At the Educational Research Centre (S.C.O.) in Amsterdam, a study determined the applicability and construct validity of ratings of speaking performances by examining tape-recordings of subjects in four dimensions. Subjects were 200 pupils of 11 and 12 years of age, and performances on four different oral tasks were investigated. The rating dimensions--defined as main functions of speaking--were reference, delivery, fluency and articulacy. Each main function was represented by a rating category that was selected in a pretest. Inter-rater reliabilities, correlations between panel ratings--within and between tasks--were calculated and some hypothetical models for correlations between ratings were tested. Results indicated that performances on two oral tasks were rated on the four dimensions in a meaningful way. Performances on the other tasks were rated in such a way that only reference and articulacy were meaningfully differentiated. Furthermore, ratings on articulacy correlated highly between tasks, whereas ratings on reference correlated relatively weakly between tasks. (Five tables of data are included, and 12 references are attached.) (RAE)