Structuralism and Semiotics

Robert Layton. Handbook of Material Culture. Editor: Christopher Tilley, Webb Keane, Susanne Küchler, Michael Rowlands, Patricia Spyer. Sage Publications, 2006.

Structuralism and semiotics provide ways of studying human cognition and communication. They examine the way meaning is constructed and used in cultural traditions. Applications of structuralist and semiotic method have raised questions concerning the extent to which cultural understandings are stable and shared, or changeable and ambiguous. Structuralism contends that the cultural significance of objects and actions derives from their place in a cognitive system. Structuralism was first advocated by the French sociologists Durkheim and Mauss (1903/1963). It became the dominant theory in the social sciences during the 1960s and 1970s (e.g. Douglas 1973; Glucksmann 1974; Hawkes 1977; Lane 1970; Robey 1973; Sturrock 1979; Wittig 1975). In 1968/1971 the psychologist Piaget argued for the relevance of structuralism to an even wider range of disciplines, from mathematics to philosophy. Piaget anticipated some of the criticisms of more formalized versions of structuralism. The failure of early structuralists to consider temporal process was addressed by Bourdieu (1977) and Giddens (1979, 1984), giving rise to what has been termed ‘post-structuralism’. Foucault (e.g. 1972, 1977) and Derrida (1976) drew attention to the relationship between meaning and power. More recently, the anthropologist Gell has challenged the claim that that anything except language ‘has “meaning” in the intended sense’ (Gell 1998: 6). He is particularly critical of the structuralist semiotic anthropology in the 1970s (p. 163). At that period, Gell writes, ‘it was customary to discuss systems of all kinds as “languages”. … Art was the (cultural) “language of visual forms”’ (Gell 1998: 164, A.G.’s parenthesis). Gell singles out Faris (1971) and Korn (1978), who argued for rather complex vocabularies of visual elements and grammatical rules for combining them into well formed motifs or compositions in the art of East Africa and New Guinea.

The Theoretical Framework

The Foundations of Structuralism and Semiotics

Early structuralists reacted against previous writers who had interpreted non-Western customs as survivals from supposed archaic stages in human social evolution. They argued that the significance of a custom depends on its place in the structure of contemporary culture. In The Rites of Passage (1905), for example, van Gennep criticized the popular procedure of taking rituals out of their ceremonial context and considering them in isolation, as historical survivals, ‘thus removing them from a context which gives them meaning and reveals their position in a dynamic whole’ (Gennep [1905] 1960: 89). Van Gennep contended that there is a general tendency among human societies to conceive of a change in status on the model of a journey from one town or country to another or, as he expressed it, a ‘territorial passage’ (Gennep 1960: 18). Van Gennep argued that territorial passage had three aspects, separation from the place of origin, transition (la marge), and incorporation into the destination. Territorial passage could therefore stand for any change of status in society, but each phase in the ceremony made sense only in terms of its place along the journey from old to new status. ‘Marriage by capture’, where the groom and his brothers ride to the bride’s house, snatch her and carry her back to the wedding, is not a survival from some original condition of Hobbsian anarchy. It dramatizes the separation of the bride from her position as an unmarried girl in her parents’ house, and her incorporation into the groom’s household as a married woman. Through a series of case studies, Van Gennep demonstrated that rituals of birth, entry into adulthood and death can all have the same structure.

In 1912 Durkheim provided a more formalized and theoretically productive structural analysis in his examination of Aboriginal totemism. Since some totemic species seemed insignificant in themselves, Durkheim deduced they gained significance from their place in the totemic system. The classification of animal species parallels the classification of social groups. The marsupial rat and the witchetty grub were valued not because they had some intrinsic significance, but because each stood for the concept of a particular clan as one segment in the social system. The species itself was, moreover, less sacred than its representation in totemic art. This, Durkheim argued, was because the artistic motif represented the clan’s identity in concrete form. He argued that the existence of society depended on such symbols. ‘Individual consciences … can communicate only by means of signs which express their internal states’ (Durkheim 1915: 230). Art and writing therefore originated to fulfil the same purpose: ‘man commenced designing … [so] as to translate his thought into matter’ (Durkheim 1915: 127. n). Painted depictions of totemic animals were so simple (consisting predominantly of circles, arcs and dotted lines) that Durkheim took them to be arbitrary and non-representational.

The Swiss linguist Saussure (1915/1959) applied the reasoning used in Durkheim’s analysis of totemism to the explanation of communication through language. The sounds of speech, like the animal species depicted in totemic art, have no intrinsic meaning. The meaning of words is established by their place in the vocabulary. The linguistic sign has two components, the signified or idea, and the signifier, the spoken sound(s) that conventionally express that idea, equivalent to the idea of the clan and its expression in the totemic emblem. Saussure realised that there were many other sign systems in human culture that could be studied using the same methods: ‘A science that studies the life of signs within society is conceivable. … I shall call it semiology’ (Saussure 1959: 16). Saussure acknowledged that some semiological systems might include signs that seem to have intrinsic significance, such as bowing submissively to a superior, but emphasized that even these gestures are based on social convention.

A semiotic system is thus made up of two parallel structures: a set of ideas that divide up experience into discrete units, and a set of material signifiers that stand for those ideas. The signifiers may be sounds, pictures or gestures. Signifiers may be completely conventional, or there may be an iconic or associative connection between the signifier and the object, the idea of which it signifies.

