This communication reviews different methods used in several research studies investigating learning processes during teaching sessions (Givry 2004, Elbilani & Le Maréchal 2005, Kucukozer, 2005). The theoretical frameworks and the consequence on the methodological choices are discussed to compare the common aspects and differences and analyse the respective interests and limits of the methods. This review is based on two main perspectives: (1) different theoretical approaches to deal with knowledge and meaning and (2) the levels of granularity with which knowledge and time are taken into account. The studies involved in this review have several common characteristics and differences: * The length of the data collection during the learning process lasts from few hours to several weeks. It infl uences the kind of knowledge that can be attributed to the learner and the way of splitting the transcription into episodes. * Various levels of granularity are implicitly or explicitly used to analyse the elements of knowledge attributed to a learner or to a group (students and teacher) in each episode. They lead to different “re-compositions of these elements” to characterize the learners' evolution or the learning processes. * The analysis of the taught knowledge can either mainly consist in listing the concepts involved in the teacher's production or be a re-construction of the teachers' or learners' scientifi c ideas based on their productions. The fi rst results show that attribution of meanings to the learner's productions is infl uenced by the analysis of the taught knowledge and by the way the students' initial conceptions are taken into account.