From village to bush in four Watchi rites

This work is licensed under the Creative Commons | © Klaus Hamberger. ISSN 2049-1115 (Online). DOI: http://dx.doi.org/10.14318/hau4.1.005

From village to bush in four Watchi rites

A transformational analysis of ritual space and perspective

Klaus HAMBERGER, Laboratoire d’anthropologie sociale, EHESS

Understanding ritual performances in terms of changes of perspective is increasingly common in anthropological analyses. However, less attention has been paid to the fact that perspectival transformations not only make up the internal dynamics of a given ritual but also connect it to other rituals. Drawing on a series of related Watchi-Ewe rituals (divinatory, initiatory, funerary, and hunting), this article proposes to analyze ritual space as a system of perspectival transformations operating both within and between rituals. By conceiving of each ritual as constructing the same relational architecture from a different point of view, it becomes possible to understand the relationship between female diviners and male hunters within the context of a larger set of interconnected relations (between men and women, humans and animals, masters and slaves, the living and the dead), realized in the virtual space of ritual performance. Understood as a controlled variation of perspective, the transformational analysis of ritual thus becomes a valuable methodological device, both to elucidate the model through which a society conceives its relational universe and to render intelligible seemingly contradictory and otherwise unexplainable ethnographic facts.

Keywords: Space, ritual, transformation, Ewe, Watchi, Togo

Understanding ritual performances in terms of changes of perspective is increasingly common in anthropological analyses (Kapferer 1997; Viveiros de Castro 1998; Empson, Humphrey, and Pedersen 2007; Ishii 2013). However, less attention has been paid to the fact that perspectival transformations not only make up the internal dynamics of a given ritual but also connect it to other rituals, each of which can thus be understood as enacting the same relations and processes from an alternative point of view. Such an approach takes proper account of the common ethnographic observation that people themselves frequently explain their rituals by reference to other rituals—as, for example, when Watchi Ewe of Southeast Togo present a hunting ceremony as a mortuary rite for animals, or describe a female initiation rite as a replica of the burial reserved for victims of violent death. As I shall argue, it is the total set of such transformations, both within and between rituals, that turns the multiplicity of ritual perspectives into a space—not just in the sense of an abstract analytical structure, but as the concrete experiential architecture of ritual performance. Conceptualized as a system of perspective transformations (rather than as their invariant framework), ritual space not only becomes accessible to transformational analysis but also provides the latter with a new grounding, by turning symbolic oppositions into vectors of intersubjective relations. Thus reshaped and reoriented, the transformational analysis of ritual space becomes a valuable methodological device, both to elucidate the way in which a society conceives its relational universe and to render intelligible seemingly contradictory and otherwise unexplainable ethnographic facts.

The transformational concept of space

Ever since Arnold Van Gennep’s Les rites de passage (1909), space has been central to the study of ritual. Not only has the focus on spatial structure and patterns of movement proved to be a fecund method for analyzing rites of passage, but generalizations of Van Gennep’s model by anthropologists such as Victor Turner (1967) have increased the importance of spatial concepts for ritual theory to the extent that David Parkin (1993) proposed that ritual action might be conceptualized entirely in terms of movement, directionality, and spatial orientation. While this mainly meant that space was appreciated as the principal medium in which ritual action takes place, an increasing number of studies have concentrated on the way in which space itself is shaped and produced by ritual.

An important step in this direction was taken with the introduction of Susanne Langer’s (1953) concept of “virtual space” into the theory of ritual (Williams and Boyd 1993; Kapferer 1997, 2004).1 Langer developed this concept to analyze the working of (plastic) art. According to her theory, art uses sensory and symbolic devices (including, but not necessarily, figurative representation) in order to produce a configuration of imaginary relations exhibiting a genuine spatial structure. This virtual space differs from an optical illusion in that it is not continuous with the physical space in which the artwork is located, but constitutes a self-contained, autonomous system, a full-fledged form of perception, which, however, is not organized in the same way as is ordinary vision. This peculiar form of perception serves a symbolic function, not in the sense that it represents external objects, but in that its structure corresponds to the dynamics of subjective experience—the morphology of feeling and emotion, rendered perceivable in the work of art (Langer 1953: chapter 5). Langer’s model was innovative in more than one respect: it conceptualized representation as the means rather than as the aim of artistic creation, perception as its product rather than as its origin, and emotion as its meaning rather than as its effect. Moreover, it analyzed “emotion” as having a relational dimension, especially when dealing with the virtual space of architecture, considered as an embodiment of the patterns and rhythms of social or religious life.

The fruitfulness of this model for conceptualizing how ritual works is obvious (see also Williams and Boyd 1993: 145 f.). First, it dissolves the opposition between two seemingly different interpretations of ritual: one that sees ritual as constituting an autonomous realm and the other that sees ritual as pointing to an external reality. The representational elements in ritual act as devices to produce a self-contained relational structure (an approach further elaborated by Houseman and Severi [1998]). Moreover, by conceiving of this relational structure both as a model of social relationships and as a mode of spatial experience in and of itself, it helps us to understand how ritual, without necessarily producing the social relationships it models (any more than art necessarily produces the feelings it expresses), can nonetheless translate them into sensory perceptions, much in the way Pierre Bourdieu (1970) has shown for Kabyle domestic architecture. As Bruce Kapferer (2004: 44) has pointed out, Bourdieu’s study of the Kabyle house can, in this respect, be considered as a model for the analysis of ritual.

Although most anthropologists today would agree that movement is the primary form through which ritual space is created and social relations are embodied, another key insight of Bourdieu’s seminal study has been relatively neglected. Movement not only mobilizes a spatial relation (between departure and arrival), but also entails changes of perspective, that is, transformations of the totality of relations that make up a given space. One of the merits of Bourdieu’s study is to have demonstrated that a transformational concept of space is not restricted to abstract geometry but is equally applicable to social space. At the same time it provided a spatial interpretation of the Lévi-Straussian notion of transformation, affording the latter a coherent epistemological framework and in a sense putting it on its feet. However, Lévi-Strauss’ (1971: 603) verdict that ritual was not only different from but opposite to the logic of mythical thought persistently inhibited the application of transformational analysis to ritual. Such approaches remain few in number (see, for example, De Heusch 1980; Albert 1985; Désveaux 2001), and still fewer of them focus on spatial structure.

One notable exception is Stephen Hugh-Jones’ (1995) exploration of Barasana ritual, which shows how changes in ritual perspective entail inversions of ritual space. The orientation of the ceremonial house changes from upriver-facing to downriver-facing, according to whether it is envisaged from the inside or from the outside. These transformations in turn involve others: the house opening becomes either a male mouth or a female vagina, and so on. Hugh-Jones’ analysis makes clear that, without a methodology that recognizes the importance of variable perspective, the superimposition of inverse symbolic structures on the particular ritual space that is the Barasana ceremonial house would have remained a mass of contradictions (which in turn might well have been interpreted as a sign of “liminal” antistructure).

This result can be generalized. Much of what has been conceptualized as a ritual condensation of opposite relational poles (Houseman and Severi 1998) may actually be the sign of a change of perspective, which, far from being an instance of antistructure, can provide a key to the understanding of both the internal structure of a given ritual and the way in which rituals are linked together.

This multiplicity of perspectives is constitutive of the structure of ritual space. Rather than model the fabric of social relations exclusively from one dominant point of view, ritual allows its participants to perceive it from a variety of perspectives; and it is through the controlled variation of these perspectives that its distinctive topology is shaped. Ritual space thus provides not so much the miniaturized image of a social and symbolic universe as the generative scheme that underlies its construction—a construction that proceeds by transformation, both within and between rituals. It is by virtue of a transformational conception of ritual space that the latter can be understood, not just as a spatialized image of a symbolic structure deriving from the model of language, but as a spatial structure in its own right, whose elements are alternative viewpoints, and whose dichotomies instantiate relations between the self and the other.

