Abstract. Conceptual graphs (CGs) share with FOL a fundamental expressiveness limitation: only higher-order logics allow assertions of properties on predicates. This paper intends to push back this limit by reifying underlying relations of CGs (is-a, a-kind-of, referent) into first-class objects (i.e. nodes) of an equivalent, labelled graphs (LG) model. Benefits of this reification, applied on a subset of CGs, namely simple graphs and rules of form "if G then H", are discussed in terms of expressiveness, succintness and robustness. We show that using the LG model as an interpreter allows us to improve and extend the results in [2].