Saussure introduced an important refinement to Durkheim’s totemic model for structuralism, the distinction between language and speech. Speech draws upon the vocabulary and grammar of the language to construct a limitless series of statements. Speech consists of what Saussure called a syntagmatic chain comprising, for example, a subject, verb and object. Each element also belongs to a paradigmatic series of alternative subjects, verbs and objects that could have been selected instead:

The child chases the ball.
The dog catches the stick. …

Sense, Reference and Abduction

Saussure’s primary concern was with sense, or signification: how sounds are conventionally related to ideas thanks to the structure of the language. The US philosopher Peirce classified signs according to the way they refer to objects in the environment. (See also Barthes 1967: 38, and Piaget 1968/1971: 115 n. 8, who identifies a similar set of terms in Saussure’s work.) Peirce identified three types of reference. An indexical sign has something in common with what it refers to: smoke is naturally associated with, and therefore points to, fire; a weather vane is an index of wind direction (Peirce 1955: 102-3). Icons (such as the motifs in representational art) look like what they refer to. As long as we can ‘read’ the style, we can recognize what is depicted. Peirce used the term ‘symbol’, for signs such as the words of language that are associated with the objects they refer to purely according to cultural convention.

Gell (1998) argues that, because art makes use of representation (iconic) images, and may be constructed from materials intrinsically associated with their subject matter (particular woods or pigments), art does not depend on convention and is therefore not a semiological system. Gell argued that art objects take effect not by communicating ideas, but by extending their maker’s or user’s agency through references to the world of objects (see Hoskins). Agency is the ability to choose between different courses of action. I exert my agency by restricting or enabling yours (see Giddens 1984: 9, 15). According to Gell, an art object can extend its maker’s or user’s agency in several ways. Sometimes the agency is psychological, as when an enemy is frightened by an awesome shield. The kula valuables exchanged between leading men on the Trobriand Islands of the Pacific circulate well beyond the personal reach of previous custodians, but their association with a powerful man makes them indexes of his bodily presence even after others have received and passed on those valuables.

Gell argues that art objects are not construed as meaningful; their references are construed through the process of ‘abduction’. Abduction is a form of inference practised in ‘the grey area where semiotic inference (of meaning from signs) merges with hypothetical inferences of a non-semiotic (or not conventionally semiotic) kind’ (Gell 1998: 14, parenthesis and italics in original). It is true that icons and indexes do not rely as completely on cultural convention as arbitrary verbal sounds (signs). Nonetheless, what icons and indexes signify, and what they refer to, will almost certainly be established by convention. Campbell spells this out very clearly in her analysis of the art on the canoes used to travel between islands in the kula network. Animals whose culturally defined qualities correspond to the values of the kula are represented as icons, but according to the conventions of local style. The appropriateness of the chosen animals is determined through local cultural symbolism (Campbell 2001). In contrast to Gell, the archaeologist Dobres (2000: 142) accepts that agency can take effect semiotically She defines agency as ‘an inter-subjective quality and unfolding process of knowing, acting and being-in-the-world’, a world that ‘is mediated by cultural reason, symbolic sensibilities, and personal and collective history’ (Dobres 2000: 151).

While Gell’s attempt to develop a non-semiological theory of art is questionable, his use of the term ‘abduction’ is fruitful. Meaning in some art systems is highly codified, while in others it is scarce and questionable. This creates a continuum between semiology and abduction in Gell’s sense, that is, a continuum between highly codified meanings and situations where it is unclear whether something is intended to be meaningful or not and, if meaningful, how it should be construed.

Durkheim understood the importance of differentiating the totemic designs of different clans. Totemic art is indeed a good example of a clearly structured system. As one Aboriginal man explained to me while discussing totemic body paintings, to dance at a ceremony wearing another clan’s design ‘would be to steal their land and their life’. It was an offence punishable by death. Panofsky showed how European Renaissance viewers would recognize a painting of a male figure with a knife in his hand as St Batholomew, a female figure holding a peach as the personification of veracity (Panofsky 1955: 54). The lily is a symbol of purity or virginity, and therefore often carried by St Mary. Traditional Chinese art also relies heavily on a codified series of images. Here, the peach is an image of long life, the peony an image of riches and honour. A Qing dynasty New Year’s painting from Suzhou Province shows a boy pulling a cap and belt on a toy cart. Cap and belt are essential parts of the official uniform of ancient China, and different ranks of officials wore distinctive styles of cap and belt. The painting signified the parents’ wish that their children would grow up to pass the imperial examinations and become officials (see Wang 1985; I’m grateful to Dr Biao Xu for teaching me about this tradition). Other traditions are much less codified. Gombrich (1972) challenged the idea that the artist could evoke the emotions s/he intended in the viewer by comparing two paintings by Van Gogh, the bedroom that expressed tranquillity and the café where the artist felt he could go mad. We know the artist’s feelings, Gombrich argued, only through letters Van Gogh wrote to his brother.

Roland Barthes’s semiological study of mid-twentieth century French culture invites reanalysis in terms of the concept of abduction, since he deals with cultural traditions where meaning is suspected but uncertain. Barthes characterized semiology as the study of:

Any system of signs, whatever their substance and limits; images, gestures, musical sounds, objects, and the complex associations of all these, which form the content of ritual, convention and public entertainment: these constitute, if not languages, at least systems of signification.

(Barthes 1967: 9)

Barthes applied this approach to advertising, clothing and food. In the ‘garment system’, he claimed, there is a ‘language’ that consists of a set of named items of clothing. Assembling one’s clothing for the day is a form of self-expression: to choose to wear a beret rather than a bowler hat is to make a particular statement, as it is to wear jeans or corduroy trousers. ‘Speech’ or ‘usage’ consists of the individual way of wearing clothes. Food comprises a similar system: there are syntagmatic chains realized by assembling a meal, and there are choices to be made between alternative dishes at each course. Bad choices may produce an ‘ungrammatical’ meal—starting with jelly and ending with soup, for example, or serving gravy with fish (cf. Douglas 1975: 249-59). Barthes also writes of furniture having ‘meanings’ (1967: 29). Barthes recognized that the identity of the signifying units needs to be studied empirically: does the choice of white or brown bread convey a different message, or are they equivalent? He does not, however, consider that for some people wearing a beret or dungarees is simply a practical decision even if for others it makes a statement.