One of the major insights offered by studies on Amazonian perspectivism (Viveiros de Castro 1998) is that relations of alterity, insofar as these correspond to transformations of perspective, always imply some sort of attention, aggression, or desire—the paradigm being the relation between predator and prey. Correspondingly, there is an intimate link between perspectivism and hunting, not necessarily as an actual mode of production, but as a relational model (ibid.: 472). These results draw on data from a particular cultural area and refer to cosmology rather than to ritual. However, once we consider ritual space, we find the connection between hunting and perspectival transformation evidenced in many cultures far beyond those with explicit “perspectivist” cosmologies or hunting economies. The perspectival approach is in fact not so much bound to specific cultural areas as to specific relational contexts. In West Africa no less than in Amazonia, rituals that deal with hunting, homicide, and violent death abound with perspective inversions, while rituals that focus on persons capable of a double perspective, such as diviners, spirit mediums, or twins, are imbued with the symbolism of hunting; and all of them are organized around a fundamental division of ritual space into a human and a nonhuman realm—“village” and “bush.” This is not just because the frontier between the home and the wild is a particularly salient division. Strictly speaking, it does not constitute a specific division at all, but rather a general relational schema that can be applied at any level and in any perspective: “village” and “bush” are the respective places of the self and of the other.

The relation between humans and animals is only one way in which this general schema is realized. Others include relations between the living and the dead, and between men and women. The mutual combinations and reiterations of these various dichotomies engender complex topologies of houses and cemeteries, parlors and chambers, courtyards and enclosures. As has long been remarked by anthropologists (see, among others, Mosko 1985 and Strathern 1998), the generative logic underlying these topologies often entails some characteristic inversion of the inside/outside dichotomy—as, for example, when the innermost part of a woman’s domicile is symbolically aligned with the “bush.” Transformational analysis helps us to understand these inversions, which, far from rendering the notions of “inside” and “outside” ambiguous, in fact give them their precise meaning.

The ethnographic setting

I will start with an enigma of the sort just described. It consists in the prevalence of hunting symbolism in the rituals performed by a class of African female diviners using bush spirits—such as the Ewe amegasi, the Mossi kinkir-baga, the Senufo sando, and the Lugbara ojou. As I have already indicated, the connection of bush spirit divination with hunting is quite common in Africa. The spirits who initiate, help, or possess the diviner are frequently the masters of game animals; and in several societies the diviner’s career starts when he kills an animal whose spirit guardian, after haunting or kidnapping the hunter, becomes his auxiliary. Hunting, however, is a male domain, so this does not apply to female diviners. Yet the spirits they work with, and the symbolism of their divinatory practice, are also associated with the hunting sphere.

One possible solution to this apparent contradiction might be sought in a gender ambiguity that is characteristic of the diviner’s liminal state (cf. Peek 1991). But this would not explain why, while female diviners, as we shall see, may ritually transform themselves into men, they never assume the role of hunters. Another explanation might draw on analogies between hunting and childbirth as activities intimately linked to fortune and peril. The initiatory “madness” of Mossi female diviners is usually released by complications following delivery (Bonnet 1982: 79 f.), and in several West African societies the bush spirits engaged in divination are both the guardians of wild animals and the carriers of unborn children (see Hamberger 2012 for a discussion). Still, this approach would not explain why the diviner’s auxiliary spirits are located at the intersection of male and female domains, as allies (or enemies) of both hunters and mothers.

There may be several other hypotheses to render account of the relational configuration that connects women, hunters, and spirits, none of which are satisfactory, precisely because the relational configurations enacted in ritual cannot be understood in and by themselves. To understand them, we have to adopt the shifting perspective methodology of ritual itself, while at the same time extending it beyond the single rite to explore the configuration as it is transformed across a series of connected rituals.

The rituals that will be analyzed in this article are taken from the ethnography of the Watchi Ewe of Southeast Togo.2 Like many other West African societies, Ewe conceptually organize their environment by means of a dichotomy between the village or house (aʃe) and the bush (gbeme or gbedzi), which is applied to numerous conceptual domains (animals, vodus, the dead, etc.). As already indicated, this constitutes less a determinate division than an overarching scheme of relative spatial construction, as becomes clear in the use of the terms: aʃe means “home” (the place of the subject), while gbedzi means “the outland.” In Ewe as in other West African societies, the “bush” is the place of the other (see Cartry 1979). This, however, does not mean that it is simply defined as land beyond the village. Whether as the home of the animals and bush spirits from whom humanity obtained its cultural techniques, or of the autochthonous population that first received the village founders, or of the Northern slaves that became the ancestresses of today’s villagers, the bush is the condition of the village rather than its negation.

This relation also holds in the context of ritual. Here, the “bush” generally corresponds to an area physically or symbolically separated from its profane environment, and at times partly inaccessible to the audience located at the “village” side of the frontier. This detachment from actuality endows the bush with the characteristic otherness that Langer (1953: 45 f.) has emphasized as a condition of virtualization. In this sense, the virtual space of ritual is constructed from the bush, centered on the other’s place. This does not mean that the village remains outside virtual space—virtual space has no outside. However, as noted by Langer (1953: 95, 72), the virtual space of religious architecture integrates actual space as its context and frame. Contrary to the virtual space of pictorial art, it contains its own border. In fact, the basic difference between the rituals we shall consider consists in the way in which this border is crossed.

In the four sections that follow, I will outline the architecture of Ewe ritual space through a series of rites (divinatory, initiatory, funerary, and hunting), considered as complementary aspects of one and the same relational structure. As I shall endeavor to show, these different aspects can be transformed into one another by inverting the basic orientation of movement that takes place within this structure. This systematic variation of perspective will enable us to understand that the female diviner’s relation to hunting is not a residual or analogue of the male diviners’ role as (real or symbolic) hunters, but its inversion: rather than a feminized hunter, the female diviner is a humanized bush creature. This conclusion cannot be drawn solely from the divination rite itself, which gives us only some hints to it. It emerges gradually as, passing through its serial transformations, women appear as foreign slaves, as victims of violent death, and as wild animals.

Female divination (xɔyɔyɔ)

Ewe divination practices are dominated by the system of geomancy known as “Afa,” which they share with the neighboring Fon and Yoruba (see Surgy 1981). There is, however, another type of divination, a form of necromancy called xɔyɔyɔ (literally “to call in the chamber”), whose practitioners—almost all of them women—are called amegãsi (hereafter written as “amegasi”).3 These female diviners work with the help of a bush spirit, called “Age” or “Aziza,” who serves as a messenger between the living and the dead. Like his Mossi and Senufo counterparts, Age is a small, dwarf-like creature, and like his Lugbara homologue, he appears cut in half with a single leg. At the same time, every amegasi is the adept (vodusi) of a spirit (vodu) that is connected to the bush dead (dzogbeku, a category comprising all those who have experienced violent deaths) and that is transferred—albeit discontinuously— along the female line (as among the Senufo, the diviner’s vocation stems from a uterine ancestress). Transmission of this spirit announces itself by illness or other calamities within the maternal family, and culminates, after a long period of initiation, in the erection of a shrine4 within the home of its new custodian. The Watchi Ewe refer to this particular vodu as “tro” (trɔ̃); accordingly, Watchi amegasis are also commonly called “trosi” (trɔ̃si).5

Although the title “amegasi” can be used of a wider class of diviners, it is applied particularly to necromancers. Yet we also find this term in ethnographic materials collected by Jakob Spieth in the early twentieth century (1906: 896) where it denotes a person possessed by, and later becoming a priest of, the bush spirit Ade. The distinctive marks of Ade priests reported by Spieth—such as the raffia cord, the cowrie bracelets, and the prohibition on eating food cooked by other people (ibid.: 850, 896, 912)—are shared by present-day amegasi diviners. However, Spieth makes no mention of divination in this context. According to him, Ade confers on his priests success in hunting, not the power to summon the dead. Hunting connotations still characterize present-day Ewe bush spirit divination. Age—the bush spirit whom Spieth’s informants called Ade (literally “hunter”) and who is the messenger of contemporary amegasis—remains intimately linked to hunting: he reigns over the bush, protects the wild animals, and thwarts the plans of any hunter who has failed to turn him into an ally. However, while hunting among Ewe was and remains a male activity, today’s amegasis are almost exclusively female.

