Computational approaches to the study of post-marital residence : a thesis presented in partial fullfilment of the requirements for the degree of Doctor of Philosophy in Statistics at Massey University, Manawatu, New Zealand

Abstract

Post-marital residence is the location taken by a couple after marriage. It is often a cultural practice, a choice based on tradition. Through its effect on family and family structure, post-marital residence influences important concepts and practices such as inheritance of property, status of men and women, initiation rites and tracing of descent. Due to its long-reaching effects and the fact that it is relatively easy to observe, through a short discussion with informant or by collecting information about marriages, post-marital residence has become an important subject of study for many anthropologists. In the past, post-marital residence was predominantly explored through association studies. However, factors influencing post-marital residence often exhibit unclear causative direction. This makes their study using association analyses difficult. This thesis will explore post-marital residence using three different computational methods: an evolutionary approach based on language trees; a data-mining approach that finds clusters of societies in an ethnographic database; and an agent-based model of warfare-induced residence change. These three methods enable exploration of post-marital residence from significantly different angles, which enables me to describe a much more complex and balanced picture. I find no evidence for the existence of global patterns of residence change. In fact, even language groups with similar demographic histories differ significantly in their patterns of residence evolution. However, I find strong evidence for the existence of more localized patterns. Based on data that describe societal properties such as the prevalent type of subsistence, sex taboos or type of housing, societies can be clustered into groups, with some groups being almost exclusively formed by societies with a single type of residence. Finally, I find that while warfare is able to induce a change of residence, it does this only when a significant portion of the society is under warfare pressure. However, warfare can also be a catalyst when another factor influencing residence change is present. My results suggest that more localized patterns should be explored. Based on the grouping of societies identified in this work, one should not assume that because two societies have the same residence state that similar factors must be in play. In fact, a multitude of factors could induce change into a specific residence state under different conditions. Thus, the factors for residence change should be explored on a case-by-case basis and societies with similar histories and pressures should be grouped together and investigated instead. Societies where the change of residence was induced by warfare could be one such group. Results from the agent-based model can help to specify the exact conditions required. Computational-based approaches enable new and interesting points of view on classical anthropological problems. However, they are limited by the existence of data and a functional knowledge of societies and cultures. These are often lacking, at least in a programatically accessible form. Thus, developing better and more accessible databases and knowledge banks with a mechanistic description of cultural concepts should be a primary future focus for anthropology. Taken together, the results of the three approaches shown in this thesis form a strong statement regarding how various factors influence a change of post-marital residence. This provides a proof of concept of benefits for tacking classical anthropology questions with computational tools. It will hopefully work as an invitation for collaboration between the two research areas.