The mere presence of sequences of actions entailing choice is not proof of a semiological system. It has been suggested that the ability to make a series of choices in the shaping of a stone tool is evidence of prehistoric mental mastery of grammar (Gowlett 1984). To make any implement, the craftsman has to put together a sequence of actions—the chaîne opératoire—that may embody practical choices at several steps. While such cognitive skills may be a precondition for modern human speech, stone tool manufacture in the Lower Palaeolithic is not proof that early hominids could represent what they were doing verbally and there is even less reason to assume each component of a prehistoric tool itself signified ideas—such, for example, as male : female, culture : nature, etc. (Dobres 2000: 155-6).

Barthes’s contemporary Georges Mounin was rightly critical of Barthes’s imprecise use of semiological terminology. Mounin readily accepted that highly codified systems such as heraldry, railway signalling, the highway code and cartography are genuine semiological systems. Road signs categorize the world into opposed ideas (advice/instruction, locomotive crossing/children crossing), each of which is represented by a sign. The use of a triangle to signify advice and a circle to signify instruction is arbitrary, but locomotives and children are represented iconically Ordnance Survey maps in Britain codify landscape features into deciduous woodland versus coniferous woodland, church with spire versus church with tower, and represent these concepts with conventional icons such round tree/conical tree. However, Mounin argued that when Barthes writes of the signification of clothing, he is using the word in a medical sense (as in ‘the spots on your body show you have measles’). In other words, clothes are symptomatic or indexical of a cultural situation, but do not necessarily signify ideas about that situation. Mounin agreed this did not make clothing uninteresting, but clothes should rather be studied as ‘diagnostic symptoms of the psychosocial ills’ of bourgeois society (Mounin 1970: 193). Barthes uses the term ‘sign’ indiscriminately, even when he is in fact discussing Peircian indices (Mounin 1970: 196), thus confusing meaning and reference. When he wrote of the semiology of clothing, moreover, Barthes suggested ‘one may say, for instance, that a certain sweater means long autumn walks in the woods’ (Barthes 1967: 43). But to whom does the sweater have this association? Possibly just Barthes and his dog! Given the human propensity constantly to be interpreting and making sense of the world, it is likely that everything around us will evoke some mental response, but many of these ‘meanings’ may be idiosyncratic to the individual.

Barthes also argued that every usage is converted into a sign of itself (Barthes 1967: 41), a claim that seems to defy the fundamental tenets of semiology set out by Durkheim and Saussure; a sign system, by definition, uses tokens of objects (words, pictures, gestures) to refer to those objects. It is the words we choose to talk about things that reveals their meaning for us (as Barthes understood; see Olsen 1990: 172). Objects can nonetheless be used as signifiers of something else. To take a local example, some shops in my home town still have oversized, nineteenth-century metal representations of artefacts hanging outside. A former café has a 4 ft wide teapot, a former shoe shop a 6 ft high boot (cf. Ruesch and Kees 1970). The boot and teapot do not signify themselves, they represent the trade conducted from the building and this would be the case even if real artefacts were used. One must have the idea of ‘teashop’ versus ‘shoe shop’ to understand their significance.

Korsmeyer argues items of food can, but do not necessarily, have meaning: croissants were first made by Viennese bakers to celebrate the Austrian victory over the Turks, while Thanksgiving and Passover are both celebratory meals. Korsmeyer (2002: 36) contends that the food eaten by a Norwegian when they get up in the morning ‘does not mean breakfast’ to a US citizen. While Korsmeyer is right to point out that breakfast is a cultural artefact, ‘meaning’ is used here in a similar way to Barthes’s notion of a sweater ‘meaning’ long autumn walks in the woods; US citizens do not associate pickled herring with breakfast. Meanings must be shown to be current. All Jews know that Passover celebrates the sparing of their ancestors’ firstborn sons in Egypt, but not all US citizens may recall that Thanksgiving is held to honour the native people who saved the lives of the Mayflower colonists during their first winter in North America. Few people eating croissants will be aware that they once signified the defeat of a people whose emblem was the crescent moon. Meanings may be lost, and normally meaningless food items can suddenly acquire signification: when France refused to back the 2003 US invasion of Iraq, over-patriotic Americans renamed ‘french fries’ ‘freedom fries.’

From Structure to Process and Post-structuralism

Such evidence that elements of culture can gain, lose or change meaning leads one to ask whether structural systems must be stable if they are to be mutually intelligible. Saussure recognized that a language gradually changes. However, like Durkheim, he saw change as evolution in the system, rather than the result of innovations introduced by individuals. He argued that individual idiosyncrasies can have no meaning, because they are not part of the conventional system. Individuals use the system, but it exists independently of them. Piaget (1968/1971: 76) attributes the predominantly synchronic character of Saussure’s structuralism to the influence of current economic and biological theory on his thought. Economists had come to appreciate that the price of tobacco in 1910 depended, not on its price in 1890 or 1920, but on then current market conditions (cf. Barthes 1967: 54-5). Equally, the function of an organ depends on its relation to other organs in the body, not on its evolutionary history. In the same way, Saussure evidently reasoned, a word’s significance depends on its relation to other words in the current vocabulary. When we look up the meaning of a word in a dictionary, we find it defined in relation to other words. A track is a well beaten path; a road is a track with a prepared surface. Saussure was unable to integrate analysis of the ‘synchronic’ state of a language at any moment and the ‘diachronic’ process by which it changed over time. It seemed as if the language must be stable if it is the function as a medium for communication.

