Go to Semiotic Review of Books Home Page
Go to SRB Highlights
Go to SRB Archives
This editorial appeared in Volume 10 (1) of The Semiotic Review of Books.
by Robert G. Bednarik
The function of boundaries is to separate and thus define, if not create, entities. In archaeology, these would be the classes of the evidence the archaeologist defines as archaeological finds. For instance, stone tools are divided subjectively, in accordance with Western concepts of style, function, form and material classes. These taxonomies of stone tools are usually not falsifiable, hence not scientific, and from the little we know about the perceptions of hunter-forager-fisher societies, they do not resemble their perception of stone tools. Traditional societies do have taxonomies of stone tools, but like those of other object classes they differ significantly from those of Westerners. In the view of traditional societies, the properties of stone tools may be related, for instance, to colour, or to some to us obscure variables of the stone, instead of the archaeologically all-important typology. Western 'experts' may perceive such traditional views as naive or ethnocentric, which illustrates the inability of Westerners to perceive their own epistemological naivety.
Taxonomies, archaeological, traditional or scientific, are all based on what are perceived to be the common denominators of phenomenon categories, and on which ones of them are selected as the crucial ones (Crucial Common Denominator of Phenomenon Category, or CCD; see Bednarik 1994: 149). That decision is subjective, which means that different metaphysical systems are likely to prompt the selection of different CCDs. In other words, they create different realities. The human species has created a number of such realities. Our tendency of perceiving the one reality subscribed to by the community that has conditioned us as individuals as real and valid is called ethnocentrism.
The same is apparent even in the way science is conducted. Disciplines could be demarcated in many ways, but Western ideology, has developed a particular taxonomy of disciplines. Archaeology is very well suited to show us how haphazard this process of developing and shaping a discipline can be. Its models are all historically contingent, in that they have depended on a more-or-less haphazard sequence of discovery. If important finds in archaeology had been made in a totally different historical order, we would have a very different archaeology today. The structure of archaeological knowledge is historically contingent at every point of time, which means that the dynamics of knowledge acquisition must be historically contingent at any time. To illustrate this point with an example, let us assume that the Altamira rock art had been discovered before the identification of Upper Palaeolithic tool industries, and that it was soon dated and recognised for what it is. No doubt science would have formed a concept of the people who had produced the art, based on its perceived aesthetic merits. Next, let us assume that, several decades later, stone tools were discovered and attributed to the same period. This notion would have been squarely rejected as preposterous by archaeology, and as an attempt to discredit the 'established fact' that these were people of a very advanced civilisation, not lowly Stone Age savages. In other words, heuristic history not only determines acceptability of models, it also stipulates the very cognitive framework within which they maybe developed. Archaeology is entirely contingent on the historical sequence of discoveries, and on numerous other factors that have nothing to do with objectivity, but are random forces producing a mythology of the human past. The divisions of the Upper Paleolithic are entirely based on perceived tool styles, they have no historical validity whatsoever, they are simply a fantasy of archaeologists. If discoveries had occurred in a different sequence, these divisions would be totally different today. One should not expect these entities to be scientific or objective or valid.
Any analytical process is based on assessing crucial denominators, and on seeking to find among them the one that is crucial in defining the category in question. Because in mathematics, geometry and physics we seem to be able to perceive valid taxonomies, the most notable being the periodic chart of elements, we tend to assume that such taxonomies should also exist elsewhere. However, they are not as clear-cut in the biological sciences, and in the social sciences they are practically arbitrary. In archaeology, specifically, most CCDs used in interpretation appear to be false, mostly because the massive importance of taphonomic logic is often not taken into account. Most models in archaeology seem to be based on misunderstanding the role of the evidence. To illustrate again with an example, consider the global distribution of hominid remains. Many archaeologists seem to think it reflects in some way regions of hominid occupation, even of population densities. Nothing could be further from the truth, which is that these are almost random localities that reflect primarily the suitable sedimentary preservation conditions and the magnitude of relevant research efforts. The distribution of hominid find sites tells us very little about the geographical extent of the species, it indicates the extent of the areas where the fossils could manage to survive and where they were found and correctly identified. We can generalise that the distributional and statistical characteristics of most types of archaeological evidence are practically irrelevant to interpretation. Essentially this is because archaeology cannot produce valid random samples of any kind of evidence, and no such evidence can be expected to correctly reflect any historical variable. Therefore we have no correctly measured variable, or common denominator, and on top of that we probably are not capable of selecting among the available common denominators the one that is crucial in defining the category of evidence.
