Abstract
The Reynell Developmental Language Scales (RDLS) have been used extensively as an instrument in clinical settings. However, little is known about its construct validity. Essential for its use both as a clinical and a scientific instrument is its discriminant validity as the scales involve separate assessment of language comprehension’ and ‘expressive language’. Two studies were carried out to investigate its discriminant validity within a longitudinal design, the first an ordinary least‐squares path analysis. The second applied a data‐analytic strategy consisting of a two‐step procedure. First, a measurement model based on four prespecified linguistic dimensions derived from the facet structure of the well‐known Illinois Tests of Psycholinguistic Abilities (ITPA) was tested as a specific case of a general covariance structural model. Second, three of the four originally hypothesized dimensions were inserted as latent dependent variables in a multivariate multiple regression analysis. Both studies supported the hypothesis that the RDLS measure two correlated but distinct language abilities.