Lévi-Strauss later struggled with the same paradox. Lévi-Strauss followed Durkheim in arguing that art can communicate only if it forms a stable system. He thus perpetuated the synchronic, even timeless, quality of structural analysis. Lévi-Strauss explained to Georges Charbonnier (Charbonnier 1961) that artists in ‘primitive’ society are careful to defend their group’s own ‘language’ because, if foreign elements were incorporated too liberally, the semantic function of the art and its role within the society would be destroyed. Lévi-Strauss saw free choice in Western marriage undermining the ability of kinship to structure social relationships and attacked the frequent proliferation of new styles among some contemporary Western artists that made it impossible to re-establish a visual language (Lévi-Strauss, in Charbonnier 1961: 93-4).

Lévi-Strauss has clearly been uneasy about the difficulty of analysing change from a structuralist perspective, and has returned to this problem in a number of his publications. He argues that certain cultures are more likely than others to achieve a stable symbolic system. The simple societies studied by anthropologists have little taste for novelty (Lévi-Strauss 1987: 275). Complex societies are more susceptible to innovation and change than simple ones. Lévi-Strauss characterized the difference as one between ‘cold’ and ‘hot’ societies. Cold societies seek to counter the effects of history on their equilibrium and continuity (Lévi-Strauss 1966: 234), while hot societies make history the moving power of their development. All societies are in history, the difference lies in how they respond to it. Those who want to sustain a stable structure must fight the contingent, non-repetitive events of history, whose cumulative effects are to produce economic and social upheavals that undermine the existing structure of culture (Lévi-Strauss 1966: 234-5).

Piaget drew a distinction between the logical structures studied in mathematics, which genuinely exist outside of time and are reversible (in the sense that subtraction reverses addition, division reverses multiplication), and the types of structure that unfold through time that are studied in linguistics, sociology and psychology The latter are not strictly reversible, but rather subject to feedback mechanisms that either preserve or modify their structure. Unlike Lévi-Strauss, Piaget therefore did not treat stability as the normal or necessarily desirable condition. He anticipated (Piaget 1968/1971: 104) that game theory would provide a more dynamic approach to the generation of economic systems by treating them as the outcome of negotiations between players in the economy and argued that analysis of the evolution of language must similarly take into account the fact that the individual develops speech through interaction with other individuals.

Bourdieu and Giddens were both critical of the timeless quality of structural analysis (Bourdieu 1977: 5; Giddens 1984: 25). Lévi-Strauss uses a single text for each South American myth that he analyses. Bourdieu argued that structural analysis of this type tends to treat variation in individual performances as though they were deviations from an unwritten score. In fact, however, there is no single, transcendent ‘myth’, only a limitless set of tellings performed by individuals on particular occasions. (For two case studies from Australia see Layton 1992: 40-4, and for a case from India see Narayan 1993.)

Turning the Durkheimian position on its head, the linguist Searle (1969) argued that language does not have any real existence outside people’s heads. A competent speaker has, by trial and error, internalized a vocabulary and set of rules for using speech, which allow meaningful and predictable interaction. Dictionaries can only report on current usage. How, then, is consensus negotiated and sustained within a speech community? Lévi-Strauss argued that, in the structure of any particular culture, ‘what we witness and try to describe are attempts to realise a sort of compromise between certain historical trends and special characteristics of the environment … and mental requirements’ (Lévi-Strauss 1987: 104). Bourdieu developed the concept of habitus to represent how this process occurred (Bourdieu 1977). Giddens and Bourdieu reject the Durkheimian notion that there is a supra-organic entity, ‘society’ or ‘culture’, which tries to maintain its own structure. Social systems are a by-product of agents pursuing their own ends through cultural strategies. (On Ricoeur’s parallel critique of Lévi-Strauss, see Tilley 1990: 58-60.) Agents use the repertoire of social strategies they have learned to construct relationships with others. In doing so, social networks are generated that, as Marx appreciated, are beyond agents’ control but enable or constrain their subsequent actions. Participants in exchange experience it as a sequence of transactions, in which each transaction is prompted by the previous offering and seeks to influence subsequent exchanges to the actor’s advantage (Bourdieu 1977: 25). Participants have a ‘practical mastery’ of how to handle social relationships. Structural-functionalists such as Parsons and Radcliffe-Brown treated social structure as a constraining force, external to human action, but it is, in fact, the agents’ activities which reproduce the conditions that make those activities possible, a process that Giddens called ‘structuration’ (Giddens 1984: 27, 162, 176).

Bourdieu studied the Kabyle people of Algeria. He describes how he and his Kabyle field assistant set out to discover the structure of the Kabyle seasonal calendar and the rituals that accompanied each phase of the agricultural cycle. Everybody gave them a slightly different answer, because each individual carried their own mental habitus. While aspects of Bourdieu’s work owe a lot to Durkheim and Lévi-Strauss, this was a radical break with their concept of the collective culture imposing its vision upon individual members. What, then, prevents individual habituses from diverging so far that no one can understand anyone else? Bourdieu argued that public rituals tend to create consensus. The culturally constructed environment had a similar effect. In the traditional Kabyle house men sit on a raised area at the back, women in a lower area in front. Children therefore grow up in a cultural environment that predisposes them to think of men as superior to women. When children grow up and build their own houses, they reproduce the material structure. Bourdieu also recognized that kin groups persist because they share a common interest in managing the group’s inherited property. Since many activities would be impossible without the help of other members of the group, ostracism is a terrible sanction and the threat of it encourages conformity (Bourdieu 1977: 35, 60, 1980/1990: 136-40; see Tilley).

