Field name | Data Constraints |
Unique Id | Required in every case. Must be |
unique, number only |
Sex | Required - must be one of 3 values |
m,f,? |
Name (s) | Text |
Date of Birth | Date format only |
Place of Birth | Must be in list of place names |
Date of Death | Date format only |
Place of Death | Must be in list of place names |
Current Residence | Must be in list of place names |
Time at current residence | Date format only |
Education Religion Economic Data Genetic/Medical Data Gossip/Other Information Marriages |
Marriage Id | Required in every case. Must be |
unique, number only |
Husband | Number only but must already be |
in the individuals files |
Wife | Required in every case. Number |
only but must already be in the individuals files |
Date of start | Date format only |
Date of end | Date format only |
86 |
Offspring set of Individual | Numbers only but must already be |
ID nos... | in the individuals files |
Gossip/Other Information |
Note that only the wife but not the husband must have a required value. This
reflects the common occurrence where fatherhood may be unknown, uncertain or not admitted. There are social systems where even this requirement may have to be relaxed - in strongly patrilineal societies the identity of a wife may be implicit from the fatherhood of a child but the actual identity of the mother may be unknown. A final variation on the way in which the data is organised demonstrates how different files each containing a limited set of data can be linked, so the sum of the data in the files in conjunction with the web of their relationships provides an economical way of storing genealogical data. Consider three different linked files: The first records personal information about individuals, the second lists the marriages of an individuals parents, and any that they themselves participate in (it therefore links the generations). The third file contains information about marriages. These files may be summarised as follows: |
Individual Data |
Unique Individual Id Sex Name (s) Date of Birth Place of Birth Date of Death Place of Death Current Residence Time at current residence Education Religion Economic Data Genetic/Medical Data Gossip/Other Information |
Generation Links |
Individual Id Parent's Marriage Id Own Marriage Id (to be repeated as many times as marriages) |
Marriage Data |
Marriage Id Husband Id Wife Id Date of start Date of end |
87 |
Other information e.g. place of marriage, dates of prenuptial events, divorce etc. One of the advantages about using this sort of data structure is safety: it is relatively straightforward to take data stored in such a fashion and transform it into the format used by any particular database programme. This therefore ensures flexibility, and some basic insurance against future developments. The real work of collecting the data (leaving to one side the onerous chore of typing it all in to the machine) will not be wasted or become unreadable when machines change. |
Browsing |
When discussing data entry I said that typing long ID numbers could be avoided. The
machine could present a list of names, one of which being selected would automatically insert the ID number associated with that name into the relavant record. The technical term for the lists of names is 'browsable lists or indexes'. The user can browse through a list of individuals, or a list of places of birth, or whichever aspect concerns them. The database automatically generates an index from the data already entered, and this provides and instant and very powerful means of accessing the data. From an entry on a list one can go to the full set of data about an individual and possibly then choose to browse a different list starting from that individual. Or one may wish to move from an individual to their parents or siblings. The computer can automatically (invisibly) combine the data from the two types of data described above (individuals and marriages) so as to display what looks like a vary conventional sort of data card about an individual including names for parents, and cross references to siblings. But when entering the data the names of parents were not entered in the same way that they are displayed. That is a critical step that computers facilitate. We can design the process of entering data to reduce duplicate typing and errors. This is completely separate from the way we access that data once it is safely in the machine! Access and entry can be separated in computer in a way that is impossible on paper. I have already commented about how powerful a representation the genealogical tree is. Computer databases can be made to generate diagrams which act as indexes in the same way as browsable lists - clicking on a symbol for a person takes you to the data for that individual. The graphical representation can be altered at will which is not possible for a tree drawn on |
88 |
paper. One can instruct the computer to lay out the diagram to stress patrilineal,
matrilineal or cognatic descent. Individuals without descendants (who add density but do not contribute to the overall topology of the tree) can be omitted as can those in certain types of marriage. Finally, one can search all the data and select those that satisfy particular criteria. This could range from common searches, such as, for all individuals called 'David' to (in some systems) searches that allow one to extract particular types of marriage. So, for example, one could find (and then get the computer to draw) all those individuals in a population who had married a particular type of cousin. This is of interest both to alliance theorists and to geneticists! The ability to extract information in arcane ways gives the system the power to be flexible, to respond to the vagaries to be found in different individual circumstances. |
89 |
Chapter Seven: Models and simulations |
Most readers will be familiar with models in one form or other. Barbie dolls, HMS
Victory made of matchsticks, Douglas Bader's fighter plane from the second world war. All of these are models, but so too are the summary accounts that anthropologists give of different societies. A model is a smaller and necessarily incomplete representation of something else. The necessity of incompleteness is worth stressing since it is often misunderstood. A model has to be reduced in some form or other else it would be a replica. If a model is really complete then it simply reproduces its object without needing any analysis. A model is smaller than the thing it represents - either physically or conceptually. In order to reduce then some features must be omitted: Barbie dolls lack freckles, moles and reproductive organs. Anthropological models of society lack reference to individual psychological quirks. A particular model has been developed for a particular purpose. This may lead to some features being preserved at the expense of others. Toys for children often preserve surface appearances at the expense of inner structure. Anthropological models reflect the theoretical preoccupations of their creators; an economic model will seek