Test validity in morphosyntactic measures for typical and SLI incipient Spanish–English bilinguals

Abstract

This study will demonstrate that group differences on a morphosyntactic measure used for the identification of specific language impairment (SLI) do not guarantee validity for diagnosis and tracking, and will exemplify this with a case study of the Spanish version of the Clinical Evaluation of Preschool Language–2 Estructura de Palabras subtest for SLI and typically developing (TD) Spanish–English incipient bilinguals. The use of standardized tests beyond identification, for planning intervention and monitoring progress, requires that validity evidence be provided if and when extended purposes are involved. Results show that although group differences on total subtest score were observed, some of the items exhibited a distinct pattern, that is, high standard deviations and lower rate of correct responses. A careful analysis of these items points to design problems and content-validity deficiencies as causes. Solutions for the different problems are proposed. We conclude on the importance of a careful linguistic analysis to support test validity in order for the test to serve its broader purposes.