Cand.ed. Marianne Klem

Title of dissertation:

Language assessment as a basis for identifying Preschool children at risk for language impairment. A model-based approach.

The topic of this article-based dissertation is language assessment as a basis for identifying preschool children at risk for language impairment. There are large gaps in our knowledge base on language assessment of preschool children. The purpose of this dissertation was to develop evidence-based knowledge on language assessment of preschool children, in order to strengthen the evidence-base for this practice within a Norwegian context.

The dissertation builds upon three articles, all of which have a quantitative methodological approach with a non-experimental design. Structural Equation Modeling techniques (with latent variables) was used as the main analytical approach in all three articles. The dissertation highlights different challenges related to the measurement of language ability and language assessment as a basis for identifying risk for language impairment in monolingual, Norwegian preschool children, aged four to six years. The articles have a particular focus on the language screening tool “SPRÅK4” [LANGUAGE4](Horn & Dalin, 2008) and an age-appropriate adapted version of the sentence memory test from the “Språk 6-16” language screening test (Ottem & Frost, 2005). Overall, the articles present empirical evidence in support for decribing the structure of language ability in preschool children in terms of a model with a single general (latent) language factor that explains a wide range of diverse Language measures. This general language factor shows a strong longitudinal stability in children from age 4 to 6 years, - a trend which is evident across both low and higher-performing groups of children. The results related to SPRÅK4 give reason to further pursue a validation of this language screening tool. At the level of latent variables, we present evidence in support for satisfactory construct validity of SPRÅK4. Furthermore, the use of latent class analysis on SPRÅK4 at item-level to identify four year olds at risk for language impairment appears to be a promising approach, which should be further investigated to determine its applicability in practice.






Publisert 5. feb. 2015 14:48 - Sist endret 6. feb. 2015 10:47