During the 1960s and 1970s, students of kinship became increasingly uneasy about the gap between formal terminology-and-genealogy-based models and data on actual behaviour. This gap–sometimes described as the problem of relating ‘prescriptive’ and ‘statistical’ models–was an important factor in Schneider’s rejection of the structural and cognitive traditions, and their subsequent near abandonment by Anglo-American anthropology. However, these developments did not resolve the problem so much as simply refuse to address it. The need for a better understanding of the relation between terminology and behaviour is still there, nowhere more so than in Europe, where quantitative historians and sociologists have revealed major macro-regional differences in kinship practices, which are associated with distinct patterns of kinship terminology.

This is where Keesing comes in. In his contribution to a 1972 volume celebrating the centenary of Morgan’s “Systems”, he, too criticized the existing work on formal models–but did not advocate abandoning it. On the contrary, he argued that it should be extended and deepened–setting aside simplistic assumptions of a direct correspondence between terms and roles in order to model the complex social and semantic processes that integrate kinship with the rest of social life. In this article, I return to Keesing’s agenda and propose a modeling approach that would fit some of the European data.