Agency and the Control of Knowledge

Although Lévi-Strauss has occasionally entered political debate (e.g. Lévi-Strauss 1994), his structuralism is characteristically apolitical (Tilley 1990: 56-7). Foucault took post-structuralism further by highlighting the hidden role of power in the way that we speak or write about the world, let alone how we act. A discourse, he wrote, is not ‘an innocent intersection of words and things’ (1972: 49). Each discourse (such as the technical language used in an academic discipline) shapes the way we experience the world. The conventions of a discourse specify the objects it talks or writes about by giving them names such as madness or witchcraft. Discourse defines the topics that are worth discussing and, most important, who can speak on them with authority. The rules of a discourse also determine what positions the subject can take towards the object of discourse: as direct questioner, observer, interpreter, etc., and defines which statements are deemed valid, marginal or irrelevant. Derrida more pointedly criticized Lévi-Strauss’s naive approach to the objectivity of scientific writing. The anthropologist has the power to describe his subjects as ‘primitive’, ‘tribal’, ‘non-literate’, both through his authority as an expert on his subject and through the subjects’ inability to answer back. The indigenous cultural space ‘is shaped and reoriented by the gaze of the foreigner’ (Derrida 1976: 113). Anthropologists have become increasingly sensitive to this issue (e.g. Clifford 1986), and some at least have moved to involve their subjects as collaborators in writing anthropological accounts (see Olsen).

Structuralism and Semiotics in Anthropology: History, Current Research and Prospects

Structuralism became the dominant school in anthropology during the 1960s and 70s largely thanks to the prolific work of Lévi-Strauss. Lévi-Strauss argued that it was the structure of human cognition that generated structure in social relationships. Humans conceive of social relationships in terms of an opposition between self and other, or us and them. The exchange of gifts and the exchange of marriage partners were forms of communication between us and them that made statements about the social relationships between individuals or groups. They should therefore be treated like language, the best-studied medium of human communication (Lévi-Strauss 1952). In his mid-career work on South American mythology, Lévi-Strauss moved away from the Durkheimian/Saussurian tradition and ceased emphasizing the arbitrariness of cultural symbolism. He had discovered certain recurrent themes in the verbal imagery of the Amazon region. Legends describing the transition from nature to culture repeatedly characterized the animal condition as one in which food was eaten raw, and individuals mated at random, while the cultural condition was one in which food was cooked, and men exchanged their sisters to create political alliances (see, for example, Lévi-Strauss 1970). My own view is that, even if certain images are recurrent, they are not universal. Despite van Gennep’s demonstration that the territorial model for status change is widespread there are other common images, such as rebirth. Saussure’s observation that even signs which seem to have intrinsic significance depend on cultural convention remains valid.

Van Gennep’s The Rites of Passage was translated into English in 1960, and also contributed to the rise of structuralism in British anthropology. Mary Douglas used van Gennep’s ideas in her book Purity and Danger (1966). Douglas argued that concepts of ‘dirt’ are based not on hygiene, but on matter that is out of its proper place in the scheme of the culture. The phase of transition is the most dangerous stage in an initiation rite because it takes people out of their stable social roles. Animals that do not fall neatly into categories (such as the pangolin, a mammal with reptile-like scales) are similarly considered dangerous and polluting. ‘The whole cultural repertoire of ideas concerning pollution and purification are used to mark the gravity of the event’ (Douglas 1966: 96).

Victor Turner also highlighted the suspension of normal social regulations while initiates are secluded during ritual. People in a such a ‘liminal’ state evade or slip through the networks of classifications that normally locate positions in cultural space. They exist in a state of ‘communitas’ (Turner 1969: 95). Turner took a more positive view of this condition than Douglas, and later argued that the most creative ideas occur not during daily routines but in liminal moments such as play and joking, that facilitate fresh perceptions of social life (Turner 1990).

Structuralism has been used in numerous studies of art in small-scale societies. Rosman and Rubel (1990), for example, carried out a structural analysis of Kwakiutl art and ritual on the north-west coast of North America using ethnographic data Boas collected between 1890 and 1895. Boas documented an elaborate iconography that made it possible to identify the totemic guardian species depicted on masks and totem poles that identify the wearer’s social affiliation, lending support to Durkheim’s theory. The beaver, for example, was depicted with large incisor teeth, the hawk with a curved beak, the killer whale with a dorsal fin (see Layton 1981/1991: 151-7). Rosman and Rubel take this analysis a step further. They identify a basic division in Kwakuitl art and culture between baxus, the secular condition, and tsetsequa, the sacred condition. Baxus was associated with the totemic, land-owning descent groups. Baxus held sway during the summer, while tsetsequa was ‘on top’ during the winter. During the harsh winters, several totemic lineages congregated in defended villages, where they lived on food that had been collected and preserved during the summer. The spirit world, associated with ancestors during the summer, becomes immanent in the village during winter. Descent groups were replaced by secret societies whose membership cuts across the lineages. Unlike the benign lineage guardians, the winter spirits are fearsome cannibals or their assistants. A liminal period (cf. van Gennep) between the two seasons was signalled by the whistling sound of spirits moving closer through the forest. Finally, the spirits entered the village and captured the young people who were to be initiated that winter into the secret societies. In contrast to totemic carvings, the winter masks convey the grotesque and exaggerated world that the shaman experienced during trance. The beaks of eagle and raven masks are grotesquely elongated, embodying spirits that have come to peck the brains from the skulls of initiates.

Following Foucault, many recent writers have highlighted the role of power in the imposition of particular interpretations upon artefacts, particularly the power of colonizing peoples, and the efforts of the colonized to subvert the interpretations of the dominant community. Indigenous artefacts have undergone transformations of value as they are appropriated by different cultural traditions. The S. BLACK bag (Peers 1999) was made in western North America during the era of the fur trade. The four decorative tassels that hang from the bag are an ancient form of Native American artefact. According to Peers, such bags were often used for carrying personal possessions. But the floral patterns embroidered on both sides of the bag are of European origin, adopted by native women from around the year 1800. On the side worn next to the body, a heart is embroidered amid the flowers, probably to be read in the European sense as a sign of affection between a woman of mixed descent and her white husband (S. Black). In 1842, after its owner’s death, the bag was bought by George Simpson, Governor of the Hudson’s Bay Company, or by his assistant Hopkins. It was brought back to England as a souvenir of frontier life and belonged to the Hopkins family for forty years, then was given to the Pitt-Rivers Museum. As long as the museum’s policy was to conserve ‘authentic’ tribal artefacts the bag’s value to the collection was tenuous and it was fortunate not to be thrown out as an ‘inauthentic’ product of cultural mixing. More recently it has become appreciated as ‘illustrating’ (i.e. being symptomatic/indexical of) fur-trade society and cultural hybridity (Peers 1999: 299).

