respond differently to treatment aimed either at improving language or working memory. It was the purpose of this pilot study to evaluate this hypothesis.

We employed a single subject design with 9 participants, 3 with SLI, 2 with SWMI, and 4 with L&WMI. Performance was measured using 3 probes (picture recall, sentence formulation, geometric puzzle completion) throughout the baseline phase (2 wks), treatment phase 1 (4 wks), no treatment (4 wks), and treatment phase 2 (4 wks). Standardized tests of language and working memory were completed before and after each treatment phase, and at 4 months post treatment. The treatment in the 2 phases involved 4 tasks (word recall, n-back, story retelling, memory updating) but differed in the strategies taught. For the language-based intervention, verbal strategies were targeted whereas for the working memory-based intervention, imagery strategies were taught as a memory strategy. Five of the participants did the language-based intervention first followed by the working memory intervention.

Results revealed that performance improved on the picture recall probe after the language-based intervention, and on the geometric puzzle completion probe after the working memory-based intervention. No reliable effects were found for the sentence formulation probe. Improvements of 10 points on standard scores were considered clinically significant. Based on this criterion, 6 of 7 of the children with a language impairment (SLI or L&WMI) showed an improvement on a language measure and 3 of 5 (SWMI or L&WMI) on a working memory measure at 4 months post onset.

This study provides some preliminary evidence of a domain-specific treatment effect in that performance on a language task improved after language-based intervention and on a visuospatial task after working memory-based intervention focusing on imagery strategies. The findings also suggest a profile-specific effect in that children with a language impairment tended to improve on a language measure and children with a working memory impairment on a working memory test. Given the small sample size, these results must be interpreted with caution.

Blogger: Lisa Archibald