Design patterns for learning are about articulating, testing and sharing the principles of problem solving in the educational context. In this way, multiple patterns are developed to solve common problems, described in various pattern language formats. Therefore, this work is about characterizing and establishing functional relationships between the components involved in the construction of a learning design pattern. The research was carried out in a Delphi study of double round individual non-contact between experts (n = 14), from different areas (educational technology, e-learning, distance education, higher education), selected through the coefficient of expert competence. The main result is a pattern language composed of attributes, components, and groups of processes. It is concluded that the language obtained allows externalizing knowledge from its consistent and shareable structure, which makes it suitable for use in different scenarios that require a formal pattern language and facilitates its implementation in online learning contexts.