Criteria of Symbolicity: Intrinsic and Extrinsic Formal Properties of Artifacts

Paul Bouissac (University of Toronto, Victoria College)

1. Two approaches to identifying symbols.

While anthropologists usually can assess the functions of most artifacts by correlating them with specific, observable behaviors, prehistorian archaeologists must construct hypothetical behaviors which can never be verified. Assigning functions to prehistoric artifacts therefore relies exclusively on inferential arguments. In practice, except in the case of replications, these heuristics are rarely made explicit and it seems that the most common interpretative strategy consists first of imagining a "culture" on the model of those produced by the ethnographic record and, then, looking at the material data from this holistic vantage point. This top-down approach has the advantage of providing at the onset a main frame of reference within which whole clusters of data can be integrated into meaningful virtual behaviors. Moreover, in so doing, blanks can be filled with both hypothetical behaviors and not less hypothetical perishable objects consistent with the general picture that has been set forth. The history of palaeontology and prehistory (Groenen 1994) offers many examples of such narratives through which remnants of the past are construed as illustrations of the stories told.

Two arguments can be put forward in support of such a top-down methodological approach: first, there is no alternative; second, a narrative, whatever its specifics may be, always forms a matrix from which propositions can be derived regarding human agencies, their relations and their actions. These tentative inferences can lead to discoveries of new data by streamlining the attention of the inquirers towards a specific range of artifacts or the byproducts of their making. Naturally, the absence of evidence, then, may count as much as relevant information as its presence would have and may contribute to revising the initial narrative. A case in point is the shamanism hypothesis developed by David Lewis-William (1995, 2002), whose epistemological strategy is made explicit in the form of a four-stage model of the production and consumption of San rock paintings based on ethnographic evidence. The model is applied to the Upper Palaeolithic parietal art of Franco-Cantabria under the double assumption that this cultural area was indeed shamanic and that the same four stages were implicated in the negotiation of social relations among the local prehistoric populations. A set of consequences are derived from this overall comparative conception and their degree of congruence with the archaeological record are assessed.

The bottom-up approach constitutes an alternative strategy. It consists of building up increasingly complex patterns from limited but precise information. It requires exhaustive observations and measurements of artifacts and their surroundings at various levels of analysis and tentative configurations and reconfigurations of the data within middle-range hypothetical interpretations such as the reconstruction of some technological spheres including the selection and gathering of raw material, and the steps leading to the completion of a stone tool or a parietal painting (chaîne opératoire) (e.g. Schlanger 1994). The proof that the method is valid rests on the success of the replicating process and the demonstration that the techniques used in the process are the only ones able to achieve this result. Both gestural and cognitive inferences can thus be confidently made. The next phase in the inquiry is to replicate also the range of behaviors made possible by the stone tools such as killing specific preys, cutting up carcasses, preparing skins, processing wooden implements or preparing pigments for the purpose of painting. It also involves reconstructing the technical gestures which must be assumed to account for the observed effects such as the shape of a tool, the making of a petroglyph or the negative representation of a hand (e.g. Bednarik 1998, Lorblanchet 1991). Partial cultural patterns progressively emerge from these processes like pieces of a jigsaw puzzle in which, however, many gaps remain to be filled regarding notably the social relations and mental representations which can be assumed to have characterized the various populations of tool makers over very long periods of time. But is it possible to apply the bottom-up method to the daunting task of reconstructing systems of beliefs in the absence of a known language that would provide fragments of the missing universes of meaning? Some do proclaim this enterprise impossible if not absurd. Others switch to the top-down method as the only way out of this apparent methodological dead end.

This paper contends that the bottom-up method can be pushed further and that attempts at reconstructing at least in part the mental universe of prehistoric populations should not be abandoned too quickly. A first step in this direction would be to determine the formal features which could qualify some physical remnants as pertaining to some symbolic rather than strictly practical behaviors. This would isolate subsets of artifacts upon which hypotheses could be tested regarding the formal consistency of their organization. Providing proof that some kinds of algorithms generated their structure should lead to a range of virtual cognitive mappings and to hypothetical representations more rigorously constrained than it is the case in the top-down method. This, of course, was the approach propounded by Leroi-Gourhan (e.g., 1992) in the wake of Max Raphael's earlier insights (Chesney 1994). Admittedly, Leroi-Gourhan's theory regarding the mapping of images of some animal species unto the topography of the caves is now mostly discredited, as are his sexual interpretations of most geometric signs. However these shortcomings do not come from the bottom up method itself but from its faulty application. Indeed , it is now generally recognized that Leroi-Gourhan's model too hastily determined the selective gathering of the data. A merely approximate fit between the theory and the data is not sufficient, mainly when further data are chosen for their congruence with the theory. In fact, the generalization of a tentative insight based on a preliminary set of data amounts to shifting prematurely from the bottom-up to the top-down method. The approach that is proposed here should be mindful of this danger.