In principle, xɔyɔyɔ divination consists in the evocation of any soul (se), including that of a living person. However, in the vast majority of cases, it is the dead who are consulted. The séance takes place in a rectangular hut within the amegasi’s walled courtyard (or “convent,” kpame). The hut is divided into parlor and chamber; a white blanket covers the door connecting the two rooms (see figure 1). While the clients take their place in the parlor, the amegasi enters the chamber, where she will remain hidden for the duration of the séance. Similar arrangements are reported from elsewhere in Eweland (Spieth 1906: 490–94; Pazzi 1976: 300; Amouzou 1979: 197–98.)

Figure 1
Figure 1: Necromantic session

Recent ethnographic accounts of amegasi divination are remarkably sparse in number,6 which might appear to confirm Albert de Surgy’s (1981: 10) prediction of the institution’s rapid decline. Such a conclusion, however, is belied by a vivid regional network of tro diviners who are regularly consulted for a variety of reasons (from vodu installation to classmate seduction), most frequently to inquire into the reasons of the death of a close relative. The following description is based on two sessions I attended at Afagnan-Gbléta in April 2004 and in August 2006.

The session starts with the sound of a bell that the amegasi rings to summon Age. His high-pitched voice is heard immediately, asking the clients to explain the nature of their request. They tell the bush spirit where to find the dead they wish to contact (that is, the houses where the dead once lived) and negotiate a moderate fee (some hundred francs CFA) for his services. Contrary to other West African forms of bush spirit divination, Age merely acts as a go-between and is not consulted in his own right. In addition to his faculty of speech, his principal qualification for performing this function is the extreme rapidity with which he is able to cross territory, a speed approaching bilocation. He returns a few moments later, announcing the deceased’s arrival. This is preceded by the sound of a rattle that the amegasi shakes, and continues to shake for as long as the dialogue between the deceased and the clients lasts. The acoustic contrast between the sound of the bell and the sound of the rattle corresponds to that between Age’s voice, which is rapid, high-pitched, and nasalized, and that of the deceased, which is slow, hollow, throaty, and frequently unintelligible (so that the amegasi has to intervene as interpreter and commentator). The deceased’s voice is said to emanate from the tro installed in the chamber.

The tro chamber is forbidden to any person other than the priestess. However, on certain ritual occasions the tro is brought out and shown to the public, so that we can describe some of its external characteristics. The tro is a huge calabash, wrapped in a white percale blanket, whose only visible content is a long stick of Zanthoxylum wood (considered as a defense against witchcraft) that protrudes from it at both ends. The nature of the other contents of the calabash remains obscure. Some amegasis told me that it contains the clothes they wore and the utensils they used during the initiation ritual at the bush cemetery. Certain origin stories speak of calabashes containing innumerable twin puppets (a common representation of Age). There also is a rumor that the key component of the tro’s contents is a human skull (which talks when animated by the soul of the dead). Surgy (1988: 215 f.) describes the necromancer’s shrine in Southwest Togo as a clay idol in the shape of a head, elevated on a “dish” (which could be a large half-calabash), and covered with a white blanket. In accordance with my own information, the whole rests on a table or rack adjacent to a shrine of Age, which Surgy describes as a couple of twin figurines.

Imaginary or real, these traits are well integrated into tro symbolism. The idea that the calabash contains clothing worn by the novice during her initiation suggests a relation of equivalence between the tro and the trosi’s body. Moreover, in Ewe and neighboring cultures the calabash is a common symbol of the female womb, and more generally of the woman. The equivalence between trosi and tro is corroborated by the image of the head described by Surgy, and compared by him to an idol used by Afa geomancers as a materialization of their own personal soul or destiny (se). Be it as a womb (the calabash) or as a head (the clay idol or skull), the tro represents the amegasi’s body—a body shared with the dead she evokes. The corporeal connection between trosi and tro can also be observed during the nocturnal procession when the newly consecrated amegasi, leaving the convent of her apprenticeship, carries the tro on her head to its new home, dancing, swinging her hips, and swaying the calabash gently. This movement ceases abruptly the moment the calabash is taken from her: symbolically beheaded, the trosi falls like a corpse into her companions’ arms. Here, the calabash appears not only to contain but actually to replace the head.

It would, however, be insufficient to consider the relation between trosi and tro only with respect to the calabash. As one amegasi put it, the core of the tro resides in the gokɔnu (“lap pocket”), the apron knot at the level of the pubic region, where the amegasis put the wooden figurines representing Age. The image evokes a maternal relation—indeed, in this context, Age was explicitly characterized as the amegasi’s child.7 This relation has another, more particular dimension, however, because the wooden puppets produced in Eweland—generally referred to as togosu, a term meaning “twin”—represent dead twins (replaced by puppets in order to deceive the surviving twins who otherwise would follow them). Now, the idea of a dead person worn in a container that serves as an external representation of its carrier’s womb also characterizes the tro calabash. It is thus hardly surprising that certain origin tales state that the calabash contains twin puppets. The tro calabash and the twin figurine represent the two poles of the maternal relation (womb and child), and at the same time two alternative ways of connecting the living and the dead. While the tro calabash accomplishes this function as the single body of two persons, Age does so as a single person with two bodies. This difference can be restated in terms of movement: while the calabash serves explicitly as a trap immobilizing souls, blocking their return from the living to the dead, Age’s extreme mobility enables him to bring the dead to the living.

This movement is hardly visible in physical space. Ewe bush spirit divination is characterized by marked disembodiment. It involves neither dance nor trance. There is almost no bodily motion: the amegasi remains seated on a stool or mat throughout the ritual, and has no visual contact with the client. The ritual takes place in a closed chamber behind a veil, in a closed hut within a walled courtyard within a house within the village. One can hardly imagine a more interior place. Yet the tro is connected to the bush dead buried outside the village, and Age is known above all as the protector and guardian of the wild animals. This association of the extreme inside (chamber) with the extreme outside (bush) is characteristic of the amegasi’s liminal position. But characterizing it in this way does not yet mean understanding it.

The amegasi’s divinatory practice in bringing the dead to the living, and the bush to the village, cannot be fully grasped without taking into account a second and inverse movement—that of bringing the living to the dead. This inverse movement is a constitutive part of the amegasi’s initiation, which takes place symbolically in the very heart of the bush; and in this ritual, movement from one realm to the other is visibly acted out.

Diviner initiation (dzogbedɔdɔ)

The formation of an amegasi is by far the most testing and most expensive of all vodusi apprenticeships in Southeast Togo. Its median duration is seven years, and some amegasis have spent more than twenty years in their mistress’ tro’s “convent” (kpame, lit. “enclosure”). Throughout the apprenticeship, the novice is subject to strict rules of obedience, poverty, and chastity. She wears a dirty white blanket around her haunches, a cotton cord around her neck, and her long hair is matted into dreadlocks. Once the training is complete, however, her appearance changes radically: she is adorned with necklaces of precious beads, perfumed, painted, sprinkled with talcum powder, and literally wrapped in money in the form of numerous rows of cowrie bracelets around her ankles, wrists, and arms.

This contrast is linked to the character of the tro. This particular vodu is closely associated (at times almost identified) with Tchamba, a vodu constructed by the descendants of slave owners (the term “Tchamba” refers to the region of northern Togo considered to have been the principal source of slaves). The cowries on the amegasis’ arms symbolize the money with which their ancestresses purchased slaves, the highest expression of family wealth; some amegasis perform while sitting on the ancestress’ cowrie-adorned stool, a component of the Tchamba vodu. At the same time, however, Tchamba is also a vodu of slave descendants, thus incorporating simultaneously the spirits of slaves and of their masters (Brivio 2007; Hamberger 2009). Like the tro, Tchamba is transmitted along female lines, as were formerly money, beads, cowries, and slaves. And because slaves (at least theoretically) became full members of the uterine family of their buyers, a female slave’s descendants are also the descendants of her mistress, and both—the slave and her mistress—become exchangeable in the Tchamba vodu.