Even more complex is the history of artefacts made for the potlatch on the north-west coast of North America, that have been returned to native communities. Saunders describes how two Kwakwaka’wakw (Kwakiutl) groups arrived at different solutions for the display of artefacts returned from the Victoria Museum in Ottawa to local museums. At Alert Bay the regalia are displayed on three sides of the interior as if at a traditional potlatch, ‘mutating into priceless Art as symbols of Kwakwaka’wakw greatness—a finely wrought fabric of historical and mythological geopolitical time frames’ (Saunders 1997: 109). Saunders infers the display is directed primarily at reshaping a white audience’s perception of native culture. The Cape Mudge museum, by contrast, is ‘a cosy little place’ (Saunders 1997: 111). The artefacts have not been ‘museumized’. Each cluttered case contains the objects owned by a different family and there are no explanatory captions. To the local community the meanings are clear and carefully nurtured, but this knowledge is confined to the community, just as traditional knowledge was, and not publicly displayed.

Povinelli (1993) shows how the Aboriginal women of Belyuen, in northern Australia, live in a meaningful cultural landscape, but are conscious of the uncertainty of the meanings embodied in the behaviour of plants and animals around them. The totemic ‘dreamings’ communicate through the behaviour of animals, but the messages are unclear. ‘People must discriminate among the numerous events that occur in the countryside and decide what is a sign and what is not’ (Povinelli 1993: 692). Was the call of a dove the cry of a deceased woman to her living namesake? Was a successful yam harvest a gift from the yam ancestor? Women confirmed their egalitarianism by not asserting authoritative readings of the landscape, but made fun of a younger woman who risked a display of competence by proposing an interpretation. Povinelli’s case study exemplifies Bourdieu’s and Giddens’s arguments about negotiation and structuration. It underlines the need for future research to test the degree of consensus in any community, and how agreement is negotiated.

Two studies indicate the direction such research might take. Derrida accepted Saussure’s argument that the linguistic sign consists of a purely conventional association of sound and meaning. If, then, language is constructed through practice, and each discourse constructs a different way of giving meaning to experience, so knowledge itself is an artefact of language and as arbitrary as language (Derrida 1976: 49-50). Webb Keane (forthcoming) argues that icons and indexes are to some extent exempt from Derrida’s indeterminacy because they are anchored to their referents by similarity or association. Material objects can become meaningful, but they also exercise constraints and opportunities on their users. Keane gives the example of middle-class Indonesian men who choose different clothes for different occasions. It is easier to kneel in a sarong, and sarongs are therefore worn for Friday prayers. It is easier to sit on a chair in trousers, therefore safari suits are worn at official meetings. Thus, while meanings are still open to negotiation, clothing has an inherent potential to signify adherence to Islamic or to Western values.

Rowlands’s study of war memorials gives an excellent example of a more loosely formalized and contested tradition. Memorials function, as Lévi-Strauss might have put it, to suppress time in a ‘hot’ society by representing recent events according to the structure of classic prototypes (cf. Rowlands 1999: 132). Rowlands asks why some monuments enable people to experience healing and reconciliation while others evoke distaste and condemnation. He identifies two types of memorial: the national or triumphalist type and the type that demands recognition of what was done, to whom and by whom. Controversies over the aptness of visual form therefore partly concern competing policies, one seeking to celebrate national or ethnic prowess, the other seeking confession, expiation and reparation. Rather in the manner that Rosman and Rubel contrasted the summer and winter art of the Kwakiutl, Rowlands contrasts the Lincoln and Vietnam Memorials in Washington as examples of structurally opposed genres. The Vietnam memorial is sunken into the ground, while the Lincoln memorial is raised up. The Vietnam memorial is made of black granite, the Lincoln memorial of white marble. The former is modernist in form, rather than the classical Greek style of the Lincoln memorial. The listing of the 58,132 names of those US citizens who died in Vietnam emphasizes the reality and individuality of the deaths. The Vietnam memorial asks questions, while the Lincoln memorial ‘is an act of closure’; it does not mention slavery or the Civil War.

Rowlands shows how controversy may arise over the aptness of the chosen images for the intended message. There are guidelines for designing an appropriate memorial, but no precise rules for success. He cites the rejected design for an Australian war memorial in which a single, naked female figure was to be depicted, hanging from a cross over the broken bodies of dead soldiers. He argues the design evoked two unpalatable ideas: that the nation itself (the naked female) had been sacrificed, and that the dead soldiers remained unsanctified. Lévi-Strauss (1970: 12) assumed the mind of the analyst can discover what meanings such structural oppositions have for others, because myth operates alike in all our minds. It is important in such innovative situations not to make any such assumption, but to interview the sculptor and the users. Rowlands quotes the architect’s intentions, and relies on the observed behaviour of visitors to the Vietnam memorial to deduce their feelings.

