Game-based learning is one efficient pedagogical concept that uses game principles to incite learners to engage into learning activities. Learning games are commonly known as digital environments. In the mean time, new technologies have been increasingly developed, thus providing new perspectives in game-based learning, particularly, mixed reality technologies that merge both real and digital worlds. For instance, they are widely used in mobile learning or learning with tangible interfaces. For the latter, mixed reality technologies make collaboration easier and provide better feedback to users. We present in the paper a brief study on the state of the art of mixed reality technologies. While the technologies have been experimented in educational settings or in games, they are used only in few learning games. Some research efforts have proven positive outcomes of the latter in learning even they are not widely applicable. Based upon the contributions of mixed reality in learning games, we point out that means employed by designers are as crucial as the pedagogical objectives. Therefore, they have to be taken into account during the design process. Our research efforts aim at providing tools and methods to support the design of mixed reality learning games (MRLG). One of the first steps during the instructional design is to write the learning scenario. However, there is no universal method to be used in the design process. Thus, in the second section of this paper, we analyze the ability of existing pedagogical specifications to model mixed reality learning games scenarios. In this view, we compare IMS-LD, LDL and ISiS, which all intend to assist pedagogical designer and teachers in the design of new pedagogical activities and in the formalization of existing ones. Design process of MRLG includes game elements and mixed reality technologies design, particularly in the pedagogical scenario writing. Meanwhile, the specifications mentioned earlier do not fulfil our needs: they all allow a representation of a pedagogical scenario in a workflow, but in our case, a more detailed description of the workflow is needed. The specifications are not suitable to describe the used technologies or the way they are used. Regarding the fun factors, none of the formalisms includes the description of rules and neither game principles nor game objectives. The last part of the paper discuss different proposals: an extension of a pedagogical specification, a combination of existing specification with a task model for more detailed description, and a new pedagogical specification for mixed reality learning games.