As noted earlier, the embodiment of the two opposite poles of a relation in one and the same vodu is also characteristic of the tro. The tro serves as a common body of diviner and invoked dead, just as Tchamba represents both slave owner and slave. It is noteworthy in this respect that one of my interlocutors identified the secret content of the tro as a slave’s skull. Moreover, the uterine bond that links the amegasi to Age as her child also links her to the tro as the embodiment of her ancestress. Despite its individualized character and the fact that it is destroyed after its priestess’ death, every tro is supposed to “return” within the same uterine family, which has then to reconstruct the vodu and provide another amegasi to replace the deceased.

The final rite of passage of an amegasi novice (which includes the construction of a new tro) consists in spending one night at the bush cemetery, the dzogbe. The term dzogbe (lit. “bush of fire”) denotes, in its narrow sense, the burial place, usually not far from the village, of those who have suffered a violent or unnatural death. In its wider sense, the term refers to the area beyond the range of human habitation and cultivation that is the domain of bush fires, wild animals, and hunters (significantly, death during a hunt is a “bad” death). Still more widely, dzogbe figures in the term used for the wilderness of the north (dzogbedzi, lit. “on the bush”) where Ewe locate the unknown home of slaves. Whether as origin or as final destination, the dzogbe is the place of those who die away from home.

It is in this no-man’s-land that the amegasi novice undergoes her transformation from epitome of slavery into allegory of wealth. I base my analysis on the ritual I observed at Tonoukondji in August 2006. The ceremony starts at midnight in front of the convent, where the novices are wrapped, like corpses, in burial mats and white burial cloth, similar to the percale that wraps the tro calabash. Senior amegasis hold up a ring of cloth to shield the novices from external view. Thus transformed into lifeless parcels, the novices are carried in a funeral cortege to the bush cemetery, where they are placed within a circle drawn with flour on the ground, around which the senior amegasis sit on their stools. Spectators and a set of drums are installed at some distance behind a straight line of flour, the space between the drums and the “corpses” serving as a dance floor (see figure 2). As the drums intonate the gaʋu (“iron drumming”) rhythm, pairs of amegasis leave the circle at regular intervals and dance toward the drummers, always returning just before they reach the “borderline.” When, at a certain moment, the drums switch to the gbenyanya (“battue”) rhythm, all the amegasis run around the novices. They then perform a war dance at the border, before rushing into the surrounding bush. The same scene is repeated just before dawn, after more dancing, but this time the amegasis break through the border, and hit the spectators with leaves and branches before rushing back into the bush. At sunrise a ring of cloth is again formed around the novices, who emerge upright, each with an iron cane in her hand, and finally return to the convent. The consecration of the calabashes takes place in front of the convent some hours later. Again, the novices stand in the middle of a circle while the senior amegasis dance around them. Many of the dancers hold a ram or goat between their legs. Afterward the throats of the animals are cut, and the amegasis, followed by the novices, carry the carcasses around the calabashes, tracing a circle of blood on the ground while singing “we are a ram” (miawoe nye agbo), a well-known vodu song. The calabashes are then carried into the convent, followed by their owners.

Figure 2
Figure 2: Amegasi initiation

My observation of the ceremony largely confirmed what I have been told in interviews, with one important exception: all my interlocutors had affirmed that the novices remain alone in the bush, in the exclusive company of the dead,8 these being visible only to the novice who spiritually visits the netherworld while physically “sleeping” in the bush (the name of the ritual is dzogbedɔdɔ, “to sleep at the bush cemetery”). Yet the behavior attributed to the dead by my interlocutors corresponds closely to that of the amegasis I observed at Tonoukondji: the dead were said to form a circle around the novices, to sit on stools, to dance, and to fight with the humans who try to retrieve their “sleeping” children by chasing the dead away. The battue dance (gbenyanya, literally “chasing the bush”) is in fact commonly interpreted by Watchi Ewe as a fight between the living and the dead. Clearly, then, the women I saw rushing among the spectators and brandishing vegetation represented not so much the attacking (living) humans as their adversaries. The gbenyanya dancers assume a human or nonhuman aspect depending on the particular direction they take (from village into bush or vice versa), and the bidirectional movement of the dance is a diachronic representation of what is actually conceived of as a simultaneous confrontation. This twofold orientation is also inherent in the relation of equivalence between the trosi and the tro, which is most manifest in the dzogbe ritual. Not only are both wrapped in burial cloth but the dead are said to construct the tro at the bush cemetery (though the tro is actually assembled in the convent) from things wrapped in the same blanket as the novice. Again, aspect corresponds to direction: while the trosi is transformed into a corpse in going from the village to the bush, the corpse is transformed into the tro in returning from the bush to the village.

The initiation ritual is thus structured by three successive scenes, linked by converse movements across a diametrically divided space (see figure 2). This fundamental structure is drawn on the ground: the line of white flour separates the sphere of the village, occupied by the living (the spectators and the drummers), from the sphere of the bush, reserved for the dead (the novices/corpses and the amegasis/ghosts). The three scenes take the form of three circles, each of which isolates, as it were, a particular term of the trosi/corpse/tro equation. First, a circle of cloth surrounds the novices on their burial mats in the village; then, a circle of white flour surrounds their corpses in the bush; and finally, a circle of animal blood is traced around the tro calabashes back in the village.

Each circle is produced by the senior amegasis (upright holding the cloths, sitting on their stools, or dancing with rams between their legs). However, while the amegasis encircling the corpses in the bush represent human beings who have suffered violent deaths, those dancing around the calabashes in the village identify themselves with animals that have met violent ends. True, the song that goes “we are a ram” is not peculiar to tro rituals. But the image of dancing dead animals emerges clearly from the account one amegasi gave of the vision she had during the night in the bush. The dead who receive the novices in the netherworld were described by her as animals, dancing in front of the novices and serving them food. Yet communion with wild animals carries the potential for a deadly outcome: they will devour any novice who has violated a taboo during her apprenticeship.

This interpretation gives further significance to the fact that the dawn battle between the living and the dead is conceived of as a collective hunt. It also explains why the tro, a vodu of those who have died a “bad” death, is intimately connected with Age, the protector of wild animals. And it clarifies why the amegasis observe strict taboos on eating game, and wild animals generally. Except on certain ritual occasions, amegasis are also forbidden to eat ram, an animal that, though domesticated, is associated with the bush. Living in Age’s home—the termite hill—the ram is conceptualized as a transformed buffalo (which in turn bears the surname agbo, “ram”). In fact, the identification of the amegasis with rams seems to be a particular instance of a more general identification with wild animals that is established during the night in the bush. At an enthronization festival I witnessed in summer 2007, the newly consecrated amegasi was called from her enclosure by a song implicitly equating her with wild game that hunters drive from the bush with fire: “the fire is falling upon the bush, the animals are hiding, if there is a hidden animal, get out!” (edzo dze gbe, e lãwo be, ne lã ɖe bea, ne to!). “Bush of fire” is the literal meaning of dzogbe, the bush cemetery (also called akladzame, “hiding-place”); and to! (“get out!”) is the standard call of beaters during a drive. Thus, the gbenyanya attack, which concludes the night in the bush cemetery, does more than chase away the beasts in order to retrieve the novices—it also targets the novices who are assimilated to beasts. In effect, the novices are targeted by humans and animals, by the living and the dead, and both parties are represented by the same group of women.

The simultaneous identification of the amegasis with rams and ram killers has yet another aspect because rams can also represent human beings. Ram sacrifices are generally considered to substitute for human sacrifices, and the term agbota, “ram head,” is a euphemism for a human skull. The amegasi who recounted her vision of the animal dance did not mention rams specifically, but spoke of beasts being “killed for the talking drums” (taʋuga). Now, while the usual victims in talking drum rituals are domesticated goats, they nevertheless represent wild animals, and their immolation is staged as a mimed shooting, accompanied by hunting songs. Moreover, the “animals” originally killed for the talking drums are said to have been humans, bought as slaves or hunted in the regions of the north (dzogbedzi). Today’s ceremonies recall this history: the hunting mime that accompanies the immolation is followed by the representation of a slave being captured for the purpose of sacrifice.

