The Sakalava Poiesis of History: Realizing the Past Through Spirit Possession in Madagascar

by Michael Lambek
The Sakalava Poiesis of History: Realizing the Past Through Spirit Possession in Madagascar
Michael Lambek
American Ethnologist
Start Page: 
End Page: 
Select license: 
Select License

the Sakalava poiesis of history: realizing the past through spirit possession in Madagascar
MICHAEL LAMBEK-University of Toronto

Time never begins or ends; life always does.

-Northrop Frye, The Double Vision: Language and Meaning in Religion

It is a commonplace today that culture is historical, but the converse is perhaps less widely recognized. From Levi-Strauss's structural account of history (1 966) follow two emphases. One ofthese, epitomized by the work of Sahlins (1 9851, asks how new events are culturally mediated, understood, and absorbed-in other words, how history happens. This account is essentially about the ongoing cultural articulation of the present of which only in retrospect, if at all, can we say, "Ahah, this is how our history was made." The other, complementary emphasis is concerned less with how the past structures the present than with how the relation of past to present is locally formulated and understood in the present, with how the past articulates with the present to give a particular shape and form to time-a characteristic historicity or historical consciousness.' It is the latter topic that has largely engaged anthropologists of Madagascar and that I take up in turn for the northern Sakalava of B~ina.~

The comparative study of historical consciousness is still in its infancy, beset by the way in which such consciousness has been seen as the defining feature of Western civilization and hence by oversimplified binary oppositions. In order to argue that spirit possession provides the Sakalava with a kind of historical consciousness, I begin by noting that any form of history is composed; it therefore has a dimension of poiesis (from the Greek poiesis, meaning productive creation). This enables us to draw on the work of critics such as Auerbach (1968) and Bakhtin (1981) who have long moved beyond the tired and misleading dichotomies (reallmythical, objective/ideologicaI) by which Western history has imagined its distinctiveness and rational- ized itsdisciplinarity. Auerbach illuminates how realism is relativetoshiftingconventions, while Bakhtin's concept of the chronotope clarifies the articulation of different temporalities. These provide tools for specifying the Sakalava case while linking it to other forms of history. In the conclusion, I show how suffering, a critical feature of Aristotle's poetics, is no less central to Sakalava history, not only as an object of representation, but also as a means of establishing that history as both real and true.

Using a broadly Aristotelian framework I propose poetic form as a means for distinguishing historicities. Ianalyze Sakalava performances of possession by royal ancestors as the creative production of a kind of history, distinguish it from a dominant occidental model of history, and elaborate the chronotope on which it is based and the heteroglossia and historical consciousness it enables. I argue that Sakalava spirit possession has a strongly realist bent and suggest the interest of poiesis for anthropological analysis and comparison more generally. [historical production, historicity, spirit possession, mimesis, poiesis, Aristotle, Madagascar]

106 american ethnologist
Central to Sakalava experience and prevalent throughout northwest Madagascar in the 1990s are spirit-possession activities and spirit shrines. The tromba spirits are deceased members of the royal descent group that has controlled the area since about 1700 and that, broken down into smaller principalities and increasingly eclipsed by other forms of power, continues to exert a certain authority today.3 The mediums, who come from all walks of life and all manner of kin groups, form part of an otherwise clan-based division of labor through which living and dead royalty have been served and ~upported.~

Royal spirits, their live descendants, mediums, tomb guardians, shrine managers, supplicants, and the general public interact in complex ways. This article is about neither the practice of mediums nor the reproduction ofthe system (for instance, how monarchs are transformed into trombas). It is about the poetic quality of spirit performances and the manner in which they shape a historical consciousness.

The argument is based on research I conducted in the port city oiMahajanga. Although never its capital, Mahajanga sits in the heartland of the once powetful Sakalava kingdom of Boina and contains on its outskirts the most significant Sakalava shrine (doany) containing relics of four ancestral figures who stand at the apex of the range of northern Malagasy polities and who form the connecting link with other kingdoms throughout the i~land.~

When people in Mahajanga speak of tantara, of formal tradition or history, it is usually to this that they refer. And while not everyone pays respect at the shrine, everyone knows not to offend those who do. The trombas can enter anyone. The large annual festival (fanompoa beidraws pilgrims from all over Madagascar and abr~ad.~

situating Sakalava history
Mahajanga, July 9, 1994

Ndramarofaly, Lord of Many Taboo [Violationls, hunches on a low wooden stool near the altar table on the sacred northeastern side of Dady (Granny) Sakalava's front room. He clutches his spear and taps it impatiently. The tromba spirit (manifestation of a deceased member of the Sakalava nobility, ampanjaka) wears a kind of red loincloth that is drawn up to cover the breasts but leave exposed the ample shoulders and thighs of his female medium. Several other spirits are arranged in a semicircle with Ndramarofaly around the altar. One of these is Ndramandaming, comfortable in an armchair and, in further contrast to Ndramarofaly, dressed in several layers of clothing, including a man's European dress shirt with gold cufilinks and a jaunty ielt hat. Ndramandaming, presently occupying the body of a thin, elderly female medium, pours beer from a bottle into a glass, while Ndramaro, Lord Many, as his name is affectionately shortened, quaffs rum from the bottle. Abruptly shaking his spear in a threatening manner in my direction, Ndramarofaly asks the company what I'm up to. Granny Sakalava, the medium who is managing the event, attempts to calm him. Ndramandaming whispers to me that Ndramaro is airaid oi me. Ndramaro is sauvage, Ndramandaming explains, and has never before seen a "European" ivazaha).

At thetime I tookthis to bea spontaneous gestureoi kindness on the part ofNdramandamingls medium, designed to put me at my ease and explain what was going on. On later reflection I realized that in fact the remarks of both trombas fit into a larger picture that had nothing to do with me personally. Each spirit-the fierce and fearful and the worldly-played the part expected of him. In fact I, too, had become an actor or figure in the collective poiesis of history. Present in Granny Sakalava's room were spirits of different historical periods. All members of the same royal descent group, in which Ndramarofaly is a direct ancestor of Ndramandaming (FaFaMoMoFa; Figure 1), they express the distinct historical experiences that different genera- tions of Sakalava have undergone. To their largely nonnoble mediums and audience the spirits are beings to whom one otfers respect and from whom one requests blessings. In addition to their manifestly political and religious functions, however, the spirits serve as icons of history. In this context Ndramaro expressed Sakalava hegemony before the arrival of the Europeans, whereas, in his comment on Ndramaro's wildness, Ndramandaming spoke as someone who had served as governor of the province under the French colonial regime and was a Christian.

Sakalava poiesis of history 107

A Ndramisara

A = founders of the hngdom of Boina


Radama A= (Merina)


AA Ndramanefa


Ndramianta AA A Ndramandaming

AL A Ampanjaka Desy (living monarch)

Ndramanisko A

Figure 1. Simplified Sakalava royal genealogy (Bemihisatra Branch) indicating tromba spirits mentioned in the text.

His perception of Ndramaro is not to be taken as the point of view of the present any more (or less) than was Ndramaro's perception of me.

The spirits juxtapose distinct historical epochs. The juxtaposition is a part of their very constitution, for they emerge through contrasting signifiers of comportment, clothing, furniture, drink, dialect, and so on. What is still more significant, the space of performance enables the simultaneous display of successive temporalities. Sakalava history is thus additive in that, in principle, later generations do not displace earlier ones but perdure alongside them.

