Abstract
The writing skills of students in primary education show large defects. It has been shown that they do not practice frequently nor are they provided with enough guided instruction in employing writing strategies. In the Netherlands the computer program TiO-schrijven (Bok, 2003) aims to offer this support to pupils in
... read more
order to enhance their writing performance. It is examined experimentally whether the claims of TiO-schrijven can be empirically supported. With this purpose in mind the writing performance of students of the eighth grade in primary education is compared in three conditions, that differ from each other in intensity of training, with or without a computer and in scaffolded help in all stages of the writing process. By using more writing assignments per measuring moment it is possible to get a better indication of the quality of the writing samples. The texts were assessed by means of a so-called holistic scale scoring, each test has been rated by multiple raters. It has become clear that positive effects are detected when students write weekly using the computer program TiO-schrijven. Unfortunately, this study does not demonstrate whether any of the didactic principles in TiO-schrijven are individually tenable. Global judgments about text quality cannot indicate precisely what features of the writing-process have caused those effects. As part of an analytical assessment an extra condition was added. This was done because data of a national study were available. By reviewing this analytical assessment it could be asserted that the sample is representative of the total population. The measurement of writing is problematic not only due to the variability of writers (i.e. a writer may write a very good text on one occasion, but a poor text at the other occasion), but is also due to the variability of raters. If different raters give different ratings to the same product, not only reliability, but validity issues are at stake as well. Therefore it has been suggested that many of these problems could be circumvented by measuring writing with a multiple choice test. However, such a multiple choice test appears to correlate poorly with observed global writing scores. Hence, we have to conclude that there is serious doubts about the validity of multiple choice tests to evaluate students’ writing ability. The computer program TiO has shown to be an effective approach to the teaching writing. The writing performance of students who have worked with this computer program during one school year is very hopeful. Future research should focus on strengthening the writing pedagogy as well as on the teacher education and training. Given the importance of good writing for students’ school success it is important that future research on designing effective writing lessons will be performed. It is possible that the computer in general and TiO-schrijven in particular can contribute to so-called 'blended learning': the teacher scaffolds the writer in the communicative situation; TiO-schrijven can be employed as a crafty tool to write
show less