This association of wild game and slaves also characterizes the amegasi’s enthronization ritual. The “animal” that emerges from its hiding-place (the convent) is a slave’s descendant, and its coming out precedes the immolation of sacrificial animals inside the tro enclosure. However, far from being shot or maltreated as in the talking drum rituals, this “animal/slave” is placed on the ancestress’ throne, which will serve as her seat in the necromantic sessions.

This multiple condensation of opposite relational poles—masters and slaves, hunters and animals, killers and victims—is more than the transitory effect of the liminal stage common to all rites of passage. While the three-staged process from the village to the bush and back that characterizes the initiation ritual conforms to the classic structure described by Van Gennep, the symbolic death followed by rebirth involved in the ritual does not just transform a novice (slave) into a priestess (master). The full-fledged amegasi is a slave on the master’s throne, a dead person among the living, an animal in human form—at least during the necromantic sessions. The liminal state is here perpetuated as a permanent latent condition that is reactivated every time the amegasi enters the “hiding-place” of the tro chamber, sits on the slave’s stool, and speaks in the foreign voices of the dead and wild animals.

Yet the amegasi’s role in divination does not just repeat her role in the initiation ritual; it inverts it. In both cases, ritual space has the same basic structure. The veiled door between chamber and parlor in the tro hut replicates in miniature the line of flour separating the bush from the village. However, the sphere of the bush is projected into the inner chamber, while contiguity with the village characterizes the outer parlor of the tro hut. This spatial inversion corresponds to reversals in movement. In the initiation rite, the amegasi twice crosses the border, becoming a guest of the dead and the wild animals while “sleeping” in the bush. In the divination session, she remains at home while the dead and the bush spirits come and go. Initiation and divination are complementary structures, belonging to a single pattern of reciprocal crossings of the border between bush and village. To explore this architecture further, let us now consider it from yet another perspective.

Funeral in the event of violent death (gaʋu)

The mock burial of the trosi novices at the bush cemetery draws on the model of the funeral ceremony for actual victims of violent or unnatural death. Yet in some respects both rituals are directly opposed. The dzogbe burial ritual is commonly called gaʋu (“iron drumming”) or kpoʋu (“stick drumming”), after the special drum rhythm for these occasions, sometimes also gbenyanya, after the battue sequence of the dance. These concepts are closely connected: the “iron” machetes and wooden “sticks” brandished by the dancers represent not only the weapons of warriors but also the instruments beaters use during a drive. This contrasts with the amegasis’ variant of the gaʋu dance, in which the dancers representing warriors are unarmed. A more obvious difference is of course the sex of the participants: while the amegasi dancers are exclusively women, the “original” gaʋu dancers are all men, and the gaʋu songs explicitly describe violent death as male.

Dzogbe burials are systematically performed in all cases of violent death (today most frequently road accidents). The following account is based on two cases I witnessed at Afagnan-Gbléta in August 2005 and in July 2006. As in the case of the dzogbedɔdɔ ritual, the funeral site is divided by a straight line traced on the ground; the section closer to the bush contains the corpse, while the drummers sit in the section closer to the village (see figure 3). Again, the bush is the sphere of the dead, the village the sphere of the living. However, contrary to the dzogbedɔdɔ case, the dancing ground is here situated between the drums and the border, that is, in the domain of the living rather than of the dead. Since in both rituals paired dancers move slowly towards the border, the two variants appear as mirror images of each other. If one were to project the true dzogbe burial and its amegasi replication onto a single space, male warriors and female ghosts would be dancing face to face.

Figure 3
Figure 3: Bush funeral

This directional contrast becomes decisive at the moment of the battue (gbenyanya) sequence when the dancers, passing over the border, deploy into the bush, beating the underbrush with their weapons and cutting vegetation to violent shouts of to! to! (“get out! get out!”). The amegasis cross the border in the opposite direction when they invade the “village,” beating the spectators with branches and leaves. While their dance only hints at “chasing the bush,” their primary ritual action consists in a counterattack on the village by the bush creatures. However, such a countermovement is also discernible in the original version of the gbenyanya. After rushing through the bush, the dancers return with branches and leaves that they throw, one after the other, into the coffin until the deceased becomes in a sense covered by the bush. By contrast, the amegasi gbenyanya precedes the resurrection of the dead, who return to the world of the living, leaning on metal canes. While each of the gbenyanya variants involves a contrastive diachronic change of direction, they do so from opposite perspectives, and with a different outcome: whereas the male warriors, after invading the bush with sticks and iron weapons, surrender the corpse to the bush (which covers it with leaves and branches), the female bush creatures, after invading the village with leaves and branches, surrender their prey to the village (which guides them back with iron sticks).

The fact that in both cases the intruder (the village in the burial rite, the bush in the initiation rite) loses the body to the other party seems at odds with the popular interpretation of the dance as a fight for control of the corpses. In fact, the song that accompanies the (original) gbenyanya—“a warrior has died, take him away!” (aʋawɔtɔɖe dzo, mi tsɔ yi boo!)—suggests that the ritual is less concerned with retrieving the deceased than with chasing him permanently into the bush. This is certainly the intention of a dzogbe burial: the tomb, together with the deceased’s personal belongings, is supposed to disappear without trace in the underbrush. Restless and haunting, victims of bad deaths constitute a menace to their living relatives, and the dzogbe burial ensures that they do not come back.

The same interpretation can be advanced for the dzogbedɔdɔ: the novices’ return to the village does not signify defeat of the bush creatures but rather prolongs their intrusion into the world of the living. The novices have become creatures of the bush during the night at the bush cemetery, and when they return, they bring the bush with them—represented by Age, the lord of the bush (gbetɔ), whom they have adopted as their child. Whereas the humans, via the burial rite, try to make those afflicted by bad death disappear forever in the bush, the bush creatures, via the initiation rite, install themselves permanently in the village.

While the funerary and the initiation rite both stage the gbenyanya dance as the alternation, or confrontation, of two opposed movements—from the bush to the village and vice versa—each rite ultimately emphasizes a different direction. The “original” gaʋu is oriented to the bush, the dzogbedɔdɔ to the village. Yet the latter is not a simple inversion of the former. Contrary to the linear movement of the funerary rite, that inscribed in the initiation rite is circular: a death ritual is followed by a resurrection, and the bush creature that returns to the village is in fact a human being who was previously sent to the bush. For a complete inversion of the dzogbe burial, we need to turn to another ritual: the ritual that is performed following an animal’s violent death.

Hunting ritual (adenu)

Every hunter who has killed a large animal (such as a buffalo or warthog) has to undergo a ritual purification. This involves the construction of a permanent sanctuary called aklamakpa (lit. “aklama’s enclosure”), which is a small enclosure in which the skulls, horns, and jaws of wild animals are piled upon buried liturgical leaves. The sanctuary constitutes a vodu that is transmitted to the hunter’s agnatic descendants (comprising, in a wider sense, all the residents of the hamlet or village quarter) who make offerings and prayers to it before going to hunt, and add skulls and jaws to it after a successful trip. The purification ritual (called adenu, “hunting affair,” the generic term for all hunting rites) is still practiced in regions such as the Mono valley where deforestation has not yet put an end to the (illegal) hunting of large animals. It has, however, become relatively rare, and I have not yet had the chance to witness an adenu other than within the framework of hunters’ funerals or festivals, contexts that have increasingly replaced that of actual hunting. The following account is therefore based on the testimony of eyewitnesses.

According to my interlocutors, the ritual begins with seven days of seclusion in a “hunting hut” (adexɔ) erected in the village.9 Afterward, the hunter is washed with lustral “hunting water” (adetsi) while the aklama enclosure is constructed in his house. Great hunters bring skulls and jaws and dance around the pile, while the characteristic “hunting rhythm” (adeʋu, also called kpokpo) is played on the great talking drums and a broken metal vessel. Hunting scenes are mimed, secret hunting stories are told, and the bones are doused with palm wine and flour diluted in water. These descriptions basically correspond to those collected a century ago by Spieth (1906: 389ff.).

