As a step in a project whose final goal is to propose a Controlled Natural Language for requirements writing at CNES (Centre National d'Études Spatiales), we intend to build the grammar of the textual genre of the requirements. One of the main issues faced when analyzing our corpus is the (sometimes subtle) difference between the terms and syntactic structures pertaining to the genre and those linked to the domain (in our case, the development of space systems) – a difference that is generally not taken into account by automated tools. In this paper, we present a methodology aimed at detecting candidate terms and textual patterns specific to the genre by combining results obtained from a terminology extractor and a data mining tool with a validated resource in use for indexing documents at CNES. The results are then illustrated by a selection of examples from our corpus.