Two studies are described which presented children with computer-based writing tasks. One compared observations of pairs and singletons within a classroom setting, with P5 children as subjects. The other video-recorded $2 learning difficulty pupils working in partnership or with a teacher and compared task performance under both dyad conditions. Focussing specifically upon evidence relating to revision during and following composition, both studies support a general observation in the field of writing research that children lack diagnostic strategies. However, there is some evidence that this finding is affected by writing task so that the problem may be in transfer, rather than lack, of skills. Developing useful teaching software to remedy this depends on an understanding of writing processes as writing competence develops; a programme for further work in this field is outlined.