Moreover, and this is a central point, the conjunction of temporalities, including the present, allows each period to serve as a locus of commentary on the others. Thus the provincial governor is able to speak condescendingly about his wild (but immensely powerful) ancestor. In the resulting arena for heteroglossia, multiple voices and alternate points of view are expressed and

108 american ethnologist

made available for consideration, without being subordinated or silenced by others.' This is a condensation of historical time within the space of the present, but emphatically it does not thereby flatten or confuse historical voices.

What this means is that historical consciousness is not reducible to a single attitude, but arises through the interplay of multiple voices. It is neither single nor static, but open. It is no mere response to the present, nor is it somethingthat can be reducedto a term like resistance. Thus the anthropologist cannot state what it is, but can only describe the forms through which it is pr~duced.~

Scenes ofthis kind occur throughout the city ofMahajanga and across much ofthe northwest (Bare 1980; Sharp 1993). History is not the explicit subject of contemplation ofthese gatherings, but is rather their means or modality. In this case, the spirits had arrived in order to contribute to the amassing of money to be sent to the main shrine on the occasion of the annual ritual in which the relics of the founders of the kingdom of Boina are bathed. In its general outline this ritual fits very closely to Bloch's famous analysis of the production of sacred authority, hasina, among the Merina (Bloch 19891, in which honor and money are exchanged for blessing. Expanding from the central highlands since the early 19th century, the Merina were rivals of the Sakalava but more similar to them than most people now imagine. The spirits present in Granny Sakalava's room are at once descendants of the kings whose relics are honored and, equally, the ancestors of the living members of the descent group who receive the material benefits oithat honoring. Thus they mediate between the further past and the present and, within the present, among the living-that is, between the nobles (ampanjaka) and the people Iboeni, vahoaka). Indeed, these spirits join with the living in collecting the money to honor the senior ancestors. At the same time, Ndramaro, the most senior spirit present in the room, ended his intervention by blessing me as I crouched before him and placing white clay (tany fotsy), the material sign of having made the transaction of honor for blessing, on my palms. In maintaining their genealogical order, the spirits demonstrate the transitive, predominantly unidirectional flow of the honor for blessing transactions.

The royal ancestors are thus far from an abstraction or undifferentiated body. Like the living they are distinguished according to seniority; advisory and executive iunctions are distributed accordingly. One problem Sakalava have is that while the senior generations of ancestors have greater authority than more junior ones, they have a lesser understanding ofthe present and so their decisions are not always welcome. At such times the spirits receive gentle tutelage in how things have changed. Negotiation with the spirits enables Sakalava to work through how various generations view present issues and thus to produce historically informed responses. A suitable response is pragmatic but acknowledges the concerns of earlier generations. Today in Maha- janga the public issues in question are largely restricted to activities at the shrines, notably renovations to the tombs and buildings, and to matters of succession. Institutionally, the scope ofthe monarchy has been increasingly circumscribed, yet its past breadth remains resonant.

History, then, is intrinsically and explicitly connected to the political and the sacred. Far from representing a place to which completed battles can be consigned (White 1987:79), Sakalava history has an insistent relevance in the present. It provides the substance of sanctity and authority and serves as a vehicle for the reproduction and legitimation of hierarchy. These iunctions constrain the heteroglossia of political expression. Spirits speak from-indeed astheir respective times, yet virtually all the spirits are members of the noble stratum and descent group or are defined with respect to it. Whatever the discordance in historical experience, almost all express something about the power of royalty, a power realized at each perf~rmance.~

There is thus an assertion of totality in which the diverse voices are embedded in lineal and political continuity.

Nevertheless, we can equally turn things around and say that in all this explicitly politico- religious activity, history itself is always an implicit, emergent subject. To invoke or evoke a particular tromba spirit is always to invoke the past and to evoke a sense of both its connection

Sakalava poiesis of history 109

with and disconnection from the present. There is, as Ricoeur (1984) puts it, a "secondary reierentiality" or "figurative reference" to the "structure ottemporality" that provides events with an aura of "historicality." lo Such historicality, as manifested in part in the spirits, is the very stuff of being Sakalava. Sakalava history neither denies the evidence of discontinuity in iavor oi a reversible time, as L6vi-Strauss (1966) argued for "cold" societies, nor does it take the manifestation of irreversible discontinuity to represent underlying reality, as may be the case for "hot" societies (ci. Bloch 1996). Like any history worthy of the name, it speaks to the relationship of continuity to discontinuity; it provides "imagined continuities." " But it does so irom a rather diiierent prefiguration oithe ground than that underlying Western historiography.

The heteroglossia contributes to the often carnivalesque quality of historical performance. A few days after the incident described above, Granny Sakalava instructed me to bring to the next gathering oi Ndramaro and his descendants some mason'piso (literally, "cat's eye"). I inquired what mason' piso was. It turned out to be nothing less than an expensive greenish bottle oithat favorite drink of French colonials, pastis (anise, which turns milky when water is added). It was only Ndramaro who did not know that pastis was the real name oithe drink and who, seeing his descendants imbibing a drink (he would not touch it himselfl that did not exist in his time (or perhaps, I should say, in the time his presence had come to epitomize), scornfully called it "cat's eye." Mason' piso thereby comes to condense the discordance between the precolonial and colonial worlds and the distinction in historical consciousness established by colonial- ism-a distinction that Sakalava historical poiesis manages to make comprehensible and thereby, perhaps, to transcend.

Conversely, people outside the spirit milieu-in Mahajanga such people are numerous-may recognize pastis but not mason' piso. Implicitly, then, the appellation also signifies the distinction between those who cultivate this particular art of history and are able to enjoy a specialized wordplay (and drink), and those who try to stay at arm's length from the activities oi the trombas and mediums. Although space precludes a full elaboration oi the social context here, it should be kept in mind that Mahajanga has a heterogeneous population crosscut by divisions of class, ethnicity, political connection, religion, and knowledge and interest in spirit affairs. The severe and increasing poverty of Mahajanga does not override the cosmopolitanism of its inhabitants, including many oi those deeply involved with Sakalava history. Citizens ot Mahajanga draw on cultural resources from the Sakalava politico-religious domain; from pre-Sakalava traditions; irom Islam, Catholicism, and Protestantism; irom Comorians, South Asians, and others who live among them; from the French-influenced school curriculum and national culture; and irom theglobal culture available on television and through the experiences of family and community members who have traveled abroad. All theseelements are significant; none unifies, encapsulates, or excludes the others. The knowledge oi, and tolerance and empathy ior, diverse points of view is evident in the multiple voices oithe spirits. In Bakhtinian terms (1 981) we could say that there was a tension evident between the centrifugal pull of the multiple voices and the centripetal pull exercised by the powers that attempt to harness the spirits to their own ends.

history as poiesis
In Western usage history has a deliciously ambiguous relationship to the past. In historical inquiry we go back in time, yet history itseli marches ever iorward, leaving the past behind, in ruins. The Western sense of historical time establishes deiinitively the pastness of the past. History thus has a peculiar relationship to memory, one that may even be perceived as antithetical to it. Memory consists in clinging to remains thrust aside by history's relentless advance. To adhere to past habit is to be "out oidate." In its most sophisticated conceptualiza- tion, as transierence, (implicit, unconscious) memory is the inappropriate retention and

110 american ethnologist

activation of past dispositions in present circumstances. This tension between time and memory may not always be successfully mediated by the profession of history. The role of the historian is a Sisyphean task, forever straining to see what continues to recede, to preserve what is in the process of dissolution, and to reintegrate what has been definitively prized apart. But the historian has success of a kind, writes a book, and, in the very act of rescuing the past, assists in the processofthe reduction andobjectification of it-indiscipliningthe past and pronouncing its epitaph. Historical writing keeps the past in focus by fixing the distance of the spectator's gaze. Scholars may argue, but it is always one objectification or another that comes to stand in place of memory. Historical scholarship, through its own distanced gaze, in effect rationalizes the distantiation and alienation, the acute sense ofdiscontinuity, that it recognizes as produced by the march of history.