Before addressing the issue of the criteria of symbolicity, a preliminary discussion of the notion of symbol is in order. While this term has a fairly precise meaning within the semiotic system of the American polymath C.S. Peirce (1839-1914), who is considered one of the founders of semiotics, it has been used with a wide range of semantic values by other scholars in a great variety of disciplines (e.g., Whitehead [1927]1955, White 1949, Douglas 1973). Peirce's "symbol" is understood as a sign based on an arbitrary or conventional link with its referent while "symbolism" is used in anthropology to designate all behaviors which are determined by ideologies, values, and beliefs rather than by purely functional considerations (Robb 1998). These two meanings do not exactly coincide but they overlap in as much that both imply a social ground as well as virtual (mental) operations which elaborate meanings and determine behaviors in the absence (in the eyes of the observers) of obvious, material forces or agencies. It applies also to all forms of computation, linguistic expression and reasoning through the manipulation of artifacts representing classes of objects or abstract entities. Naturally, the distinction between material culture and symbolic culture is an artifact of the research which presupposes that an objective ground exists to positively distinguish the two. In fact, from the point of view of a particular, culturally homogeneous group, the two are intimately blended in actual behaviors. However, because of the apparent lack of a better method, it seems justified to make this distinction at least as a tentative step toward a fully integrated representation of the daily life and history of human populations which are so distant in time that attempting to reconstruct their mental universe at first might seem futile as some have contended (e.g., Hawkes 1954) . Therefore, assuming that symbolic behaviors, based on the storage, coding and communication of information, generally leave some material traces in the form of artifacts and their collocations, it is reasonable to scrutinize objects which do not appear to have any obvious technological functions. But rather than relying on creative imagination in order to sort them out, it will be proposed here to use a formal method of discrimination, keeping in mind that a single criterion can not be sufficient to establish the plausibility that an artifact was endowed with a primarily symbolic value. It will remain to be determined whether some sets of criteria can be considered to be robust enough to lead to confidently assigning a plausible symbolic function to an artifact, whatever this function may be. It is indeed important to distinguish symbolic plausibility from interpretation, the former being a prerequisit for the latter. Naturally, it is assumed that the objects considered bear indeed the irrefutable marks that they have been formed or collocated by the activity of human agencies and that the approximate age of these transformations have been established according to scientific standard procedures.

2. Intrinsic properties.

Five intrinsic criteria will now be examined. These criteria are intrinsic in as much as they pertain to the features of the artifacts themselves irrespective of the various contexts in which they have been found.

The first criterion bears upon the relative dimensions of an artifact. Let us assume that a functional stone or bone artifact's dimensions are determined by its congruence with the dimensions of the span of the human hand, its skeletal and muscular structures, so that it can be efficiently manipulated. Ergonomic and ballistic characteristics, taking into consideration the natural variability of human physical dimensions, can at least suggest minimal and maximal thresholds beyond which an artifact is likely to lose its functionality. Very large objects that are difficult to be moved by a single person or very small ones that require particular attention to be distinguished from their surroundings are unlikely to have possessed practical functionality. However, such thresholds are difficult to establish in absolute terms since cooperative manipulations of large objects and insertion of microliths in wooden tools remain always a possibility. Moreover, on the one hand, symbolic artifacts may also require to be congruent with the dimensions of the human hand so that they can be handled, for instance in rituals, and, on the other hand, magic objects whose possession is considered crucial for the success of some activities can be neither too large so as to interfere with these activities nor so small that they can be easily lost. In spite of all these qualifications, it seems that dimension nevertheless constitutes a relevant criterion as long as it is complemented by other characteristics.

The second criterion concerns density. The relation of dimensions to weight is relevant to the notion of portability, but, more importantly the density characteristics of an artifact determines the functionality of its impact and its degree of degradation through use. Lower density material, while often easier to transform, are inappropriate for certain basic functional requirements. There may indeed be a scale of increased plausibility of symbolicity based on the lower density or malleability of the material used for an artifact. However, this also needs to be qualified because very high density objects with weights disproportionate with respect to their dimensions can conceivably be endowed with symbolic values because of their rarity or appearance. Nevertheless, clays and paints at the extreme end of the density scale are usually associated with symbolic expressions.

