|Computers and Kinship
Kinship-related computer applications represent the earliest efforts by anthropologists to use computers in research. (Kunstadter 1963; Coult and Randolph 1965; Gilbert and Hammel 1966; Hackenberg 1967). Anthropologists have shown considerable interest in the use of computers for analysing kinship and genealogical data. Everyone seems to have 'too much' kinship information and computers have been suggested as a means to make this intractable quantity of data do some work (Gilbert 1971:135-137).
Just as kinship is one of the most complex areas of social anthropology, the computing resources required to assist in the analysis of kinship data are relatively complex and varied. Anthropologists have used computer based methods in:
1. creating and maintaining databases;
2. analysing genealogical data or models in connection with other data or models;
3. presentation of kinship and relational diagrams.
Computer based databases of relationships have been developed to draw out specific examples of relationships or to establish relationships between two or more people. This has been an application on the wish-list of many anthropologists for some time (Coult and Randolph 1965; Gilbert 1971; Chagnon 1974). There have been significant difficulties; issues include constraints on the size of the database (how many people and relationships can be represented), and how to specify the kinds of relationships in which you are interested without being 'swamped with types of relationships which do not interest' you (Gilbert 1971:136).
Computers have been used to help analyse genealogical data or models in context with other data or models. Relationship databases support further analysis based on other variables associated with the group or unit under study. This may be further supported by computer-generated diagrams (§6.5), or incorporated with simulation modelling (§8.1). Surprisingly, there has been relatively little published work (but see: Ottenheimer 1988; Davis 1990; Bagg 1991; Read and Behrens 1992; White and Jorian 1992) or even speculation (Gilbert 1971; Chagnon 1974; Davis 1984b) A notable exception has been the use of computer-based modelling to examine how demographic structure influences marriage behaviour and vice versa (Kunstadter 1963; Randolph and Coult 1968; Hammel, McDaniel and Wachter 1979; Dyke 1986). It may be that the apparent difficulty intimated in the past (Gilbert 1971: 137; Findler and McKinzie 1969) with using computers for flexibly analysing genealogical data has limited speculation in this area.
Computers have also been sparsely used for theoretical work in kinship, either with terminologies, or structure. Examples of such work include Read and Behren's kinship algebra expert system (Read and Behrens 1988, 1992) for terminologies, White's P-Graphs (1992) for examining larger structural connections between individuals and groups, Ottenheimer's work on modelling (Ottenheimer 1988), applications derived from graph theory (Garbett 1980; Hage and Harary 1991) or statistical studies aiming to develop behavioural correlates.
A number of programs have been written by anthropologists which can assist in the preparation of kinship diagrams for most computers (Ryan 1987, Ottenheimer 1985 and 1988, White 1992). Two basic approaches to automatic diagram support have been developed. Theoretically motivated diagrams present in principles of kinship structure using imaginary individuals. Actual kinship diagrams present examples of relationships and diagrams between the people in an area of study. It is easier to present an idealised kinship chart than to deal with actual populations. Besides the more obvious problems of deviance from conventions, there are problems of different kin cluster sizes, sheer quantity of people, and deciding precisely what kinds of relationships to diagram. By diagramming actual people in actual relationships, we are introducing both mechanical and conceptual problems. For example, in a physical layout, it is difficult to present both the spouse and sibling relationship clearly in the same diagram. This is true in non-computer generated diagrams as well, and the ordinary solution has been to introduce only relationships significant to the current discussion. (This is the computer based solution as well.) A mechanical problem is solved conceptually. Unfortunately, the reverse rarely proves true: conceptual problems are not solved by mechanical means, even when the machine is a computer. A clear conceptual model of the material is necessary for any analysis , traditional or computer-based . The first step in computer-assisted analysis is a clear development of the structural scheme we apply to a body of data.
Computers are useful tools for integration. Kinship-related studies satisfy (and are easily shown to satisfy) most of the requirements for computer-based analysis. Most computing methods require relatively clear schema for successful application, and many of the models and methods anthropologists have used for analysing kinship meet this criterion.
Next section: Conceptual requirements