Variation in legal communication amounts to different use of legal language. Therefore, systematic recording and understanding of linguistic stratification is crucial both for the communication and the translation process. In this paper, we study existing classification models for legal communication and analyse their prerequisites. Based on the result that the presented models focus on different aspects of legal communication and are too general in their design for them to bring the complexity of legal communication across, we propose a more sophisticated model, which combines all the aspects of the analysed models. The model is discussed, using the language of German criminal law as an example. At the end, we explain the importance of the proposed model for translation-oriented analysis.