The third criterion concerns the degree of complexity of an artifact, that is, the relative quantity of information it offers in terms of shape, structure, and patterns. "Information" is not taken here in its commonsensical sense but with the meaning of its definition in information theory: a measure of uncertainty or unexpectedness. For prehistorians, the discovery of new artifacts triggers a classificatory behavior depending on the features which are usually recognized as belonging to a particular type of objects dating from a particular period. Thus, assigning a position in a matrix, at a place where it is expected according to the current standard knowledge, yields little information. It is simply "more of the same". But if there are obvious discrepancies between some features of the artifact and what would be expected, then the information value of this artifact may increase to the point that it becomes unclassifiable because it does not fit anywhere. Its characteristics with respect to its presence in the layer where it has been found or its sheer appearance is "mind boggling", "incomprehensible". Of course, this complexity is relative to a particular state of knowledge, and the first reaction of the archaeologists will generally be to try to reduce its information by testing whether it is the result of random or taphonomic factors. If none of these information-reducing strategies hold, then this morphological complexity will be attributed to the cognitive complexity which must be assumed to have generated it. A recent example is the Blombos piece of ochre with its geometrical markings. Let us note, in passing, that this particular object also satisfies criteria #1 and # 2.

The fourth criterion that is proposed can be called complementarity. Two or more artifacts complement one another if they can be shown to belong to a whole of which each one is a part. In cultures where there is no evidence of mechanical devices, parts which form coherent sets such as strings of beads and painted or engraved representations are likely to have symbolic rather than practical value. This criterion is related to the previous one in as much as it can be a form of complexity, mainly if some unexpected logic rather than pure randomness is discovered in the composition. Moreover, complementarity can be shown to go beyond a few items and to encompass large sets spread on large cultural areas and showing temporal depth. This characteristic leads to the notion of type / token relationship which introduces the next criterion. (See appendix on the type / token distinction)

The fifth criterion consists of evidence of replication applied to artifacts which meet the above four criteria. The replication of complex patterns with or without variations on diverse scales and supports is probably the surest indicator that these artifacts were endowed with symbolic value. First it provides absolute proof that the patterns are not due to random or taphonomic causes. Secondly, it allows the inference that they were pragmatically important even if their other morphological and material characteristics disqualify these artifacts for being of any practical use. Further, it may suggest two conclusions: either the variations are the results of copying errors, or they are the results of systemic manipulations of the patterns.

In the former case (that is, variations in the copying of identifiable motifs), it is possible to infer the existence of types with respect to which the various realizations which are observed are tokens . As for isochrestic artifacts which are related to a single practical function, tokens can be classified according to their congruence with the distinctive features of their respective types. For instance, Leroi-Gourhan interpretation regarding the geometric signs as either male or female symbols presupposes the existence of a relatively abstract mental representation of two types which can be implemented in a variety of styles, scales and supports. The same applies to his theory of the mapping of animals unto the topography of the caves in as much that species and topological categories provide the types which are the sources of the tokens (each painted or engraved bison shows variations and each cave has its own structure combining narrow passages and wider spaces). As it was pointed out earlier in this paper, it is not the method itself but its loose application which makes Leroi-Gourhan conclusions questionable.

In the latter case (that is, the variations are systemic), it is theoretically feasible to reconstruct the algorithms which generated the systemic variations. On purely morphological grounds, it is possible to infer a calculus from the variations of a close set of elements. Of course, demonstrating that a system of signs governed by rules can be inferred from the data does not mean that the code can be deciphered, although it is a first step in this direction following the bottom up approach.

3. Extrinsic properties.
While the previous discussions of some formal features concerned bounded items or sets of items taken in isolation, other aspects of an artifact such as, for instance, the place where it was found may be relevant to the determination of its functioning as a symbol at the time of its making and use. Three such extrinsic properties will be now examined: location, distribution and context.