It is this kind of opposition between history (the dispassionate representation ofthe past) and memory (the subjective continuity with it) to which the experience from non-Western historical fields, such as that ofthe Sakalava, invites alternatives. In order to grasp this, clearly we cannot restrict our attention to the facts and narratives that Sakalava enunciate about the past, but must include all the forms and practices (although space here restricts me to spirit possession) through which historicity is realized. I call this broader creative production poiesis. Poiesis is a useful word because, in its sense of making, it comprises what in much of social thought have been separated and opposed as the material and the ideal, production and creation, ritual and narrative, the making and the made.I2

In Sakalava understanding, the past is far from escaping the present; the past calls upon the present with a literally imperious insistence, yet moves continuously in and out of focus. Historical distance is measured, yet the distances are traversable in both directions (not only the present back toward the past, but the past forward to the present) and the gaze is not unidirectional. For example, the colonial governor responds to the past but addresses his remarks forward to me in the present. The relationship with history is largely one of direct engagement, even embodiment, rather than of distanced objectification. Yet, at the same time, a discursive space opens up; moreover, the reproduction of the Sakalava past is highly formalized and is no mere subjective preoccupation with an infinity of particulars. The content concerns discrete personages and events arranged according to a chronicle.

My point is that, as a record of past events, this is history; what we are not used to is the form. Hence my emphasis is less on whatthe Sakalava historicize than on howthey historicize it-n how actors, acts, and processes are distinguished and emplotted, how the very relationship of the past to the present is constituted, legitimated, and realized. My interest here is not that of an oral historian, concerned with evaluating what Sakalava say in order to construct a more authoritative version, although I do not discount such an approach and I recognize that the Sakalava constitution of the past is itself historically situated and undoubtedly different now from what it was in the various times of and from which the Sakalava speak (see Feeley-Harnik 1984, 1991 a; Sharp 1995). My interest lies less with the content as such than with, to borrow Hayden White's helpful phrase, the content of the form (1 987).

I use the word poiesis explicitly and in preference to a currently more fashionable term, namely, mimesis. Mimesis is the term by which Plato opposes identification with the content of what one is performing or imitating to reason, that is, to rational, reflective thought. In Havelock's rendering of the opposition (19631, mimesis entails an immersion in the concrete and in the flux of endless events, whereas reason distantiates, enabling abstraction and freedom from time conditioning in order to reflect upon it. To be sure, Plato assigns all poetry and narrative to mimesis, but this depiction of artistic performance and reception as a kind of unreflective "sleepwalking" is far too extreme. In other words, while the Platonic concept of mimesis is helpful in pointing to the centrality of identification, to circumscribe the distinctive- ness ofthe sort of practice I have begun to describe by such a label is to give far too much away.

Sakalava poiesis of history 111

We need not make a radical choice between abstract reason or distanced representation and subjective identification.I3

That mimesis is not sufficient to describe possession is made apparent by the articulation of the spirits with one another; their synchronization is obviously a form of objectification and a departure from mimesis. Moreover, mimesis alone cannot account for the silences. Ndramaro's fear of me and the ensuing direct contrast with Ndramandaming neatly evades direct reference to the entire mercantile period (during which Ndramaro lived), a time in which Sakalava power was maintained by the slave trade with Yankees and Europeans (albeit largely through the mediation of lslamicized coastal people, the Antalaotra).I4

Aristotle offers a ready alternative to the pervasive and seductive Platonic dualism. He presents a trichotomy, the terms of which are not defined by direct opposition or by the mutual exclusion of each other and therefore do not lend themselves to evolutionary typologies demarcating whole epochs or modes of thought.15 History may be conceptualized with reference to all three of Aristotle's kinds of "intellect": contemplating (th~ria),practical deliberation or doing (praxis),and productive creation or making (poiesis)(Aristotle 1976[4]:2). And, while poiesis certainly does not preclude a measure of identification by artists with their subjects, such identification does not fully explain the artistry or its reception. Indeed, Aristotle begins his Poetics (1 947) by distinguishing various modes of imitation according to their means, objects, and manner.I6

Aristotle's terms are abstractions, and it is clear that in practice any one ofthem entails the others. Thus, whereas liberal academic history (or anthropology) might characterize itself in terms of theoria, and Marxism in terms ofpraxis, these are choices ofemphasis. Sakalava differs from European history most strikingly in its creative means, and it is this that I emphasize here. Yet the concepts are mutually complementary and overlapping; poiesis must always be understood in the context of the other two modes. I have alluded to the practical in my own acceptance of blessing and subsequent purchase of pastis and in the legitimating functions served by history. Whereas spirit possession does not have the aim of studying past events as such, the contemplative is evident in attempts to evaluate or reflect on performances and narratives or piece them together.

To begin with Plato's opposition means to remain confined by questions of objectivity and of truth narrowly conceived. Even were we to move beyond the dichotomy of myth versus history, the Platonic dualism limits our focus to questions such as whether the Sakalava version of their history is correct and whether the Sakalava, as historians, are sufficiently detached from their object. Alternatively, in today's current critique and inversion ofthe model, we would be asking whether, as anthropologists, we were sufficiently connected. These questions are restrictive in large part because they stay suspiciously close to the legitimations of Western historiography. Other histories are inevitably judged inferior by our standards and, no less inevitably, the underpinnings of these other histories are ignored.

Yet to part with Plato is not to surrender to relativism but rather to suggest the lines along which different histories, both within any given cultural context and between them, can be compared. It is in part White's argument that a poetic base is immanent in any historical account of the real that has emboldened me to speak about Sakalava imaginative performance as explicitly historical." White argues that history cannot be written without what he calls "tropological prefiguration of the historical field" (1973: xii) and without particular strategies of formal argument, emplotment, and ideological implication. He demonstrates the distinctive styles of particular 19th-century European historians and philosophers of history according to specific combinations of these modes. These in turn are built from limited sets of universal tropological structures and rhetorical strategies that provide means of compari~on.'~

The critical point is that recognizing the poetical dimension of history does not require conceptualizing it in terms of the mythical, the irrational, the nonempirical, or the unreal. It

112 arnerican ethnologist

serves not so much to oppose historical constructions like those ofthe Sakalava to Western ones as to provide a bridge of comparison. The emphasis is simply on the creative side of any production of history, without additional baggage. My argument is clearly not that the dominant Western and Sakalava modes of representing the past are identical, but that the flamboyance of Sakalava practice-with its puns and liquor, its royal corpses rising to life and mimetically enacting their deaths--does not discount it from being a discourse ofthe real.

I hope that rendering the centrality of poiesis in these terms enables me to evade the twin risks of ethnocentrism: the exoticization and diminution of Sakalava achievement on the one hand, and its reduction to Western terms on the other. Turning this around, I also hope that it preserves and illuminates both the complexity, sophistication, authority, and significance of Sakalava history and its difference from the kind of work we customarily describe as history.