As indicated by the use of the broken vessel drum (which otherwise is only played at dzogbe burials), the aklamakpa construction represents an inverted gaʋu: instead of human victims of a bad death being interred in the bush (leaves covering their buried bones), animal victims of a bad death are entombed in the village (their bones covering buried leaves). True, while the gaʋu ritual is carried out to appease the victims, the adenu dances are destined to purify the killers. But these are two aspects of the same purpose. A hunter who does not undergo the ritual not only fails to make further kills. The spirit of the animal attacks and renders him mad, causing him to hurt or kill his fellow humans, whom he mistakes for animals. According to other informants, the dead animal—or its surviving spouse—transforms itself into a human being and comes to the hunter’s home to take its revenge. In a widely known Watchi tale, the vengeful animal becomes the hunter’s wife (Hamberger 2011: 580 f.). In both instances there is a confusion of roles: in the first, the hunter mistakes humans for animals; in the second, he mistakes animals for humans. This confusion of roles is an inversion of perspective. Significantly, according to Spieth (1906: 830, 850), persons possessed by Ade have “an animal’s look” until ritual purification transforms them into Ade’s priests, whom Spieth’s informants call “amegasi.”

In fact, a series of similarities links the aklamakpa to the vodu of the amegasis, including contemporary female necromancers. Both aklamakpa and the tro are devoted to those (animals or humans) who have suffered bad deaths. Both are constructed on a skull (visible or invisible), situated in an enclosure, and conceived of as a sort of “trap” for the bush creatures they “call” into the village.10 Like the future tro priestess, the owner of aklamakpa is said to dream of dancing animals, to which the skulls in his enclosure belong. In order to understand this affinity between male hunting and female divination, we need to take a closer look at the notion of aklama.

The term “aklama” is today almost exclusively used in the sense of a person’s “luck” and “good fortune,” especially in escaping from a great danger. Such luck is typical of a hunter returning from a successful hunt—the animal he killed could just as easily have killed him. Aklama is personified as a sort of guardian spirit, and in a still wider sense identified with a person’s soul or destiny (se). Surgy’s comparison of the tro with an altar of the se, represented in the shape of a head, corresponds strikingly to how some of my interlocutors interpreted the aklamakpa: according to them, aklama is a person’s spirit (se), represented by her head. By taking the animal’s skulls, and thus their spirits, to his home, the hunter keeps them from working against him in the woods. The same informants, however, subscribe to the general view that aklamakpa is a representation of the hunter’s own aklama—as if the spirit of luck represented by the skulls were simultaneously that of the killer and the victim.

This apparent ambiguity is in fact a distinctive feature of hunting. The destinies of prey and hunter are inextricably linked—good luck for one is bad luck for the other. In a sense they share the same guardian spirit, who, in protecting one, abandons the other. This is indeed the Ewe conception of Age’s role in hunting: on the one hand, he is the protector of animals and the hunter’s deadly enemy; on the other hand, he is the hunter’s indispensable ally, for only when Age abandons an animal, can it be shot. Still more explicitly, the western Ewe call the protector of animals Ade (“hunter”). This series of synonymous bush spirit names—Age, Aziza, Ade—also includes the term “Aklama.” In Spieth’s account (1906: 515, 811, 840),

Aklama is expressly identified with Ade. Less explicitly, my interlocutors define aklama as “Age’s force,” and all agree that offerings to aklama are a way of making contact with the lord of the bush. This intimate connection between the personal soul and fortune in hunting is essential to the Ewe concept of the person—a concept that includes fundamentally the link to another person in a relation of equivalence and opposition. Aklama is not only conceived of as a person’s double; it is a double person. Hence, the standard representations of Age as a pair of twins, a bisected creature, a bilocated runner.

The twofold direction of Age’s attacks makes manifest the inversion that this redoubling involves. On the one hand, he is said to kidnap the hunter and hold him captive in the termite hill, releasing him to return home with dirty clothes and long, matted hair. On the other hand, he is said to persecute the hunter, following him home and “living with” him (that is, possessing him). Both movements—from the village to the bush and vice versa—are aspects of the hunter’s displaced status: a human among animals, or an animal among humans. The construction of aklamakpa puts an end to this condition. But at the same time, it perpetuates the inversion of bush and village: as one origin story explicitly states, “aklama’s enclosure” is no other than a reproduction of Age’s home.

The aklama enclosure is not the only example of the projection of the bush into the village. The tro convent, Age’s village domicile, is another. The amegasi’s apprenticeship, as indicated by her matted hair, replicates the hunter’s captivity, and the reciprocal movements that characterize Age’s relation with the hunter (kidnapping and possession) also characterize his relation with the amegasi (initiation and divination).

This correspondence is not surprising in itself. Both hunters and necromancers act on the frontiers between the village and the bush, the living and the dead. The hunting rituals summon living animals to join their dead companions, while the necromantic sessions summon dead humans to visit their living relatives. As already noted, in several West African societies diviners are former hunters. The specific feature of Ewe amegasis is that they are women, and that the tro, unlike the aklamakpa and everything else that has to do with hunting, is transmitted through the uterine line. The almost perfect homology between hunters and amegasis makes their opposition all the more radical. Far from representing a female hunter, the amegasi is forbidden to eat wild animals, and the hunting songs refer to her as game.

In fact, the hunting rite provides direct evidence of an equation between women and wild animals. As I mentioned, I have not yet witnessed an aklamakpa construction rite. But I was able to attend a hunting ceremony organized in January 2010 by the hunters’ association of Kouvé as part of the funeral rites of a deceased comrade. The sacrificial animals had already been immolated within the aklama sanctuary when the public part of the ritual began in the courtyard of the mortuary house. It essentially consisted in dancing around a huge pile of skulls, jaws, horns, tails, and guns heaped up around a big bowl (kolo) resting on a warthog’s jaw, while dousing the pile with palm wine and flour water, drinking large quantities of palm wine from horns (a privilege reserved to buffalo killers), and performing extensive hunting mimes. It was thus largely a reduced version of what elderly hunters (and earlier ethnographers) had described, with one major exception: the participants included women.

The same song introduced the hunting mime as accompanies it in the talking drum ceremonies: egli be lã me le o vɔ—“Egli [or any other hunter’s name] says: there are no more animals,” meaning that people have no more animals to eat. In talking drum ceremonies this song continues with the line ma yi kpɔ dzogbedzi (“I am going to the bush”), accompanied by the mime of a hunter shooting the sacrificial animal. In the adenu rite at Kouvé, however, the role of prey was not played by livestock. As the song continued with the line lã kpo ye aɖu (“he will eat only animals!”), the male participants took horns and skulls from the pile and put them on the heads of the women, who, thus transformed into buffalo and antelopes, retired to the rear of the yard, which represented the bush. Somewhat later the hunters followed, miming the tracking and shooting of buffalo: after determining wind direction by allowing sand to trickle through their fingers, they crawled towards the animal, “shot” it, “cut its tail and ear” to mark it as their property, and covered the carcass with leaves to protect it against the shafts of sunlight. At the same time, the women, far from acting as frightened or enraged animals, danced in front of the hunters with smooth, seductive movements, gently balancing the horns on their heads. A smooth and gentle walk is said to be characteristic of buffalo when unaware of a hunter close by. However, the buffaloes’ dance, performed directly in front of men with guns trained on them, almost seemed to enact the popular Watchi tale of the hunter who surprised a female buffalo in the shape of a beautiful maiden.

