Fine-grained assessment of children's text comprehension skills

Abstract

Text comprehension is an essential skill for achievement in personal, academic, and professional life. Therefore, it is tremendously important that children's text comprehension skills are actively monitored from an early stage. Text comprehension is, however, a complex process in which different reading abilities continuously interact with each other on the word, sentence, and text levels. In educational practice, various tests are used to measure these different reading abilities in isolation, which makes it very difficult to understand why a child scores high or low on a specific reading test and to adequately tailor reading instruction to the child's needs. Dynamic assessment has the potential to offer insights and guidance to teachers as cognitive processes that are important for learning are examined. In dynamic tests, students receive mediation through instruction when answering test questions. Although computer-based dynamic assessment in the reading domain holds potential, there is almost no support for the validity of dynamic measures of text comprehension. The aim of the present study is to determine design principles for the intended use of computer-based dynamic assessment of text comprehension. Based on the dynamic assessment literature, we developed a model for assessing the different reading abilities in conjunction. The assumption is that this model gives a fine-grained view of children's strengths and weaknesses in text comprehension and provides detailed information on children's instructional needs. The model was applied in a computer-based (fourth-grade) reading assessment and evaluated in practice through a three-group experimental design. We examined whether it is possible to (1) measure different aspects of the reading process in conjunction in order to obtain a full understanding of children's text comprehension skills, (2) measure children's learning potential in text comprehension, and (3) provide information on their instructional needs. The results show that while the model helped in explaining the children's text comprehension scores, unexpectedly, mediation did not clearly lead to progress in text comprehension. Based on the outcomes, we substantiate design principles for computer-based dynamic assessment of text comprehension.