Sakalava objective and embodied historical knowledge
It should be clear that I wish neither to reduce either Western or Sakalava history making to a single mode or dimension nor to oppose them to each other.19 Nor is to speak of poiesis to deny either the contemplative or the embodied side of Sakalava history. Regarding the former, there are many occasions when people seek out correct genealogical relations, sift among distinct versions of a story, reflect on the sources of such divergence, or simply try consciously to compose a coherent, consistent narrative. Mediums and their performances are subjea to critical evaluation. Moreover, mediums themselves are often contemplative and critical when out of trance. On many occasions they have told me that they did not know which version of a story about the past was correct and that they had tried to reason it through or seek the answer elsewhere. Often I was referred to the spirits themselves and on certain occasions (although not all) I received detailed chronicles or genealogies from them. It is of considerable interest, however, that the knowledge of history, much like its embodiment, is subject to a division of labor regarding obligations of commemoration and rights to speak. Thus many spirits and mediums can report on specific segments of history that are relevant to them, but very few have a view of the whole. Nevertheless, when compiled the pieces fit together remarkably well. Moreover, certain spirits are known to be better historians than others; thus I was referred to one spirit, Ndramanisko, who had been a student of history during his lifetime.20 Spirits and mediums are also constrained by what they have the right to speak about; more marginal people may have looser tongues but lesser authority.

Moreover, if becoming possessed is a kind of mimetic surrender to history, when the spirits rise they are understood as rational historical agents. Although rarely asked directly about the past (such questioning would be perceived as rude inquisition about their personal lives, which were not always exemplary), spirits may be called upon to discern things in the present or give direction to current enterprises.

Conversely, despite the significance of contemplative historical thought among Sakalava (or the fact that I have learned a good deal from local intellectuals), I would argue that a picture of history as abstract or theoretical knowledge is insufficient. It cannot grasp the medium (replaced in each generation) who lives permanently at an isolated shrine in the bush, bearing the tromba of the wife of one of the founding kings, who murdered her some 250 years ago. The medium signifies her burden by strenuously avoiding the town, the main shrine, and all the sites associated with the husband. When these barriers are traversed, her physical distress is pronounced. The pathos ofthis rarely seen and rarely discussed, but ever-present, figure in the Sakalava drama simply could not be so powerfully evoked through words alone. Perhaps because I have not met the medium, her image crystallizes for me the passion of an embodied poiesis of history.

Sakalava poiesis of history 113

chronotope and character: the content of the form
A distinction between the mythical and the real (or the ritual and the commonsensical) can be faulted for its binarism, for missing the diversity of forms operative in any cultural arena, and for homogenizing the mythical as internally consistent and all of a piece. It also ignores the fact that the commonsense "real" world itself is culturally constituted and to a degree contingent and variable, capable of foregrounding only certain features of the environment and connec- tions among them at the expense of other features and connections. This Boasian point is reinforced by Auerbach's remarkable analysis of the development of Western literature (1 9681, which demonstrates that realist accounts themselves come in many forms. Any representation must be selective. But the principles of selection-how widely or how narrowly the lens is focused-need not be ranked according to an external logic. We can therefore compare the "realism" of a writer like Stendhal with that of a writer like Virginia Wo~lf.~'

This is obvious as well when we shift from literature to historical writing, some of which focuses on moments of time, some on the longue duree, some on political events, some on family life, and so on. One of these forms is not necessarily more "realist" than another, although, to be sure, they all follow certain canons of historiography.

What each kind ofwriting has to offer is a particular set of stylistic conventions or constraints by means of which a time and space for the action is laid out, with a concomitant formulation of a specific "ratio" among actor, act, and scene (Burke 1945). Thus the depiction of the social milieu characteristic of the 19th-century novel or historical account retracts in later fiction like that of Virginia Woolf, or in psychohistory, as the focus narrows further upon the actor, and time and scene are virtually internalized.

It may be useful to speak of these varieties of narrative and performance, following Bakhtin (1 981 ), as providing alternative chronotopes-particular configurations of time and space, or "differing ratios of time-space projection" (Holquist 1981 :xxxiii)-that organize and emerge from particular cultural productions. Bakhtin applied the concept to literature but argued that such an "intrinsic connectedness of temporal and spatial relationships" could be found in other areas of culture as well; "the image of man is always intrinsically chronotopic" (Bakhtin 1981 :84-85). Thus, instead of making unilateral and binary distinctions between the "mythical" and the "real," we can delineate the play of chronotopes at hand.22

The chronotope connects the former lives of deceased royalty, their genealogical links to one another across time, and their actions in the present, shaping these around the places of their living and dying, as well as their tombs and the mediums, contexts, and occasions on which they are called to perform in the present and around the spaces that are inaccessible to them. To begin with, the space of action is not restricted to the original life span of the ancestor. In that spirits perform in the present, their chronotope includes contemporary action. Moreover, in stories about the past it is not always immediately apparent whether it is the live person or the tromba who was the agent, nor is the distinction always salient.

In my experience, most Sakalava historical accounts are not now explicitly articulated in the form of a lengthy chronicle, as a series of events in sequence.23 Nor are most people in the spirit milieu concerned with establishing an abstract overview of the spirits as a group. Even genealogies tend to be remembered in fragments. Each spirit must be able to name its parents and offspring and, in this way, the whole genealogy is reproduced-with occasional gaps, but always piece by piece. Nevertheless, the pieces do presuppose a chronological grid (see Feeley-Harnik 1978:411). Some nobles and intellectuals keep written genealogies, and it is commonly asserted that each shrine contains a book with a written record, but I have met only one person-a devotee of tromba activities who had traveled extensively in the region-who could recite complete sections of the royal genealogy directly from memory, and he is famous for being able to do so. This, the assembled pieces I received from so many others, and the few

114 american ethnologist

written genealogies I saw largely confirmed each other (with certain exceptions derived from either waning engagement with or conflict of interest over certain portions). The narratives concerning the lives of royalty exist likewise in fragments.

If the underlying chronology is implicit and fragmentary when it comes to recitation, the situation is different in embodied performance. The conjunction of spirits from different times is never confused; their relative chronological positions are always carefully maintained and each generation acknowledges its ancestors and recognizes its descendant^.^^ The genealogical distinctions are evident in the ways in which spirits manifest themselves and in which human supplicants approach them. Spirits of earlier generations are treated by the living with greater respect and fear. During a night's performance the most senior spirits arrive, it at all, only just before dawn; their appearance is more fleeting, and their pronouncements bear more weight. Their dialect is antiquated, making their speech harder to understand. For the mediums, the experience of active possession by genealogically senior spirits is more intense and exhausting.

Genealogical relations thus form the basis for petformative coherence, the relationship of the parts to each other (Becker 1979). Most critically, the temporal and genealogical dimensions are cast in a spatial idiom. The seating arrangement of the spirits provides an accurate map of genealogical seniority in which the northeast has precedence (see also Kus and Raharijaona 1990). Offspring sit to the west of or below their parents, and sometimes briefly on their parents' knees. In addition, junior spirits are generally more mobile than senior ones, moving around within the performance space and sometimes even outside it rather than sitting still and waiting for others to come and greet them. This spatialization of time and hierarchy (or temporalization of space) is equally evident in the microcosm of Granny Sakalava's front room and in the macrocosm of the western Malagasy landscape, where it is configured by means of a network oi shrines (doany), active noble burial grounds (mahaboi, and the movement of supplicants, tribute, and ritual artifacts among the shrines as well as in the scheduling of ceremonies at particular shrines.