The division of ritual space between bush and village in the adenu ritual at Kouvé (see figure 4) thus corresponds to two different representations of wild animals: as dead skulls piled up in the village, and as living women dancing in the bush—the transition being effected by placing the skulls on the women’s heads. It is noteworthy that the amegasis proceed in a like manner when putting the tro (which is said to contain a skull) on the head of a dancing woman, the newly consecrated trosi, whose movements recall those of the “buffalo.” In one sense, she resembles an inverse buffalo: instead of being killed in the bush and buried inside the aklamakpa in the village, she is “killed” in the village (before the tro convent) and buried in the bush. However, one should remember that the village is not, strictly speaking, the amegasi’s starting point. She entered the tro convent to represent another woman, the ancestress issued from the northern wilderness. This wilderness, dzogbedzi, described as a hunting ground in the pivotal song “there are no more animals,” is the origin point not only of buffalo but also of slaves, the mythical female ancestors, and, ultimately, of all women. Like slaves and wild animals, women are not buried where they are born11—as one Tchamba priestess emphasized in order to explain why dzogbe burials are the responsibility of the maternal family. One of Spieth’s informants even affirms that all women are buried in the bush, as they “have no home of their own” (Spieth 1906: 634). Viewed from this angle, the revival of the buffalo and the bush burial of the amegasis appear as two aspects of the same event, just as the novice’s dream had described. For hunters and dead warriors, the way to the bush is the way to the outland. For buffalo and women, it is the way home.

Figure 4
Figure 4: Hunting ritual


The four rituals we have discussed construct the same virtual space from different but complementary angles. These differences can be expressed in the way each ritual envisages the border crossing between the village and the bush. The funerary rite for the victims of violent death (gaʋu) centers on a movement from the village to the bush; the hunting rite (adenu) on a movement from the bush to the village; the initiation rite for female diviners (dzogbedɔdɔ) entails a movement in both directions (from the village to the bush and back); and the necromantic session (xɔyɔyɔ) evokes the same back-and-forth movement but in reverse.

To be sure, none of these rituals can be reduced to a single movement. As we have seen from the battue sequence that is common to the funerary and initiation rituals, or from the animal (buffalo or ram) dances that figure in the initiation and hunting rites, each ritual is a complex composition of to-and-fro movements, spatial subdivisions, and changes of perspective. Yet these changes of perspective are more than reiterated inversions of a single basic relation. The passage from the village to the bush (or vice versa) does not take the same form for men and women, for humans and animals, for the living and the dead. All these oppositions have a spatial expression, and the common line dividing bush and village that traverses all four rituals is only the abstract of a multiplicity of relations that generate a complex, pluridimensional space.

It is by considering several rituals as complementary views of this space that we can grasp its architecture and understand how its constituent relations are interconnected. Taking each ritual in isolation would have lead us to envisage separate hypotheses to account for the fact that women represent men (in the initiation rite), animals (in the hunting rite), and dead persons (in the divination rite): the gender ambiguity of liminal roles, the violence inherent in sexual relations, or the symbolic affinity of childbirth and violent death. Taking the rituals together, we understand how women dancing on the “bush” side of the boundary in the initiation and the hunting ritual can represent both dead men and living animals, according to whether they are considered as mirror images of dead animals (accumulated buffalo skulls or immolated rams on the “village” side), or of living men (who, in the funerary ritual, perform the same dance in the opposite direction).

The specificity of the divination rite consists in activating these transformations within the village itself, by projecting the bush into a woman’s chamber. This projection of the extreme outside into the extreme inside constitutes in a sense the divination rite’s characteristic movement, whose visible manifestation is reduced to the amegasi’s entering and leaving the chamber. At the same time, it reveals the fundamental topological feature of Ewe (and much of West African) social space: its innermost place, the female dwelling (or body), is a transformation of the bush, the place of the other. The amegasi’s sanctuary represents in concentrated form the home of all those who are not buried at home—women, slaves, victims of violent death, and wild animals—, and it is in this capacity that it serves as a passage for souls that speak through a foreign body. Clearly, this connection between the diviner and the bush is not the same as in the case of male diviners, considered as (actual or virtual) hunters. More precisely, it is the same connection, but viewed from the opposite angle. The rich ethnographic record regarding hunting and (male) divination only seems inapplicable to the case of female diviners if we hold to the hunter’s perspective. Once we add the animal’s point of view, the relational configuration characteristic of female divination and initiation becomes recognizable as a transformation of the configuration that obtains in male hunting and burial rites.

These interritual transformations provide the context of meaning for the dynamics at work within each ritual—not just in the sense of an analytical device but as an effective ritual technique. By quoting a bush burial, which in turn quotes a collective hunt, the novice’s ritual resurrection links itself to the correlative movements that specify amegasi initiation: the return of the bush dead to the village (the making of the tro), and the return of the animal victim to her bush home (the succession of the slave’s descendant to the throne of her ancestress). Each ritual engenders virtual space by evoking other rituals as latent alternative perspectives. This does not mean that the rituals examined in this article should be thought of as parts of one well-integrated superritual. A number of other Ewe rituals—such as the rite performed following the birth of twins, the talking drum ceremony, or the installation of the Tchamba vodu—could have served as points of departure to retrace the construction of the virtual space discussed here. As emphasized by Langer (1953: 84), virtual space does not exist by itself. It only exists as far as it is produced and supported by the elements to which it gives form, and there are countless ways in which it can be presented. If we can nevertheless recognize the same virtual space in many different rituals, this is because its creation is, by its very essence, transformation.


Fieldwork between 2004 and 2012 has been supported by grants from the Wenner Gren Foundation, the Agence Nationale de Recherche, and the Ecole de Hautes Etudes en Sciences Sociales. I am much obliged to Marc Chemillier, Stéphan Dugast, Gideon Freudenthal, Michael Houseman, Emmanuelle Kadya, Karen Middleton, Ismaël Moya, and the anonymous referees for their many valuable comments on earlier versions of this article. Its title renders homage to the work of the late Michel Cartry.


Albert, Bruce. 1985. “Temps du sang, temps des cendres: Représentation de la maladie, système rituel et espace politique chez les Yanomami du sud-est (Amazonie brésilienne).” PhD thesis, University of Paris X—Nanterre.

Amouzou, Koami Aziagbe. 1979. “Religion et société: Les croyances traditionnelles chez les Bè du Togo.” PhD thesis, Paris: EHESS.

Bonnet, Doris. 1982. “Corps biologique, corps social: Les Mossi de Haute-Volta.” PhD thesis, Paris: EHESS.

Bourdieu, Pierre. 1970. “La maison kabyle ou le monde renversé.” In Echanges et communications: Mélanges offerts à Claude Lévi-Strauss, edited by Jean Pouillon and Pierre Maranda, II, 739–58. La Haye and Paris: Mouton.

Brivio, Alessandra. 2007. “La mémoire de l’esclavage à travers la religion vaudou.” Conserveries mémorielles 2 (3): 18–26.

Cartry, Michel. 1979. “Du village à la brousse ou le retour de la question: A propos des Gourmantché du Gobnangu (Haute-Volta).” In La fonction symbolique: Essais d’anthropologie, edited by Michel Izard and Pierre Smith, 265–88. Paris, Gallimard.

De Heusch, Luc. 1980. “Physiology and cosmogony: Rites de passage among the Thonga.” In Explorations in African systems of thought, edited by Ivan Karp and Charles S. Bird, 27–43. Bloomington: Indiana University Press.

Désveaux, Emmanuel. 2001. Quadratura Americana: Essai d’anthropologie lévistraussienne. Geneva: Georg.

Dim Delobsom, Antoine. 1934. Les secrets des sorciers noirs. Paris: Nourry.

Empson, Rebecca, Caroline Humphrey, and Morton A. Pedersen, eds. 2007. “Inner Asian perspectivism.” Special issue, Inner Asia 9 (2).

Friedson, Steven M. 2009. Remains of ritual: Northern gods in a southern land. Chicago: University of Chicago Press.

Hamberger, Klaus. 2009. “Matrilinéarité et culte des aïeules chez les Ewé.” Journal des Africanistes 79 (1): 241–79.

———. 2011. La parenté vodou: Organisation sociale et logique symbolique en pays ouatchi. Paris: Edition de la Maison de la Science des Hommes.