Location refers to at least three possible kinds of information. First, naturally, the geological level which provides information regarding an artifact's temporal depth. Indeed, once taphonomy effects have been considered and the possibility of fraud eliminated, this is the main indicator of its contemporaneous material culture, hence the possibility of relating and comparing this artifact to others and to establish stylistic constants. Secondly, the geographical position yields information about the boundaries of cultural (or tribal) areas and the possible spreading over time of specific behaviors. Thirdly, when the baseline level of action has been reliably preserved, like in some caves or shelters, or can be legitimately assumed like in burial sites, the position of an artifact with respect to human anatomy, either vertically or horizontally, offers crucial information by relating the artifacts to the deliberate gestures which can be presupposed by its position. Gestures and postures can indeed be inferred from the location of a natural object or an artifact as well as from the particular position of the skeleton and the associated artifacts in a burial site. The focus on the original location of an object of interest in absolute and relative terms is a rather recent concern in archaeological research. Naturally, inferences from a location must be made with caution since many causes besides deliberate movements may be responsible for a particular position, such as animal agencies, floods and earthquakes (or archaeologists' mindlessness). However, comparisons can yield information by suggesting consistent cultural constraints regarding the gestures which can be inferred from the position of artifacts.

Distribution applies both to the bounded context and to larger cultural areas. In the former case, how individual artifacts such as geometrical signs and representations of animals or engraved and otherwise marked items are distributed over a surface or a site can indicate a range of constraints which are neither purely practical nor random. If it can be proven that a close set of items was produced or maintained within a bounded span of time by a single population, it is then possible to test their degree of systematicity or, alternatively, to demonstrate their randomness. In the latter case, that is, on the level of extended regional space, mapping the distribution of well defined items or sets of items over large areas may reveal not only significant concentrations but also help locate new sites by projecting structures inferred from fragmentary patterns such as the choice of particular geomorphs (typical landscapes, natural beacons or vantage points) whose practical affordances are not obvious.

Context can be usefully distinguished from co-text. The former refers to the immediate and distal surroundings of artifacts. It is a very fluid concept whose delimitation often depends on the a priori interpretation of these artifacts. Is the context the total culture or "stylistic" era, or the climate and related plausible modes of survival among the corresponding fauna and flora ? Or is it the cave, the shelter, the camp site, the burial site? It is intuitively considered that information provided by the context can orient the interpretation. The most common ascription of symbolicity to artifacts comes from their presence in the context of a burial. However, in accordance with this latter example, it might be more productive to restrict the conceptual vagueness of context to the more precise notion of co-text, that is, the consistent collocation (or spatial association) of two or more artefacts within an objectively bounded space. To be heuristically useful this property requires of course that the exact spatial disposition of the artifacts with respect to one another be a part of the archaeological record as precisely as all the intrinsic properties which were listed above. A corollary of this requirement is that the same principle should apply to items such as paintings and engravings of identifiable objects (let them be animals or geometric signs) which have been collocated in a permanent manner upon the bounded surfaces of a cave or another artifact by those who created them.

4. Concluding remarks.
Taken individually, all the criteria which have been proposed above in relation to both intrinsic and extrinsic properties of artifacts may appear trivial or questionable. However, the position of this paper is that not a single one is sufficient to indicate the plausibility that an artefact was symbolic rather than practical at the time of its creation and use. It remains to be determined whether all of these criteria are necessary or whether a subset of them would provide a satisfactory ground for establishing the likely symbolicity of specific items. The possibility of devising additional criteria is also left open. Moreover, it is important to underline that most of these criteria are currently used by archaeologists, usually in a informal manner, in their interpretive reasonings. But it is equally important to emphasize that they do so most often without specifying with enough clarity whether they follow a top-down or a bottom-up methodology. This leads to circular arguments notably when their primary interest is to attempt to correlate a hypothetical stage in cognitive evolution, sometimes confused with cognitive development, with some characteristics of the archaeological record. The weight of presumptions can be very heavy indeed in the intuitive assessments of the plausible functions of artifacts. Powerful images and metaphors have taken root in the various disciplinary cultures of prehistoric archaeology since their inception in the nineteenth century and still guide to a large extent the specialists' "educated guesses" which are now more than ever the source of fierce debates of the sort usually triggered by uncertain knowledge. One way to emancipate oneself from these biases is to adopt the counter-biases of the scientific method which has empowered human cognition in many domains. The method is not without flaws. In particular, it is long and arduous, and it requires both some intellectual audacity in the formulation of its hypotheses and some readiness to acknowledge that an attractive hypothesis can be irrevocably falsified. It is, at first, much less rewarding than the epic imagination which fills the gaps in our knowledge with vivid images that meet the expectations of the moment. The proposed approach is based on the certainty that the archaeological record, even in its current fragmentary state, can yield much more knowledge than has been extracted from it so far.