The relationship of time to space is organic rather than technical here in the sense that the times and places of individual events are not mutually interchangeable in any simple sense (Bakhtin 1981 :99).25 Rather than only subsisting sequentially (by having the spirits called up successively), the times ofthe past are evoked in parallel, redistributed on a spatial grid-across the landscape, or within a room. The spatial distribution and temporal sequencing of present- day ritual events are iconic of past events and relationships. When spirits from distinct epochs arrive together, persons and events formerly separated in time are juxtaposed in the same spaces. Conversely, spatial exclusions, where two spirits cannot appear together or on each other's territory, signal either past conflicts or delicate balances of power.26 Consequently, spatial relations today make contiguous what was once temporally distinct and separate what was once contemporaneous.

To say that the chronotope is organic is also to say that it is meaningful, and specifically that it contains complex political messages. Gatherings ofthe spirits speak to the power, endurance, and cohesiveness of the royal line, whereas separations indicate tensions. Thus Ndramianta, a spirit generally found among his close kin from the colonial era in town gatherings, is conspicuously absent from larger events at which Ndramisara, the founding figure of the entire monarchy, is likely to appear, and he is excluded from the main shrine, which is Ndramisara's. Ndramianta may never come face to face with Ndramisara because, when he was alive, he refused to acknowledge Ndramisara (who lived some 200 years earlier), referring to him scorntully as Makoa. According to one version ofthe past, Ndramisara was a moas, a powerful diviner, who came from East Africa, but the epithet "Makoa" later came to connote a slave (since many slaves were taken from the Makua of present-day northern Mozambique). Ndrami- sara responded by exiling the arrogant Ndramianta. Today their conflict not only is a reminder of the ambiguity surrounding Ndramisara's origins (and hence the possibly extramonarchical

Sakalava poiesis of history 115

source of royal power) but brings forward for appraisal past attitudes to slavery and the racial expressions of the colonial-period elite. (Ndramianta is portrayed as a Muslim colonial official wearing both a European jacket and a fez.)

It is relevant that for every public performance the number and identities of the spirits to be invited is carefully planned beforehand. The roles are "cast": specific mediums are invited to perform as vehicles of particular spirits. In most cases each medium is assigned only one spirit, even if that medium also receives other spirits. This ensures that all the spirits anticipated can be present simultaneously. Each medium arrives prepared with the necessary clothing; the clothes are stacked neatly on a table near the altar until they are needed. Those in the know can thereby anticipate who will be appearing2'

Sakalava do not deny temporal succession-indeed they emphasize its poignancy-but they set succession in a plane of simultaneity. Most smaller performances include a group of relatively closely related spirits rather than those from a broad range along the genealogical span. Each spirit arrives as the royal personage was at the end of life, yet each interacts in well-defined kinship relations with others. Thus Ndramboeniarivo, as the quintessential son, always bows to his mother and may sit briefly in her lap, even though she would no longer have been alive when he was the chronological age at which he is portrayed. She spoils him and treats him with much greater affection than she does the father. This same son relates as a father to his own children and as grandfather to the rest. In this way a series of "family romances" depicts relations of love or hatred, jealousy, and conflict as eternal even though not everyone knows the stories of the original actions that underlie them.

These relations of contiguity among spirits continue to be relevant offstage and are partially reproduced among their mediums. The medium of Ndramboeni' refers to and addresses the medium of Ndramboeni's mother as "mother," and so on. This is complicated by the fact that each medium has more than one spirit. Which spirits can appear together in the same medium is consistent with what is evoked in the performances. Amediumwho has Ndramboeni's mother often has his father as well. But mother and son never possess the same medium; that would be too incestuous. Conversely, just as Ndramboeni's wife never appears at the same perfor- mance event as Ndramboeni', so she would never possess the same medium. The mediums themselves are thus understood as loci of the spirits in much the same way as houses or tombs. Such considerations constrain the marital and residential patterns of acknowledged mediums. The spirits permeate their mediums' lives, and the mediums represent them even when they are not specifically enacting them. (Thus, to the degree that the mediums' social and psychological realities are shaped by the myth models, the distinction between the mythical and the real becomes spurious.)28

The end of this general-though never complete-displacement of the syntagmatic and the diachronic onto the paradigmatic and synchronic fields is that narratives and history are broken down into individual personages, personages who evoke a particular moment in a longer sequence of events but who can represent that moment at any time. Ultimately, and with the exception of those parties who must avoid each other, all past monarchs could in principle be present together in the same room, in the present.

It will be evident then that the central "literary function" of Sakalava possession is character. The chronotope brings together a cast of characters who establish their relationships to one another by means of greetings, gifts of drink, and the like, but who generally subsist side by side in a kind of"parallel play." Often they come together on the basis of having been buried in the same location, the space of death taking precedence over genealogical proximity. Each has a past life that provides a source of identity in the present, but only specific elements of this past are relevant to present identity. These have to do with parentage, procreation, manner of dying, and place of burial, but may also include aspects of personal achievement and comportment. For example, the highly educated Ndramanisko was recommended to me as a good source of

116 american ethnologist

historical narrative (tantara), another likes to sing, and a third is remembered for a love affair. In many instances characters evoke stories, generally traumatic or poignant ones, of which they were once part. They may speak to specific political achievements, alignments, or compromises from the past. But not everyone will recognize the allusion. The narratives are condensed in the spirits' appearance and character traits. Sometimes their narrative function is reduced to a name or nickname.29

These Sakalava historical figures are not "private persons," yet they are individuals. Their story

does not, strictly speaking, unfold in biographical time. It depicts only the exceptional, utterly unusual moments oi a man's life, moments that are very short compared to the whole length oi a human liie. But these moments shape the definitive image of the man, his essence, as well as the nature of his entire subsequent life. [Bakhtin 1981:1 16, emphasis in original] 30

Time leaves a trace, but only as a result of exceptional events, determined by chance, in which the character's guilt, error, or commitment can play an initiatory role.

The trombas also evoke a sense of the human cycle; each of them represents a moment in the genealogical succession of the noble line, and they are frequently present with parents or offspring. Characters are portrayed as unchanging in their own time. In their contemporary appearance their biography changes little; the events of their lives become fixed in a kind of permanent display, except insofar as they address the concerns of those who consult them in the present and insofar as the biographical time of their mediums is intimately entwined with them. The individual completion yet cyclicity ofthe characters is reinforced by their punctual rather than durational appearance (Geertz 1973:393). Entries and exits are explicitly marked (just as the entries and exits of the living into and from shrine enclosures are marked) and largely anticipated. There are days of the week and times of the lunar month at which both spirits and shrines are inaccessible. Each spirit is unavailable on the day of the week on which the person it manifests died. This affects the scheduling of large events, since, whichever day is selected, some spirits will necessarily be absent. Similarly, the lives of the mediums are regulated in that they observe stronger dietary taboos on the death days (such as abstaining from rice, the staple food in this part of Madagascar). Spirit mediums pay close attention to the calendar and always know what day it is.

It is because action is not the main object of representation in the spirit performances that plot is not the critical function. Hence, despite its evident theatrical qualities (cf. Leiris 1980), possession differs from Western models of drama derived from Aristotle. In possession the action is secondary to the characters, helping to constitute them rather than the reverse. Nor is it character development that is at issue. Nor even is the discrimination of good from bad characters critical to their depiction; most are complex and amoral with the potential for positive and negative action. The spirits, especially the most senior, are presented as mashiaka: punitive, demanding, and fierce, eliciting fear but also respect. Ndramarofaly is remembered as among the most violent oi all the monarchs; while alive he was reputed to slit open the bellies of pregnant women in order to see how babies were made, and, by some accounts, was banished or even put to death by his kin in order to stop him. Yet he is among the most popular spirits in present-day Mahajanga, where his violence is contained and frequently displaced by solicitude for his numerousdescendants and for clients undergoing therapeutic initiation into mediumship. Or, to take another figure, that Ndramianta insulted Ndramisara is presented without a strong moral gloss.

