Cette étude vise à caractériser les analogies utilisées en classe pour apprendre l’arithmétique à l’école élémentaire et à mettre en évidence les mécanismes d’élaboration de ces analogies en situation écologique. Vingt séances de cours de mathématiques en cycle 2 (6 à 8ans) et en cycle 3 (8 à 11ans) ont été filmées, puis analysées en utilisant un codage qualitatif proposé par Richland et al. (2004). Les analyses montrent que des analogies sont spontanément élaborées verbalement en classe dès le cours préparatoire (« premier grade ») ; elles sont principalement élaborées à l’initiative des enseignants. Ces analogies portent à la fois sur des concepts mathématiques et sur des procédures de calcul ou de résolution de problèmes. La nature des situations cible et source varie suivant le niveau des enfants tout comme le contexte d’utilisation des analogies. Ces résultats sont discutés dans une perspective développementale.

According to experimental studies, analogical reasoning play a central role in mathematics learning and problem solving. However, little is known about how analogy is used in everyday instructional contexts. The aim of this study is to characterize the analogies naturally used in mathematical lessons in elementary school in order to highlight the mechanisms of generation of these analogies. Twenty mathematical lessons were videotaped, 10 sessions in first and second degrees (“cycle 2”, 6–8-year-old), 10 sessions in third, fourth and fifth degrees (“Cycle 3”, 8–11-year-old). Qualitative codes proposed by Richland et al. (2004) were used for the analysis: the study examines patterns of teacher–learners’ participation, source and target construction, and contexts for analogy production. These data show that verbal analogies are spontaneously generated from the first to the fifth degree. They are mainly generated by teachers and completed by children. These analogies are both used to discover mathematical concepts and to apply procedures. The nature of analogy generated differs according to the level of children. In first and second levels, sources are concrete familiar situations. In higher levels, most of sources are decontextualized problems. Context of use analogies also varies depending on the level of children. In higher level, source and target construction is more often related to whether the analogy is in response to students’ needs for help. These results are discussed in a developmental perspective.