References:

Bednarik, Robert (1998) The technology of petroglyphs. Rock Art Research 15 (1) : 23-35

Chesney, Shirley (1994) Max Raphael (1889-1952) : a pioneer of the semiotic approach to palaeolithic art. Semiotica 100 (2/4): 119-124

Douglas, Mary (1973) Natural Symbols: Explorations in Cosmology. Harmondsworth: Penguin

Groenen, Marc (1994) Pour une histoire de la préhistoire. Grenoble: Jerome Millon

Hawkes, C. (1954) Archaeological method and theory: some suggestions from the Old World. American Anthropology 56; 155-168

Leroi-Gourhan, André (1992) L'Art pariétal: Langage de la préhistoire. Grenoble: Jerome Millon

Lewis-Williams, David (1995) Modelling the production and consumption of rock art. South African Archaeological Bulletin 50 : 143-154

Lewis-Williams, David (2002) A Cosmos in Stone: Interpreting Religion and Society through Rock Art. New York: Altamira Press

Lewis-Williams, David (2002) The Mind in the Cave. London: Thames and Hudson

Lorblanchet, Michel (1991) Spitting images: replicating the spotted horses of Pech Merle. Archaeology 44 (6) : 25-31

Robb, John (1998) The archaeology of symbols. Annual Review of Anthropology 27 : 329-346

Schlanger, Nathan (1994) Mindful technology: unleashing the chaîne opératoire for an archaeology of mind. In The Ancient Mind: Elements of Cognitive Archaeology. C. Renfrew and E. Zubrow, (eds.). Cambridge: Cambridge University Press.

White, Leslie (1949) The Science of Culture. New York: Farrar, Strauss and Cudahy

Whitehead, Alfred North (1955) [1927] Symbolism. Its Meaning and Effect. New York: Putnam

Appendix A

When is an artifact a symbol? The type / token distinction. (Paul Bouissac)

Archaeologists are interested as much in artifacts which look alike as in those which look different. A great deal of information can indeed be inferred from observable similarities across time and space. But the search for resemblance is rife with problems. On the one hand, there is a range of degrees of similarity between exact morphological identity and approximate resemblance of appearance or function (e.g., isochrestic tools). On the other hand, two or more artifacts may look alike for a variety of reasons including chance, copying, imitation or multiple implementations of a single algorithm. These problems are compounded by issues such as whether there exist an original model or whether similar ecological constraints can give rise to convergent artifactual morphologies. Natural, functional and cultural forces can equally play a determining role in the emergence of artifacts that look alike, at least in some respects.

The main purpose of this note is to examine a particular case of the generation of artifactual similarities: the type / token relation. This conceptual distinction was first proposed by Charles Sanders Peirce as a part of a triadic system. It gave a modern, semiotic form to the old debate in medieval philosophy between those who believed that individual objects or words had conceptual meanings only in as much as they were the reflects of real but immaterial Ideas (in Plato's sense) and those who looked into the production and use of each instances as their only true possible source of meanings. The former were called the Realists, the latter were called the Nominalists. By proposing the type / token / tone distinction, Peirce contrasted the pure quality of an experience (tone) with its interpretation (token) as the result of a rule (type). In archaeology, these notions would correspond to the purely physical description of an artifact (tone) followed by an interpretation of its particular function such as a "bead" (token) through assigning it to a general class of artifacts such as "ornaments" (type). All occurrences of beads are beads because they relate to a general function expressed as a rule, an algorithm which specifies a behavior.

In further philosophical and logical developments Peirce's triadic distinction was reduced to a dichotomy: type vs. token, to which Peirce sometimes referred as sign vs. replica or legisign (from the Latin lex [law]) vs. Sinsign (sin stands for singular or single). Later, other philosophers reformulated the relation as "sign-family" vs. "sign-vehicle" (Morris 1971), or "sign-design"vs "sign-event" (Carnap 1961), thus denoting single objects or events with respect to their class. The logical implications of these distinctions and reformulations are complex (Russell 1940) and will not be addressed here.

Focusing on the type / token relation, as it is exemplified, for instance, to explain how the letters of the alphabet (or any other system of writing) can take many graphic forms without losing their functional value as elements of distinctive words, it is possible to explore the applicability of this relation to artifacts. A theoretical difference can be established between mere copying and implementing a rule of construction with respect to a system of signs. First, copying implies the presence of an artifact (or a natural object) which serves as model for the replicating behavior while producing a conventional sign does not require the presence of a model but simply a mental algorithm that specifies the necessary actions. Secondly, the criteria which determine whether a copy is good or poor are not the same as the criteria which qualify the production of a conventional sign as functional. When children learn how to write or draw they painstakingly produce copies. But once they have mastered the code they have assimilated some basic rules of construction. Their writing skill is made of a set of algorithms such as "draw three intersecting lines to produce the capital letter A". The variety of fonts and personal handwritings shows that the type (i.e., the algorithm) can produce an open-ended list of tokens (i.e., the many graphs that counts as letter A in a particular population whose members share the knowledge of the type and its value with respect to the other types belonging to the same system). Depending on the occasions, this letter can be implemented minimally or rendered with various emphases and ornaments including colors and decorative elements.