The extent to which individual habituses overlap will vary according to the way knowledge is held and transmitted, and this is a research area that urgently needs further study. Extrapolating from Bourdieu, it could be said that there are three preconditions for the persistence of a cultural system (or should that be ‘a congeries of habituses’?):

  • It must generate behaviour appropriate to the environment. Totemism, for example, is associated with hunter-gatherer territoriality and expresses the association between groups and areas of land (see Layton 1986). Kwakiutl lineages came together in their winter villages for mutual defence against warfare (Maschner 1997), and therefore benefited from rituals that helped to merge their separate identities.
  • It must make sense of experience. Rowlands’s study of war memorials shows how memorials attempt to make sense of the loss of life in warfare.
  • It must be mutually intelligible and transmissible. Artists, for example, learn how to read the style and iconography of existing performances and use that knowledge to create new performances within the cultural idiom. These new works are in turn ‘read’ by an audience, whose expectations are shaped by their experiences. Where artist and audience share a similar habitus, readings will be more or less consistent, but the degree of consistency of readings needs to be verified and explained through fieldwork.

A good example of this phenomenon is given in Mulvaney’s (1996) and Merlan’s (1989) studies of rock art in the Victoria River district of northern Australia. Mulvaney was able to work with the children of known artists. These people were in their seventies and eighties when Mulvaney interviewed them. He found the paintings were ‘memory triggers for whole stories, events and remembering people of the past’ (Mulvaney 1996: 18). The paper deals only with paintings attributed to named people who were witnessed painting by Mulvaney’s instructors. Some depicted memorable incidents that befell particular people while foraging, others portrayed ancestral and legendary beings. Totemic beings are typically depicted as large figures, and placed in shelters associated with the travels of the ancestor in question. Painting was only ever practised within one’s own country.

Coupled with the tendency to depict ancestral beings at sites on their track, this facilitated interpretation by succeeding generations. Once incidents have faded from living memory, little can be said about paintings that record individual cases of hunting success other than general remarks on traditional hunting techniques and the importance of particular species in the diet (see Layton 1995). The Wardaman people Merlan (1989) worked with had been displaced during the violent years of early pastoral colonization (c. 1880-1930). When confronted with an unfamiliar site, Wardaman rely on the style and iconography of the art, their knowledge of the totemic landscape in which the art is placed, and the legends describing the ancestors’ travels to interpret it. They were relatively confident about arriving, through discussion, at the identity of ancestral figures. However, while a minority of the man-made figures were attributed to named individuals or categories of people, the majority were simply described as the work of ‘old people’, specific incidents having been forgotten.

Further work on the durability of interpretative traditions will clearly help elucidate the transmission of habitus in different cultural contexts.

Structuralism and Semiotics in Archaeology

During the first half of the twentieth century the dominant interpretation of Upper Palaeolithic rock art, advocated by Breuil (1952), was that paintings and engravings had accumulated more or less randomly in the caves as the product of sympathetic magic. In the late 1950s,

Laming and Leroi-Gourhan proposed a radically different, structuralist interpretation. They argued that figures were deliberately placed in certain regions of the cave and that the juxtaposition of different species expressed cognitive oppositions in Palaeolithic culture. In the French cave of Lascaux many panels seemed to repeat the same juxtapositioning of different species: horse and bison, mammoth and wild cattle. Predatory animals seemed to be confined to inaccessible locations (Laming 1962: 271-85). Leroi-Gourhan surveyed Palaeolithic cave art in France and Spain, and claimed that some species (deer, ibex, horse) always occurred in narrow passages, while others (bison and wild cattle) occurred in large chambers. Where the two classes occurred on the same panel, deer, etc., were peripheral, bison and cattle at the centre. Leroi-Gourhan claimed to have found the means to interpret these oppositions in the simple geometric ‘signs’ that earlier writers had construed as weapons or huts. In his view, they were all simplified representations of human sexual organs; ‘female’ signs were associated with bison and cattle, ‘male’ signs with the periphery (Leroi-Gourhan 1958a, b, 1964).

Leroi-Gourhan’s claims were re-examined, and rejected, by Ucko and Rosenfeld (1967). His distribution patterns failed to be replicated in caves that were discovered later. While his particular interpretations have since been abandoned, there is no doubt that only a few of the possible pairings of species actually occur in Upper Palaeolithic caves (Sauvet and Wlodarczyk 2000-01). Sauvet et al. (1977) classified signs strictly according to visual similarities in their form, without making any attempt to guess what they represented. They found statistical regularities in the association of different ‘signs’, reporting that only a limited number of the mathematically possible combinations are actually found. This led them to conclude that each of their twelve categories of sign probably constituted a single unit of signification in a system governed by ‘grammatical’ rules, but they did not attempt to infer what the meaning of the signs might be.

Following Barthes’s lead, the study of clothes and artefacts as signifiers—particularly of ethnic identity—has been taken up by several archaeologists (Wobst 1977; Hodder 1982; Wiessner 1983). There remains plenty of scope for ethnoarchaeology to elucidate how the use of artefacts as signifiers is negotiated, held constant or reinterpreted. Such findings can allow a more fine-grained interpretation of the archaeological record.

A number of archaeologists, including Tilley (1991) and Richards (1996), have extrapolated from Bourdieu’s analysis of the Kabyle house to reread archaeological landscapes as the material construction of messages about power and gender. However, the relative indeterminacy of cultural meanings, both between cultures and (as post-structural studies emphasize) within cultures, makes interpretation difficult. Tilley (1990: 67) noted the potential value for archaeology of Lévi-Strauss’s structural analyses of the relationship between native American social organization and settlement structure. Colin Richards’s study of the Neolithic landscape on the Orkney Islands of Scotland offers a definitive interpretation of one prehistoric settlement. Late Neolithic houses on Orkney have the same cruciform layout as the nearby tomb of Maes Howe. As Richards puts it, people dwelled around the ‘life-maintaining’ central hearth of the house. The tomb of Maes Howe has the same cruciform structure, but lacked a central hearth: it was a house of the dead. Maes Howe was covered by a clay mound that ‘positions the dead as being below the surface of the humanly inhabited world’ (Richards 1996: 202). Remarkably, the entrance passage is oriented so that, just before and just after the midwinter solstice, the sun shines directly into the tomb. Richards draws, not unreasonably, upon the kind of widespread imagery recorded by van Gennep, and by Lévi-Strauss in South America, to infer ideas signified by the opposition of tomb to settlement, and its orientation to the sun. The direction of the entrance is interpreted as signifying a passage rite, the death of the old year and the birth of the new; a time of celebration.