What, then, accounts for possession's appeal or interest? Where does the tension lie? In possession it is the spectacle, or stage appearance, that for Aristotle (1947) was the least important element of drama, that takes precedence over plot. Performances are vivid and concrete. The arrival of spirits is marked by mental dissociation and physical disruption; it contorts and batters the bodies of mediums, interrupting the ordinary flow of time, but it is always particular to the spirit. Each arrival is elicited or marked by a particular praise song

Sakalava poiesis of history 117

(kolondoy).The presence of the historical character requires a changeot clothing and demeanor and is accompanied by music, tobacco, and drink. There is sharing, joking, and sometimes dancing; the application ot clammy white clay soon marked with the runnels and smell of perspiration, overlaid with large doses of perfume; fear, amusement, affection, and exhaustion. Relations between individual characters and those who witness their arrival are marked by a combination of intimacy and distance. People show respect and reverence for the trombas but also laugh with-and sometimes at-the antics of the more junior ones, sit close to them comfortably, and hold and are held by them. The spirits' lives literally interpenetrate with those of the mediums and others who attend ceremonies. Pity and fear, the emotions required by Aristotle for a good plot, are certainly aroused in the ~pectacle.~'

The most salient events of the characters' lives to be inscribed in performance are their deaths, although this is by means of a series of negations. The deaths of royalty can only be mentioned euphemistically; royalty are said to "fall," and a specialized vocabulary and set of attendant practices are available for dealing with the corpse. Yet in possession the manner of their dying is embodied; in addition, the day of the week on which it occurred and the foods they ateduring the last day are remembered by taboos impinging on their mediums (Feeley-Harnik 1978). It appears that at each occasion of active possession the characters go backward through their deaths in order to reestablish themselves as social actors. Mediums prepare to enter trance by applying whiteclay to the partsof the body whose traumata led todeath. Often they must retreat under a shroud during the transition. When the spirits emerge they dress themselves in their personal clothing and then begin to enjoy life with a vengeance, counterposing exuberant vitality and full presence to their previous absence. They partake of their favored substances: chewing tobacco, rum, cigarettes, perfume, and so on. Only when bodily established and fortified do they acknowledge thosearound them. Greeting behavior is critical as it reestablishes the spirit's social position and provides the next piece of the emerging tableau. Yet identities are no longer the same as they were in life. When spirits announce their names, the places they identify themselves as "owning" or coming from itompin) are in fact where they died, not, as in life, places of birth, of residence, or even of an ancestral burial ground. The opposition between death and life is at once implicit, understated, and yet exaggerated and highly salient.32

Once thecharacter has fully appeared its function is largely to represent itself, to acknowledge and be acknowledged as such. Trombas are characterized by the beginnings and endings of their lives. Their birth determines their place in the social hierarchy and their position in the genealogy. But it is their death that primarily determines their "moral purpose," defined by Aristotle (1947) as what characters seek or avoid. In performance time the trombas' personal histories begin always with the foundational moment oftheir deaths. This means that the features by which good Aristotelian plots are organized-where characters learn from earlier episodes of their lives, where there is moral development or suspense, or where justice prevails-are either givens, part of the background, or irrelevant. In fact, these features are displayed elsewhere. Of the three parts of the plot described by Aristotle, peripety (the change from one state to its opposite) and discovery (the change from ignorance to knowledge) are found less in the unfolding of plot over the lives of the individual characters than in the contrasts between one character, taken as a completed whole, and another (as we saw in the opening episode), and in the relationships of the spirits to the mediums and other living people in the present. A good spectacle is one in which numerous spirits rise and subsequent juxtapositions are lively. Peripety and discovery are of course also central to the practice and interests of individual supplicants to the trombas. The third element of plot identified by Aristotle, suffering, is critical because the spirits are portrayed by means of their deaths and manifestly affect the bodies of their mediums. I will have more to say about this shortly.

In sum, poiesis has proved useful for describing the distinctive way historical experience is shaped, and thus also for purposes of comparison. The centrality of character is of particular

118 american ethnologist

note, as is the manner in which characters are constructed primarily by means of synecdoche (use ofthe part to represent some quality presumed to inhere in the whole) as the refinement of the once-living person. Indeed, the "heart" of the person is to be discovered in the form ofthat individual's death, and the posthumous praise name further distills the quality. The personage is, in turn, a synecdoche for an epoch, epitomizing its central qualities. It is striking how similar all this is to what White distinguishes as the Formist type of historical argument:j3

The Formist considers an explanation to be complete when a given set of objects has been properly identified, its class, generic, and specific attributes assigned, and labels attesting to its particularity attached to it. . . . The Formist mode of explanation is to be found . . . in any historiography in which the depiction of the variety, color, and vividness of the historical field is taken as the central aim. . . . To be sure, a Formist historian may be inclined to make generalizations about the nature of the historical process as a whole, as in Carlyle's characterization of it as "the essence of innumerable biographies." But in Formist conceptions of historical explanation, the uniqueness of the different agents, agencies, and acts which make up the "events" to be explained is central to one's inquiries, not the "ground" or "scene" against which these entities arise. [White 1973:141

the past in the present
The suspense or urgency of possession cannot lie simply in the representation of a known and potent past but rather in its implications for an unfolding present. While spirits are in a synecdochic-and hence integrative-relationship to the past, they bear an ironic relationship to the present. If irony is the trope that negates on the figurative level what is affirmed at the literal one, spirits are intrinsically ironic, at least vis-a-vis the manifest bodies of their contem- porary human hosts, although an attempt is often made to conceal the irony and doubt behind illness or other corporeal symptom^.'^ Moreover, since the trombas are to a degree portrayed as live agents in the present, there is always a tension between their being bound to the limits imposed by the past narratives in which they are embedded and the possibilities for new action offered by the present. Ndramandaming's address to me is an example of such ironic intervention.

Acts in the present do not appear sufficient to undermine or transform the identity or narrative function brought forward from the past.j5 And yet, although spirits remain in character, their performances differ in one fundamental respect from the sort of historical narratives described by White (1 973) or Ricoeur (1984) in which action and plot are quickly fixed. Inthe petformative mode, characters continue to act. History is not simply past, but continuous in the present; not merely subject to additional or alternate readings, but relatively open; well known, but, at the same time, "in suspense." That Ndramandaming addressed me, or indeed any ofthe supplicants, is a new event.

A small incident further illustratesthe possibilities aswell as returning ustoa specific historical theme with which we began. The scene was a gathering at the house of one ofthe senior spirits at the shrine on the night of June 30, 1996 to acknowledge the spirit's receipt of new clothing. This itselfwas an unusual event, part ofan emerging process of history in the making. The spirit's previous outfit had mysteriously disappeared some weeks earlier, possibly in protest over planned renovations to the shrine's main building. A large number ofspirits, covering a line of descent spanning some 300 years, was called upon to witness and approve the gift. Many living people were present as well.