Can archaeologists distinguish whether a single artifact is a copy or a token from the mere examination of this artifact? This is doubtful because the distinction becomes an issue only when there are more than one artifact. In this case, it can be hypothesized that there is a scale of similarities with a threshold in the rate of variations indicating that it is likely that the craftmen were concerned with implementing a type rather than producing a copy.

This distinction suggests two possible kinds of symbolism in relation with two hypothetical kinds of motivation for the replication of artifacts: First, the assumed magical power of some natural or artificial objects such as a charms or idols can be believed to be harnessed through making rigorously exact copies of the archetypes. Secondly, some basic distinctive features will suffice to implement the symbolic values of a set of conceptual types through their approximate realizations in replicas which preserves their distinctive structural identities. In this latter case, it can be expected that we are dealing with a set of types which form a system of contrasts such as would be the case for the symbols used in reckoning, possibly based upon the various configurations of the human hands. For this sort of artifacts, the strongest evidence that they are objects to be interpreted as tokens of types ultimately rests upon the demonstration that their diverse positions in clusters are not random but follow some compositional rules. Naturally, the second kind of symbols can be copied by craftmen who do not understand the relation of these tokens to the system of their types, following the process known as "cargo cult".

These considerations on the type / token relationship, as opposed to mere copying, are particularly relevant to the evaluation of rock art. Most interpretations, in main stream archaeology, remain focused on the identification of natural objects, usually animal species construed as preys (Mithen), mythical figures (Leroi-Gourhan) or shamanistic personae (Lewis-Williams ). On the primary level, parietal paintings are most often described piece meal in the literature, with an emphasis on the most striking pictural effects in term of resemblance with assumed actual models. Whenever composition is taken into consideration, this is done according to artistic principles, underlying for instance the realism or the stylization of the figures, their apparent dynamic, their blending with the morphology of their natural support, some perspective effects and the like. On the secondary level of analysis, they are taken as documenting the environment, and the individual subjects which have been identified are reorganized in the form of lists with indications of the number of items for each category. Conclusions may then be drawn regarding the composition of the contemporary fauna or the relative importance of some species in hunting or for symbolic thinking.

All these interpretations are equally plausible and equally unverifiable. The type / token distinction can, however, suggest another, less intuitive hypothesis. Following the tentative suggestion that some rock art could be hieroglyphic, i.e., made of representations of sounds rather than animals as individuals or as species (Bouissac 1994), Hans Bornefeld (personal communication) notes that in Lascaux, for instance, two complex sequences appear to reproduce each other in reverse order and heuristically proposes a phonologic rendering based on tentative reconstruction of archaic languages. From the point of view developed in this note, the issue is not whether the claim that these "inscriptions" have been deciphered can be substantiated, but whether the sequences indeed show compelling evidence of iterativity and systematicity. The latter should incite archaeologists to further investigate the syntactic forms of comparable clusters and could lead to the conclusion that the representations which "decorate" the cave walls and other surfaces should be treated as tokens of a limited number of types whose symbolic value could then be established at least as virtual systems.

References:

Bouissac, Paul (1994) Art or script? A falsifiable semiotic hypothesis. Semiotica 100 (2-4): 349-367

Carnap, Rudolph (1961) Introduction to Semantics. Cambridge: Harvard University Press

Esposito, Joseph (1998) Type and token. Encyclopedia of Semiotics. P. Bouissac (ed.). New York: Oxford University Press (622)

Morris, Charles (1971) Writings on the General Theory of Signs. The Hague: Mouton

Niklas, Urszula (1979) On the type-token distinction.A case against Nominalism. Jeltudományi Dokumentumok. Abstracts and Papers. . Semiotic Terminology, 28 June - 1 July 1979, Budapest, V. Voigt, ed. (71-79)

Peirce, Charles S. (1958) The Collected Papers of Charles Sanders Peirce Vol. VIII, E. Burks (ed.). Cambridge: Harvard University Press.

Russell, Bertrand (1940) An Inquiry into Meaning and Truth. New York: Norton