It is unlikely that most prehistoric sites will be as unambiguous as Maes Howe. Tilley used ethnographic evidence for the significance of the distribution of rock art in the landscape to derive several possible readings of 4,000-year-old rock engravings at the Swedish site of Namforsen. Tilley acknowledged that ‘one of the features of the Namforsen rock carvings … is their inherent ambiguity’ (Tilley 1991: 78). He was conscious of the power that lies in the hands of archaeologists when the site’s original creators cannot dispute much later archaeological interpretation. To ‘read’ Namforsen, Tilley argues, one must first identify the units of signification and rules for their combination into statements. The panels are complex, but certain recognizable motifs such as ships, elk and human footprints appear many times and each therefore probably conveyed a discrete meaning within a cultural structure. The justaposition of elk and fish, for example, seems to signify a cognitive opposition between land and sea. One interpretation of the various oppositions is to read them, following Lévi-Strauss, as the expression of a totemic system in which (for example) a land-oriented clan is juxtaposed to a sea-oriented clan. However, no single reading can be persuasive because the text does not seem to match what that reading predicts. ‘Understanding is a process in which we need to try out alternative readings of the text in order to see how to make sense of it from different positions’ (1991: 117). Tilley therefore turns to the ethnography of the Evenk, recent hunter-gatherers from the same region, and tries a second interpretation, of Namforsen as the liminal space occupied by shamans. Tilley finds many of the features that Turner considered diagnostic of liminality reproduced in the distribution of images.

Looking for structure in the distribution of rock art sites and the intensity of occupation, Bruno David has discussed how far into the past the ‘archaeological signature’ of Australian Aboriginal beliefs can be traced. For example, Ngarrabullgan (Mount Mulligan), in north Queensland, was avoided by local Aboriginal people during colonial times. People have lived in the region for more than 35,000 years. Ngarrabullgan was intensively occupied from 5400 bp to 900 bp, but rock shelters on the top of the mountain were then abandoned, even though other rock shelters in the region continued to be used. Current beliefs must therefore be only about 900 years old. ‘Abandonment … appears to have been mediated by the onset of a new system of signification that rendered the mountain inappropriate for habitation’ (David 2002: 46, my emphasis). Abandonment of the single site of Ngarrabullgan carries less weight than the regional rise in rock painting at contemporary sacred sites in central Australia, which David also discusses, between 900 and 600 years ago. Here, a much older stylistic tradition of geometric motifs becomes distributed across the landscape in a new pattern. In my view (Layton 1992: 231. ff), the ethnographically documented modes of mapping on to the landscape expressed through clan totemism can be traced to the appearance of the Australian ‘small tool’ tradition, associated with more permanent base camps, about 5,000 years ago. I consider David is unlikely to be correct when he claims the very foundations of the Aboriginal theory of being are no more than a few hundred years old, but do see his method for tracking change in Aboriginal cultural practices as fruitful.

In the field of structuralism and semiotics, I believe the most pressing research questions for anthropology concern the ways in which mutual understandings are negotiated and the extent to which they are achieved. Archaeological research can benefit from Anthropology but must address different questions. Archaeology needs to consider what conditions make cognitive structures apparent in the patterning of the archaeological record, and question carefully the extent to which researchers can ‘decode’ the material expression of prehistoric cultures. Archaeologists cannot gain the intersubjective insights that come from participant observation, but they can investigate how uniformity of practice is sustained for a sufficiently long time to become archaeologically visible and, potentially, what kinds of political process lead to the expression of cognitive structures in the landscape.

Archaeology needs to develop systematic procedures for assessing reconstructions of meanings from past cultures. While the physical structure of monuments and distributions of rock art in the landscape can sometimes be identified beyond reasonable doubt, are there any criteria for deciding which readings are merely possible and which are likely on the balance of probabilities to be correct? Bourdieu’s explanation for the transmission of habitus through the construction of the Kabyle house proposes a two-way process. The Kabyle house refers to men and women, and it signifies their relative status. Richards’s ‘reading’ of Maes Howe somewhat similarly depends first on identifying references that artefacts such as tombs and houses make to things in the real world (the hearth in the Orkney house, the winter sun). The meanings of these things are then inferred from plausibly universal cognitive structures. This is by no means always possible. Tilley’s analysis of Namforsen takes us further into plausible conjecture, and can draw on regional ethnography that may well have some continuity from the period when the rock art he discusses was created.

By the time we extend research as far back as the Upper Palaeolithic, the structure of meaning is irrecoverable. We can recognize paintings in Upper Palaeolithic art that refer to horse and bison, but we do not know what horse and bison signified in the cultures of the Upper Palaeolithic. Breuil’s assumption that Upper Palaeolithic cave art accumulated randomly can be challenged through statistical analysis, but Leroi-Gourhan’s interpretation of simple, variable and ambiguous geometric motifs as signifiers of gender was hopelessly optimistic. Shamanism has been argued as a motive for Upper Palaeolithic art (e.g. Lewis-Williams 2002), and this line of research has offered another way of investigating the distinctive cognitive structure of Upper Palaeolithic culture. Since the Upper Palaeolithic art of France and Spain represents the earliest incontestable art tradition, and is associated with the arrival of fully modern humans in western Europe, the structure of the art (expressed in the range of motifs and their distribution within and between sites) and the evidence that gives of cognitive skills, are probably of more interest to modern researchers than the specific meaning any images may have held in their original cultural context.