Studies on lexical spelling acquisition have two aims: analyzing the cognitive mechanisms involved in this learning and understanding the processes on which they are based. Dictation and copying are known to be tasks leading to acquire orthographic knowledge. They are used by teachers and researchers. If some specific task models report on cognitive processes involved in verbal written production, they do not allow to get a comparative vision. This thesis aims at examining dictation and copying in the process of lexical orthographic learning. The model we propose includes the two tasks (dictation and copying) and the different underlying memories. We analysed the productions of children of 1st and 2nd grades under dictation according to 4 training conditions: dictation without training, reading aloud, reading aloud and copying, just copying. Children were asked to produce words or pseudowords. Words were selected according to their orthographic complexity and familiarity or their frequency, and pseudowords were more or less complex. Overall, experiments 1, 2 and 3 point out an effect of both grade and training. But, in experiment 4, copying doesn’t seem better than reading aloud: copying could be efficient only when combined with reading aloud. If the association of decoding and copying seems to be the best training to get orthographic knowledge, one might wonder about the presence of a graphomotor representation linked with lexical representation in Long Term Memory. This presence could explain why we need to write a word when we are sceptical about its spelling.
