Self-explanations are verbalizations that readers give to themselves while reading a text, in order to better understand it. Our implemented system is designed for automatically analyzing self-explanations in order to allow teachers to better grasp the comprehension of pupils of the previously read materials. Our method uses specific natural language processing techniques for French language and it is conceived for use with primary school pupils. Furthermore, we have integrated a word-based heuristic in order to measure similarity between the initial texts and the pupil's verbalizations.