Senior spirits are the last to rise. Some, because of their genealogical precedence, advanced age, or illness at time of death, are seated on chairs. But there were only two chairs present and these were already occupied by Ndramandaming, the provincial governor, and his father Ndramanefa. Ndramandaming is always portrayed as immobile; his medium enters trance seated in an armchair and leaves it from the same position. Ndramaneia gave up his own chair for the senior spirits but he refused to allow the company to take Ndramandaming's. When

Sakalava poiesis of history 119

people lowered his son to the ground, Ndramanefa walked out in protest. Once outside, his medium regained consciousness and awaited the end of the ceremony sitting placidly under a tree.

This event was subsequently much discussed by other mediums. They felt that Ndramanefa had acted inappropriately; in his excessive concern for his paralyzed son, he had not shown proper respect for his ancestors. Clearly this was not part of the scenario. To understand the import of his actions it is necessary to realize that Ndramanefa is quite Merina in his dress and is a Protestant, a descendant of the union between a Merina king and Sakalava queen (although his medium at the time was a Muslim woman). His son Ndramandaming, as I have indicated, is depicted as a powerful provincial governor during the French occupation. The issue, then, is that of who carries more authority and should be granted greater respect: the earlier rulers, who manifest Sakalava autonomy and power, or the later rulers who were incorporated into the Merina and colonial states and who here also represent the westernized sector.36 This remains a live concern in present-day Sakalava politics and even on the national scene, where there is debate about decentralization. Moreover, the disregard of modernizing junior generations for the authority of their ancestors is an issue to which all Malagasy can relate. The spirit may also have been making an indirect allusion to the actions of the shrine renovators who had ignored the recommendations of the senior spirits. Ndramanefa's action thus condenses a great deal and, while speaking as the past, nevertheless manages to evoke issues live in the present. Once again, it bridges time while maintaining a sense of temporal depth.

If Ndramanefa's act was inappropriate, he must have known it would also be futile. In their domain theolder spirits have to be shown respect since they are the ultimate source of authority accruing to subsequent generations. Here, then, it is a gesture designed to say something and it carries specific ideological implications. The Malagasymestern (gasylvazaha) contrast that has been noted as highly salient by students of the Merina (Bloch 1971 ) is complicated in the northwest by the fact that the traumatic experience of the 19th century that marks the collapse of autonomy is not French colonialism so much as invasion by the Merina, whose King Radama attempted, as he famously said, "to make the sea the limits of my ricefield." Thus Antankarana history emphasizes alliance with France in opposition to the Merina (Lambek and Walsh 1997; Walsh 1998). The conflict isepitomized throughoutthe northwest by the ubiquitous Antandrano (waterdwelling) spirits, a group of trombas who are depicted as having drowned while escaping from Merina soldier^.^' Yet the line of Ndramandaming is the product of a union between Sakalava royalty and Radama himself; Ndramanefa is portrayed as virtually Merina. The living members of this branch of royalty underplay the conflict; their argument is less one of Sakalava distinctiveness from the Merina than of the pan-Malagasy quality of r~yalty.'~

In fact, however, Ndramanefa's act of supposed resistance to the senior Sakalava spirits served slyly to emphasize the latter's continued superiority to the Merina interlopers. Thus it was really a pointed message to the living ruler about the importance of maintaining Sakalava tradition and an expression of resistance not to the dead but to the living. The story also illustrates the possible refinements of character deployment; Ndramanefa acted toward an end that was not strictly his.

Despite such disruptions, Sakalava spirit performances approximate what White calls the comic mode of emplotment, in which "hope is held out for temporary triumph of man over his world by the prospect of occasional reconciliations of the forces at play" (White 1 973:9).3' These are symbolized in festive occasions, which for Sakalava are literal enactments of festivity such as the main pilgrimage or the collective receipt of the clothing. At such events the drowned spirits themselves are among the gayest participants. Yet always in counterpoint to the comedy or harmony of ritual are the tragedies of the individual narratives, summarized in the modes of dying of the protagonists. Such comic resolution is implicitly conservative, "inasmuch as one can legitimately conclude from a history thus construed that one inhabits the best of possible historical worlds, or at least the best that one can 'realistically' hope for, given the nature of the historical process" (White 1973:28).40

120 american ethnologist

The attribution of conservatism, here in the face of actual change, is reinforced when we consider how commoners must acquiesce to noble history. The perdurance of royalty is counterposed to the mortality of commoners, who must serve the trombas or become ill. What is apparent is less a "parallelism" of two distinct "worlds" than their impingement and mutual dependence. Royal perdurance used to be predicated quite literally upon the mortality of subjects; in the past the "falling" of the monarch was eased by a human "mat" or sacrifice. Royals perdure, yet they perdure in the act of their dying. Nonroyals die, yet they serve the royals in the course of their lives and, in extremis, by meansof death. In thedistinctions between kinds of people, Sakalava experience and play out throughout their lives those oppositions between permanence and impermanence, ancestors and living, that other Malagasy enact at funerals or reb~rials.~'

Many societies commemorate the past. Often, however, this is the reenactment of a fixed moment in the past, a prototype that is itself ahistorical. What makes the Sakalava case so interesting is that they do not simply commemorate a past event or even a past time, but the thickness and the passing of their history; what they repeat is not just a founding moment but a series of temporal vignettes and exemplary figures that, juxtaposed, evoke the passage of history. History is understood not in the senseof a static, timeless "past," nor in that of a simple repetitive endless cycling, but as the series of steps from then to now, always in complex relation to the uniqueness of persons and the irreversibility of death.

Yet there is less an effective displacement of one time or person by another than their cumulation and mutual resonance. Unlike the dominant mode in the West, where history recedes and the historian goes back, the Sakalava past is carried forward. The heroic figures are not the historians but the trombas. The diverse, consecutive periods of the past are brought together with each other-and with the present-so that each epoch speaks through allusion to the others. This cumulation is less a confrontation than a reconciliation; representatives of thedifferent epochs do not directly challenge each other so much as mutually address the living. It is their copresence in the present that invites interest and provides a lively sense of historical passage and transformation. If the past lends its authority to the present, the mutual presence of multiple generations of historical personages suggests ultimately an acquiescence in history. Might we say also that it is a kind of "working through"? 42

Alternatively, the inclusiveness and exuberance of Sakalava representations today may be compared to those of the novel. Despite its hierarchy, possession does not presume a supreme authority. Each voice is self-assertive, undercuts those of the others, ironizes them. This seems akin to Bakhtin's depiction of "the novel's joyous awareness of the inadequacies of its own language. . . . aware of the impossibility of full meaning, presence, it is free to exploit such a lack for its own hybridizing purposes" (Holquist 1981:xxxiii).

epilogue: truth and suffering
Poiesis and history may seem to represent antithetical claims and the reader may have begun to feel that, in emphasizing form, I have evaded questions of reference. After all, modernist poetry derives its meaning intrinsically, whereas history is supposed to derive its meaning extrinsically. Yet the whole point has been that poiesis must be discernible in every genre of cultural produ~tion.~~

One model would distinguish scientific truth, which is discovered by a process of knowing, from poetic truth, which is created by a process of making. These are known respectively as the correspondence and the coherence theories of truth. As Grant puts it, while the former defers tothe fact and attemptsto render it with fidelity, "in thecoherencetheory . . .theepistemological process is accelerated or elided by intuitive perception. Truth is not earned by the labour of documentation and analysis but coined, a ready synthesis, and made current-as is any

Sakalava poiesis of history 121

currency-by confidence, 'the confidence of truth.' Evidence is replaced by self-evidence." He continues, "In the first case the truth is true to something, in the second it is true as a line or edge is said to be true when it is straight, flawless-containingthe truth, not simply representing or alluding to it" (Grant 1970:9).