———. 2012. “Traces des génies.” In La terre et le pouvoir: A la mémoire de Michel Izard, edited by Dominique Casajus and Fabio Viti, 197–214. Paris: CNRS Editions.

Houseman, Michael, and Carlo Severi. 1998. Naven or the other self: A relational approach to ritual action. Translated from the French by M. Fineberg. Leiden: Brill.

Hugh-Jones, Stephen. 1995. “Inside-out and back-to-front: The androgynous house in northwest Amazonia.” In About the house: Lévi-Strauss and beyond, edited by Janet Carsten and Stephen Hugh-Jones, 226–52. Cambridge: Cambridge University Press.

Ishii, Misho. 2013. “Playing with perspectives: Spirit possession, mimesis, and permeability in the buuta ritual in South India.” Journal of the Royal Anthropological Institute 19: 795–812.

Kapferer, Bruce. 1997. The feast of the sorcerer: Practices of consciousness and power. Chicago: University of Chicago Press.

———. 2004. “Ritual dynamics and virtual practice: Beyond representation and meaning.” Social Analysis 48 (2): 35–54.

Langer, Susanne. 1953. Feeling and form: A theory of art developed from Philosophy in a new key. New York: C. Scribner’s Sons.

Lévi-Strauss, Claude. 1971. L’homme nu. Paris: Plon.

Lovell, Nadia. 2002. Cord of blood: Possession and the making of voodoo. London: Pluto Press.

———. 2005. “The Watchi-Ewe: Histories and origins.” In A handbook of Eweland: The Ewe of Togo and Benin, edited by Benjamin Lawrance, 90–114. Accra: Woeli.

Mosko, Mark. 1985. Quadripartite structures: Categories, relations and homologies in Bush Mekeo culture. Cambridge: Cambridge University Press.

Parkin, David. 1993. “Ritual as spatial direction and bodily division.” In Understanding rituals, edited by Daniel de Coppet, 11–25. London: Routledge.

Pazzi, Roberto. 1968. L’homme evhé, aja, ge, fon et son univers (dictionnaire). Lomé.

Peek, Philip M. 1991. “African divination systems. Non-normal modes of cognition.” In African divinatory systems: Ways of knowing, edited by Philip M. Peek, 193–212. Bloomingdale: Indiana University Press.

Rosenthal, Judy. 1998. Possession, ecstasy, and law in Ewe voodoo. Charlottesville: University Press of Virginia.

Spieth, Jakob. 1906. Die Ewe-Stämme: Material zur Kunde des Ewe-Volkes in Deutsch-Togo. Berlin: D. Reimer (Ernst Vohsen).

Stépanoff, Charles. 2013. “Dessins chamaniques et espace virtuel dans le chamanisme khakasse.” Gradhiva 17: 145–68.

Strathern, Marilyn. 1998. “Social relations and the idea of externality.” In Cognition and material culture: The archaeology of symbolic storage, edited by Colin Renfrew and Chris Scarre, 135–47. Cambridge: McDonald Institute for Archaeological Research.

Surgy, Albert de. 1981. La géomancie et le culte d’Afa chez les Evhé du littoral. Paris: Publications orientalistes de France.

———. 1988. Le système religieux des Évhé. Paris: L’Harmattan.

Turner, Victor. 1969. The ritual process: Structure and anti-structure. London: Routledge & Kegan Paul.

Van Gennep, Arnold. 1909. Les rites de passage. Paris: Emile Nourry.

Viveiros de Castro, Eduardo.1998. “Cosmological deixis and Amerindian perspectivism.” The Journal of the Royal Anthropological Institute 4 (3): 469–88.

Williams, Ron G., and James W. Boyd. 1993. Ritual, art and knowledge: Aesthetic theory and Zoroastrian ritual. Columbia: University of South Carolina Press.

Du village à la brousse en quatre rites ouatchi: une analyse transformationnelle de l’espace rituel

Résumé : Les analyses anthropologiques accordent une attention croissante aux transformations de perspective comme outil de compréhension des rituels. Or, le fait que les transformations de perspective contribuent non seulement aux dynamiques internes des rituels, mais servent aussi à les relier entre eux, a reçu moins d’attention. Cet article s’appuie sur des analyses d’une série de rituels éwé (divinatoires, initiatiques, funéraires et cynégétiques) pour conceptualiser l’espace rituel comme un système de transformations de perspective qui se déploie à la fois à l’intérieur et entre des rituels. Si l’on conçoit chaque rituel comme un processus de construction d’une même architecture relationnelle d’un point de vue différent, il est ensuite possible de situer la relation entre devins féminins et chasseurs masculins dans le contexte plus large d’une série de relations interconnectées (entre hommes et femmes, humains et animaux et les vivants et les morts), qui se réalise dans l’espace virtuel de la performance rituelle. L’analyse transformationnelle d’un rituel, entendu comme une variation contrôlée de perspective, devient donc un puissant outil méthodologique qui permet en même temps d’élucider le modèle qu’utilise une société donnée pour comprendre son univers relationnel et de rendre compréhensible des faits ethnographiques qui, autrement, auraient l’air contradictoires ou inexplicables.

Klaus HAMBERGER is Associate Professor (Maître de Conférences) at the Ecole de Hautes Etudes en Sciences Sociales (Paris). He has undertaken regular fieldwork in Southern Togo since 2004 and is working on space, kinship, and ritual.

Klaus Hamberger
Laboratoire d’Anthropologie
Sociale Ecole des Hautes Etudes en Sciences Sociales
52 Rue du Cardinal Lemoine, 75005
Paris, France
Phone : +0033 (0)1 44 27 17 56


1. While Williams and Boyd apply Langer’s concept of “virtual space” directly to the analysis of ritual (Williams and Boyd 1993: 15), Kapferer’s interpretation draws equally on Deleuze’s concept of “virtuality.” The two concepts are to some extent complementary. While Langer emphasizes the discontinuity of virtual and actual space as a condition of virtualization, Deleuze conceives of virtuality as a field from which actuality is continuously produced. As applied to ritual, the first approach focuses on the construction of ritual perspectives, the second on the reshaping of quotidian perspectives through ritual. For yet another use of the concept in the analysis of ritual see Stépanoff (2013).

2. The field material used in this study was collected between 2004 and 2012. The xɔyɔyɔ, dzogbedɔdɔ, and gaʋu rituals are presented in greater detail in Hamberger 2011.

3. Amegã (literally “big person”) is a title of both ancestors and chiefs, and the particle si (literally “wife”) is the common suffix characterizing the initiated adepts of a vodu. One possible interpretation of the term draws on the fact that victims of “bad” death are collectively called “the big ones” (amegãwo).

4. The term vodu denotes both the “spirit” and the “shrine,” which do not exist as separate concepts.

5. Actually, trɔ̃ is a euphemism that in western Eweland denotes all vodu. The western Ewe identify this particular vodu by descriptive terms such as dzogbekutrɔ, “vodu of the bush dead” (Surgy 1988: 129).

6. Rosenthal (1998: 177) and Friedson (2009: 2) mention amegasis in passing; they do not figure in Lovell’s 2002 monograph. Lovell (2005: 108) remarks on female tro cult leaders but without reference to divination. Brivio (2007) describes amegasis as adepts of the Tchamba cult but not as diviners. For a review of earlier accounts see Hamberger 2009.

7. This is common in female divination: for example, the spirit helper of the Mossi kinkir-baga starts the divination session by asking, “what’s on, mum?” (Dim Delobsom 1934: 55).

8. Surgy (1988: 217), who witnessed the ritual at Vogan and Akoumape, similarly states that the novice is left alone in the bush.

9. According to some informants (as well as Surgy 1988: 147), the seclusion hut is erected in the bush.

10. Aklamakpa serves both to arrest the dead animals’ spirits and to attract their living conspecifics. As my interlocutors put it, “the heads call more heads.” In Spieth’s (1906: 390) description, flour and palm wine offerings to the skulls are accompanied by an invitation to call their fellows.

11. Ewe society is virilocal, and (at least theoretically) women are buried in their marital houses.