When these significant epistemological problems are introduced into rock art research, the usual procedure is to begin by describing an assemblage in preparation for some form of statistical analysis. To facilitate this it is clearly essential to create a taxonomy, to create boundaries between perceived motif types. The number of variables available for this is practically unlimited, and could include variables of morphology, technology, style, form, size, geology, geomorphology, petrology, hydrology, acoustics, relative orientation, orientation to other motifs at the site, to other features at the site, or to other sites, and so forth. Most archaeologists arbitrarily select some subjective characteristics, notably 'style' and motif morphology as perceived by the subjective alien observer and proceed to create taxa.
This can only be done arbitrarily. Let us consider this hypothetical situation: we have a site with hundreds of motifs, all of circle and radial combinations, none of which are quite identical to any other. They fall within the ranges depicted here (Figure 1). It will be obvious that any two researcher'attempting to create a taxonomy will design different taxa, unless they have been conditioned in precisely the same way, and that these taxa will not correspond to the taxonomy of the artists themselves. As in the case of the stone tools, the boundaries will be irrelevant for any scientific purpose other than to study the perception of the observer. Frankel's (1993) simile of the sculptor and the block of stone is most relevant: the individual archaeologist 'finds' interpretation just like the sculptor discovers a statue in a block of stone. In both cases, the interpretations are products of creativity, and naturally they will differ between any two practitioners.
In the case of rock art description this means in practical terms that the description of thecorpus will be valid as a subjective description, but that it would be quite wrong to draw any further conclusions from motif counts or similar information. Not only must the taxonomy be considered arbitrary, we have no way of knowing how the surviving evidence relates to the originally created rock art. How much rock art has been lost through repatination, exfoliation, rain water, frost shattering, insolation, sediment cover, erosion, dam builders or whatever other factors? More importantly, what characteristics of the art itself may have influenced its demise or longevity? Most of the taphonomic processes are obviously systematic rather than random, they can select in favour of certain characteristics, and the greatest mistake archaeology can then make is to regard such characteristics as being crucial in defining perceived entities, such as styles, traditions or cultures. These kinds of mistakes lead to fundamental misinterpretation of rock art. Boundaries are thus designed without understanding the nature and history of the evidence, and they are consequently of very limited use. They can tell us a great deal about the perception of the rock art observer, they can tell us very little about the rock art itself, or the way it was perceived by the artist.
In summary, we can say that boundaries in the way we perceive rock art are necessary to describe what we perceive in terms we are capable of understanding. In this sense they are necessary. But they and the taxonomies they can create are of no reliable value in defining an objective essence of the rock art, and they are of no consequence to the scientific study of rock art. They are unfalsifiable propositions as soon as they are taken beyond the status of simple empirical observations. In other words, we can test the proposition that there are twelve circle petroglyphs at a particular site, but we must not read any significance into this statement, or derive any other qualitative or quantitative deduction from it. Therefore boundaries and the statistics they create may be useful in describing what is on the rock today and how we perceive it, but archaeologically they are irrelevant unless they are used in very sophisticated ways, for instance in semiotic models. Let me again illustrate by means of an example. If in our hypothetical rock art site we were to consistently find a particular variable systematically associated with some of the motifs, we could draw relevant conclusions. In my theoretical example (Figure 2) we note that the little star motif always occurs together with motifs of six or eight radial lines with a common focus but never with any other combination. This is a form of associative syntax, and we can propose from it that it defines a valid taxon. Such a semiotically created boundary is testable and thus scientific.
Are boundaries appropriate, or merely convenient? In an epistemological sense they are not appropriate, they are convenient for certain limited purposes, but even then they are at best a necessary evil. At worst they are misleading, often ethnocentric and always anthropocentric. However, boundaries whose parameters comply with the demands of taphonomic logic or some other scientific framework are considerably more robust, and may well provide a sound basis for interpretation.
Bednarik, R. G. 1994. "On the scientific study of palaeoart." Semiotica 100(2/4): 141-168.
Frankel, D. 1993. "The excavator: creator or destroyer?" Antiquity 67: 875-877.
Robert G. Bednarik is editor of Rock Art Research, the journal of the Australian Rock Art Research Association (AURA) and of the International Federation of Rock Art Organizations (IFRAO). He is the author of numerous articles on epistemological and methodological issues in prehistoric rock art research.