We must first define the conceptual terms to which we
intend to apply a computer-based analytic procedure. A 'structural schema' is a
set of objects, their properties, and the relationships between objects and
properties. The conceptual terms must be determined, in whole, by scientific
requirements rather than computing requirements. The structure and definition of
conceptual terms are independent of whether or not a computer is to be used. It
may be the case that an external schema which is adequate to represent a given
conceptual schema will be difficult to implement using existing or custom
purpose-designed computer applications. Using a computer may not be useful in
analysing that domain. Computers are not applicable to all problems in part or
Most simple-mindedly, anthropologists do anthropology, and
they do it using conceptual schemes determined by disciplinary conventions. The
first consequence of this is that anthropological structural schema take
precedence over computer structural schema. If the fit is good, the
anthropologist has an efficient and useful tool. The second consequence is that
the anthropologist needs at least enough appreciation of computer structural
schema to recognise problems (and felicities) in the fit of the two schema . In
the domain of kinship, anthropologists have traditionally conceived a schema
which can be usefully and productively represented on a computer. The
applications discussed in mainly use existing schema associated with genealogical
modelling in social anthropology (Barnard and Good 1984:23-26).
The structural aspects of the genealogical model are
relatively uniform; we record the genealogical connections between individuals as
a means of describing some aspect of social relations which has been of value in
social anthropology for describing social structure. The conceptual schema
influences the collection and interpretation of the structural model which
results from establishing the genealogical connections.
A computer application may provide a generalised record
keeping and relationship-establishing function common to all sorts of schemata.
An anthropologist may simply want to establish links between individuals in a
population using defined kin categories. This will influence and simplify
preparation of reports, error checks and audit functions. This is a generic
function of the computer, and computer support functions required are
However, a computing application may include functions which
are specifically related to the conceptual schema intended. If there are special
interpretations or assumptions included in the anthropological investigation of
the material, specific support requirements in the computer applications may be
required. For example, if an applications is simply required to establish links
between individuals in a population the support requirements for any particular
schemata are limited. If we wish to perform an analysis of a more specialised
sort, such as calculating inbreeding coefficients on a purported biological
pedigree, specific support requirements are required
For this reason it is usually best to define requirements in
terms of generic functions and specialised functions. All forms of genealogical
analysis require the establishment of different sorts of links between
individuals, so this is a generic function. Very few analytic models require that
these links be defined as biological links, as is required to legitimately
calculate inbreeding coefficients. However we do not necessarily require two
independent applications for these two purposes. One generic application can
calculate the genealogical connections and the results input to another more
specialised application which assigns the specialised interpretations and
operations on the proposed genealogical structure.
In terms of a genealogical study, the conceptual schema would
consist of definitions of people, the kinds of links recorded and maintained and
associated theoretical statements about how these interact or are defined in
terms of each other. Barnard and Good present one conceptual schema based on
common genealogical models This consists of primitive objects, people, a set of
properties of people (sex, relative age), a primitive set of relationships
between people (F, M, B, Z, S, D, H, W), a set of rules for building compound
relationships (FF, MM, FB, MZ etc.) based on primitive relationships via other
people. (Barnard and Good 1984:3-9)
This might appear to be a great deal of detail, but it is
nothing more than any anthropologist applies to a genealogical model. However, in
ordinary discourse much of this detail can be left unstated simply because it is
so conventional. Computer programs will, however, require some form of this
knowledge of conventions be incorporated into the program. It is important to be
able to specify this detail at the conceptual level, especially if you are
attempting to specify your problem to a programmer. One particularly problematic
aspect of genealogy programs and computer programmers (who are not also
anthropologists), is that the unsaid detail will often be inferred by the
programmer since 'every one' understands genealogies and family trees and such.
Their inferences, derived from their particular genealogical models, will rarely
match your requirements.
Even if you intend to write the program yourself, it is a
good idea to record this information. It will provide a base to which you can
compare the eventual computer implementation of the schema. As with other
analytic methods, there will be simplifications and compromises, and it is useful
to identify explicitly where there are, since deviations from the conceptual
schema impose limitations for interpretation of the results.
In specifying the conceptual schema you can use formal
looking statements and diagrams, such as in Figure 6.1. This can be rather
difficult and cumbersome for complex definitions, and is not strictly necessary.
Many people will be more comfortable clearly writing the schema using their
native language with examples of each type of definition or rule and a few simple
diagrams where useful (Table 6.1). As long as the specification is clear,
relatively complete and relatively unambiguous just about any form will be
satisfactory and useful.
This presents an example of informal specifications for a
genealogical model. This model will be customised for the group involved: in my
own work, birth order, multiple marriages, and other information is significant
in analysis. This information will be included in your informal
In summary, theoretical characterisation of the research
problem defines conceptual schema. If scientific representation of the area under
study appears to be adequately represented in the computer, we outline precisely
what general and specific functions a computer can perform in the analysis of the
material. If we decide that computer based analysis is possible and useful, we
translate our information into a form computers can use (programs) or computer
programmers can understand (detailed specifications), or look for commercially
In the next section, we will discuss more completely how this
is done, and the advantages and disadvantages to each approach. (The optimistic
might think that simply buying a software package is easy. It's not.)
Informal specification of conceptual schema for
- The object of the schema is to model people
and relationships between people. People will be represented by a set of
properties, specifically, each person is an individual entity with properties of
gender, age, and generation.
- Relations between people are of two
- Basic relationships.
- Basic relationships correspond to
Parent, Child, Sibling and Spouse, and where necessary the gender marked
relationships which correspond to the categories Father, Mother, Son, Daughter,
Brother, Sister, Husband and Wife, relative to a given individual denoted as Ego.
These relationship are defined in terms of indigenous social judgements.
- Each relationship has a reciprocal
relationship; e.g. if an individual has a relationship Father relative to an Ego,
Ego has the reciprocal relationship Son or Daughter to that individual. This
reciprocal relationship is different in the case of Parent-Child, the same in the
case of Sibling-Sibling or Spouse-Spouse, in the more gender-marked forms
different if the genders are different, and in the case of Siblings may be
different if relative age is different (and significant).
- Compound relationships.
- Compound relationships are
established between people by at least one common intermediate relationship,
which itself may be compound. Compound relationships are sequences of basic
relationships which relate one person to the other. There may be more than one
compound relationship between any two individuals.
- Reciprocals for compound
relationships are the sequence of reciprocal relationships between the same
Next section: Specification