Family Ties

Kinbank – a new database for kinship terms from around the world

The words we use to describe are family are such a visceral part of life that the thought that it might be organised differently is rarely considered. For example, if you’re reading this you speak English and are likely to live in a western industrialized society. In this context mother and father have the strict cultural definition of a monogamous couple raising their children in a household separate from their parents. Most Europeans hold similar views on family, such as the responsibilities of the parents to their child and their nuclear family’s role in the wider community. This ‘typical’ style of family is so embedded in society, it is often thought to reflect genetic relationships and not the result of cultural beliefs. But this is objectively not true. For example, in English we use different words for female and male siblings, but the same word for female and male cousins, despite the average relatedness being the same between these pairs. Family organisation is cultural, and for English speakers, sometimes gender distinctions are important, and other times not. In the UK, these norms of family organisation transcend the cultural realm into law, dictating custody rights, inheritance, and even a child’s nationality. This means cultural differences in understanding how and why family differs can have serious implications when dealing with an increasingly globalised world.

While in contemporary western society who can become a parent is rightfully becoming more varied, family, or kinship, organisation isn’t. Across cultures however, family organisation is amazingly diverse. Few Europeans would consider that the term ‘mother’ should extend to include a mother’s sister, or ‘father’ to include a father’s brother. Aunts and uncles are a different category of relative, and that distinction is important. But in Tonganthis is precisely how family are classified, where the term “fa’e” is used to refer to both a mother and her sisters, and “tamai” for a father and his brothers. What is different in Tongan society that makes this classification important? My research looks to investigate these kinds of questions, and also quantitatively explore how variable kinship really is, what makes certain distinctions important, and whether those things have the same impact in different parts of the world.

Anthropologists have been studying kinship and what social structures determine kinship organisation for over 100 years and have developed numerous theories. For example, if society places importance on the nuclear family, then you might give specific terms for the people within a nuclear family and lump together more distant relatives. This is the case in English, which is categorised as an ‘Eskimo-type’ system1. Kinship systems have been categorised into 6 types, which focus on how siblings and cousins are organised and are name dafter the society in which they were first identified(early anthropologists didn’t study themselves very closely!). I collected a number of theories that suggest a particular social structure predicts the use of a particular terminology type. Theories tend to focus on marriage practices (who can you marry, and how many people), patterns of descent (do you align with your father’s family line, or mother’s, or both?), and patterns of residence (when you marry, do you move to the husband’s or wife’s house, or start a new house?). These social structures are thought to influence which kin are brought together and which are separated, and that in turn changes who needs to be directly identifiable and who doesn’t. We also look at whether the relationships between social structure and family organisation are the same in Austronesian, Bantu (sub-Saharan Africa), and Uto-Aztecan (North America) cultural groups, which exist in different environments and have different time-depths. This is to test whether social structures always have the same effect on kinship organisation, or whether the effect is specific to language groups.

Most previous work in this area suffers from an annoying statistical hitch, named Galton’s Problem. Imagine, for instance, two neighbouring societies that both traditionally start a new household after marriage; it is more likely these two societies were once the same and maintained that tradition, rather than it being two independent incidents. In the past it was difficult to include the historical relationships between groups into any formal analyses, but now statistics has developed sufficiently for them to be included. When we incorporate the relationships between societies in our models we find no universal patterns of change and actually, many of the proposed relationships between kinship organisation and social structure are not true.

There are two possible reasons for this. Firstly, and simply, the theories were wrong and perhaps were the result of poor statistical understanding. This is a reasonable conclusion, since this has occurred in other domains which suffered from the same problem. However, to be thorough, a second possibility is to consider that the way kinship systems are being categorised is incorrect. As mentioned above, we use a 6-piece kinship typology, and within each category we know there is variation. For example,we know that most ‘Hawaiian’ systems in the Pacific distinguish older siblings from younger siblings, but that this doesn’t really happen in North American ‘Hawaiian’ systems. Is this an important distinction and why does it happen? Despite the similarities between societies’ kinship systems, there are also differences – but with the current typology, we can’t tell which are important and which are not. So, the next project is to build a database of kinship terms, rather than classifying societies by system type. This will allow us to test whether the typology of kinship system is appropriate and explore more granular models of change and their relationship to other parts of society.

Our database, KinBank, currently holds 150 different kin relationships for over 1000 languages across the world. The collection includes terms ranging from grandparents to grandchildren (G-2 to G+2 for the kinship nerds), parent’s siblings, and their siblings’ children. We also collect terms for relatives by marriage (nuclear kin’s affines). Where they exist, we also record terms that differ for sex of speaker (where men and women speakers have different kinship organisation), relative age (e.g. different words for older and younger brother), and by age of connecting relative (father’s older brother’s children vs father’s younger brother’s children). Each term has a referenced source, and each language is linked to other cultural/linguistic databases, such as D-PLACE and Glottolog. Which we hope means KinBank data will be in a format usable by scholars asking all sorts of questions.

KinBank is a key part of the European Research Council project VariKin. While it is work-in-progress, our current collection already contains 14 major language families. We’ve focused our data collection on societies linked to language phylogenies, enabling the use of phylogenetic comparative methods to further explore the problems discussed above. We hope to have our first release in Summer 2019 (UK). For future updates on KinBank, visit 

1 The term “Eskimo” is considered derogatory in some parts of the world. I do not intend any derogatory meaning, but instead am following the current nomenclature. See for more info.