to give an accurate albeit imperfect and reduced representation of an economic system - if this is an anthropological model then necessarily aspects of other social formations such as kinship and religion will occur but they may not be central. A model presents a perspective on its object, it is, if you like, a theoretical summary. The attraction is that by reducing the scale, by summarising and leaving things out we are helped to understand some of the basic mechanisms at work, unencumbered by the mass of detail. Of course there is a lot of theoretically controversial work to be done since the detail of everyday life is what we have to deal with. The models we create must be able to account for the ways in which the underlying armatures they identify both produce and are constrained by the flux of everyday life. The point of discussing the status of models is that they can be more or less formalised, and they have an interesting relationship to simulations. Most anthropological accounts of societies are models in the widest sense of the term. Some models may be called formal models. These have the terms of the account explicitly laid out; the it is made clear exactly what is being discussed;, they state which aspects they are trying to understand and which they are leaving out. Strictly a formal model must be more than that: it must be defined in terms similar to a formal logic. I note that nothing I have said so far requires quantification, although most formal models do use numbers. It is simply easier to construct a formal model using numerical data, but kinship algebras and the like demonstrate that one can formalise a completely non quantified domain. There are several advantages to formalising (whether or not one quantifies, whether or not one fully succeeds). The exercise of attempting to produce a formal system helps in making conscious the background assumptions that would otherwise not be clearly stated. It is far easier to see if a formal model succeeds in representing its object than an informal model where it is possible to produce an account that sounds impressive but ends up saying little of substance. In short, aspiring to formal systems -where they are possible - and taking some steps towards realising them where they are not can increase rigour. It makes it harder to hide behind a comfortable smokescreen of obscure jargon and fancy hegemonic concepts which cannot withstand scrutiny. If a model has been sufficiently formalised it can be tested by using simulations: a computer can be instructed to behave like the model and the effects of varying starting points can be investigated. One controversial subject in anthropology has been the relationship of demography - patterns of populations structure - and kinship, particularly when there are rules for selecting marriage partners from designated kin categories (such as second cousins). Simulations have been used to show how population size affects the number of possible |
90 |
91 |
pursue is that the words do not do justice to the experiences they are based on,
that there is Multimedia, classifying and juxtaposition
|
92 |
information - they have authors titles publishers dates. So a book - an object -
as being locatable by virtue of a set of attributes that it possesses which are indexed by card catalogue or by computer. |
BOOK | Indexed | BOOK |
Attributes |
From this point it is not such a big step to replace books by photos or film clips
and the result are multimedia catalogues. So my first, very modest, claim is that by cataloguing photos in ways hitherto impracticable, access is achieved and the use of photos and films as research material becomes serious in a way not previously possible. For example, a collection of photos can be searched by keyword to turn up relevant items. Having found a set of items that satisfy the criteria the next step is to juxtapose. The idea of putting similar things together to emphasise their similarities AND to reveal the ways in which they differ is basic to anthropology. The comparative method hinges around trying to hold some features constant and then see what else differs. For example, consider the way that the ethnography of gather-hunter societies constantly revolves round the three-way contrast of rain forest pygmies, those living in semi-desert areas in southern Africa and Australian aborigines. Some features vary others are held constant. Now if that idea holds true of ordinary anthropology how much more so does it apply to visual anthropology! We want to juxtapose photographs, film clips and sound recordings with texts and with other clips or prints. There is an analytic point to be gained - we can see patterns - we can begin to detect patterns of influence |
93 |
94 |
literature. By making a wider range of material available two things happen.
First, it changes the types of criticism that are possible: when I made a digital recording of a published transcript available in 19931 it became possible for linguists to claim that I had incorrectly transcribed the recording - previously they had to take the transcript on trust. Second, it can save a lot of effort at making verbal descriptions of non-verbal actions. Rather than have to try and describe every movement made by the actors in, for example, a complex ritual or a dance, the use of short video clips can accurately and vividly convey the aspect that is being described - and these clips can then be referred to as part of the analysis. Here the metaphor of pages becomes inadequate. The sorts of things that are possible (and now some cheap easy-to-use software makes this easy to perform) are to annotate photographs, sound recordings or video clips. Texts containing a commentary or an analysis can linked to particular sections of images, or to particular frames of video so the reader can read the commentary at the same time as they see the image or hear the sounds in question. Works such as the Yanomano Axe fight in which Asch took a few seconds of film, and in order to analyse it, repeated the film several times, become easy to replicate. Asch's film has been widely used in teaching. The example it provides has not been widely adopted mainly, I believe, because it was extremely hard to realise using film. The Axe fight has recently (1997) become available as a multimedia CD- ROM where it joins some other examples such as Wiyuta in which Assibone storytelling and sign language is presented and analysed. Here we have come along way from the word processor and indexes to fieldnotes, but let us not be misled. The types of multimedia document that I have been describing are boons to anthropology, they encourage the process of analysis and its communication, but in no wise do they replace or alter the essentially human relationship that is at the heart of the anthropological encounter. Oxford - Casole D'Elsa - Canterbury - Somie 1992-8 1The transcript was published in Zeitlyn 1993 and the digitised audio recording is available from the Virtual Institute of Mambila Studies via the following URL: 'http://lucy.ukc.ac.uk/dz/' |
95 |