The Sakalava would appreciate the metaphor. Indeed, whole coins signify and enable royal authority and speech. In order to become trombas royalty must be buried with coins in their mouths and before a tromba can first speak truly it must rinse its mouth and spit out the coin. Silver coins are immersed in the mixture of white clay and water that trombas use to provide blessings. Coins also signify the moon, the lunar cycle being an idiom of life and death like the spirits and regulating access to them. But of course there is no reason for Malagasy practices to be bound by the limitations of Western genres and dichotomies. Sakalava possession maintains a productive tension between the literal and the allegorical in a way that more highly rationalized genres do not.44 In fact, there is evidence that for Sakalava representations of the past need to meet both correspondence and coherence criteria of truth value (cf. White 1987:40).

In 1994 1 had the idea that each spirit or its senior mediums would be the authoritative repositories of their stories. I had heard from someone else excerpts from the story of the king whose tromba possessed a senior medium named Salim and I hoped to gain from Salim a more complete account. Who would know better, I reasoned, than the medium? Moreover, I felt that Salim's account would complement those of the mediums of other spirits who took their own points of view on the past. My assumption proved to be only partly correct. I spent much time cultivating Salim as well as a medium of one of the main female trombas, hoping to speak to their respective spirits and gain permission to learn more about their lives. Salim proved to be a wonderful informant on related matters and demonstrated his support of my work by putting me in contact with other experts. But my attempts to elicit stories about their spirits from the mediums or to meet these particular spirits in person were always put off; either the medium was too busy, or it was a taboo day for the spirit, or some necessary ingredient was missing.

I began to assume the mediums would not speak because they simply did not know the stories or were afraid of having the limits of their knowledge exposed. My hostess denied this, asserting that the mediums were afraid they would suffer if they spoke. So, when her schoolteacher husband asked how I was getting on, I explained that I had finally realized that the mediums were precisely the people who could not suffer their stories and that I was embarrassed for having put them in an awkward position. Nonsense, he replied, they just do not know the facts.

Each of these explanations is correct in its own way, and each is interesting for what it tells us about the clash between Western and Sakalava assumptions about historical knowledge. First, it is evident that there are limits to the mediums' knowledge. They do not know all the details of their spirits' lives, nor could the trombas speak as much about themselves as could an ordinary living person. But it is only in a Western conception of history as the amassing of infinite and precise detail that this becomes relevant. In the Sakalava view it is sufficient that the spirits come from the past and that they evoke and particularize it. The spirits bear certain images of the past, but most people are not explicitly interested in the past "for its own sake," or at least were not interested in such abstractions until the rise of historicist tendencies and spectatorial distance produced by national schooling (and possibly anthropologist^).^^

The other explanation brings to the fore what underlies both positions, namely, that the relationship to history is not one of distanced objectivity. For spirits whose lives were associated with excessive acts of violence, their mediums are precisely the people who cannot tell their stories, because to tell them would be somehow to suffer them. The sign of the truth of the story is not its retelling but its silence or the punishment that accompanies narration. Senior trombas and their stories are weighty and difficult (raha sarotro) and known to be so even in the absence

122 arnerican ethnologist

of the details that might demonstrate the reasons for it. The difficulty of speaking about the past is a salient index of its significance.

it is noteworthy that the mediums maintain silence precisely about their own spirits. Their knowledge is intimate, potent, and embodied and therefore not ready, necessary, or appropriate for dissemination. It is literally too painful. If the mediums do not reproduce the past primarily in discursive form, then what is their function? of what are they the vehicle? That they suffer when they speak the secrets of their spirits exemplifies and reinforces the fact that the "history" relevant to them is the trauma, not the stories about it. They do not "represent" this suffering, they suffer it. Suffering is a primary means of embodiment as well, perhaps, as the true subject of history. In any case, it is the presence of the spirits-the presence of the past-rather than objective knowledge about the spirits or about the past that is critical. This presence, even in suffering, even in silence, is pregnant with meaning.

The medium's situation highlights the burden of history as well as the organic division of historical labor. Salim lives a life of which the continuous (rekasting of a specific king forms an intrinsic rather than extrinsic part, one that Salim can never afford to forget because the clay of his (relcreation is nothing less than his own flesh. In addition to his having been an honored and powerful king, son of the founder of the kingdom, and ancestor of all successive rulers- aspects of his identity that are widely known by the general public and contribute to the tremendous respect he enjoys-Salim's tromba is also remembered by those in the spirit milieu for having brutally murdered his wife in a jealous rage. He is the husband of the woman whose medium lives in virtual isolation in the bush and mimetically reproduces the effects and consequences of her murder.

This is evocation of a myth, to be sure. But what could be more real--or more realist a form of portrayal-than suffering?

Acknowledgments. I thank Karen Middleton, Jackie Solway, and Emmanuel Djacoba Tehindrazanarivelo ior close readings of a iirst draft; Maurice Bloch, Fenella Cannell, Gillian Feeley-Harnik, David Graeber, Wendy lames, and Shirin Sotoudeh ior helpful comments; and Michael Herzield and lane Huber for exceptionally enthusiastic and acute editorial advice. I delivered a shortened version as a Munro Lecture, Universityof Edinburgh, on February 13,1997. Other drafts have beneiited from critical appraisal by seminar audiences at the Satterthwaite Colloquium on Airican Ritual and Religion and the University oi Uppsala (1 995); the Universities of Bern, Leuven, and Toronto (1 996); Oxiord; and the London School of Economics (1 997). Research conducted in Mahajanga for iour to six weeks per year irom 1993-96 has been supported by agrant irom the Social Sciences and Humanities Research Council oiCanada. I am indebted to the people in Mahajanga who welcomed me so hospitably, to Elie Rajaonarison and lean-Aime Rakotoarisoa ior facilitating my visits, and especially to Ernmanuel Tehindrazanarivelo ior inspiring the study. I owe my rapid entree into the community to his teaching and introductions (see Lambek 1997b ior an account oi our work together). This article is dedicated to him. Any iailings are entirely my own responsibility.

My usage here has a slight ditierence in emphasis from that of the Comarofis, less implicitly critical than constitutive oireality in its temporal dimension. Yet it shares with theirs an attention to poetic form and a recognition oi "the implicit language oi symbolic activity" (1992:157). Like them, I am interested in developing an "ethnography of the historical imagination" (1 992:31). Imagination may be a more appro- priate word than consciousness here since I do not wish to imply a measure oi their (Sakalava) recognition of what we (Western, leitist intellectuals) take to be the reality oi their situation.

Bloch (1986) and Feeley-Harnik (1991a) are notable ior the way they weave various approaches together. Bare (1 9801, Feeley-Harnik (1 991 b), and Sharp (1 995) describe historical consciousness in adjacent northern Sakalava principalities. The material and argument presented here are most directly linked to Feeley-Harnik's notable essay, "Divine Kingshipand the Meaning oi History among the Sakalava" (1 978).

1 here depart from the transcription itrumba) I have used in my work on Mayotte and accede to standard Malagasy orthography despite the fact that it is not best suited to the northern dialects. The first vowel in tromba is pronounced identically in Mayotte as in Mahajanga, somewhere between the sound in "look" and "truth."

By clans I refer to named groups whose recruitment is by means oi bilateral descent. Not all such groups posit an apical ancestor. Members constitute, in Astuti's useiul terminology (1 995), a speciiic "kind"

Sakalava poiesis of history 123

of people ikarazanai rather than necessarily each other's kin. On the division oi labor see Feeley-Harnik 1978, 1982, 1984, 1991 a, and 1991 b.

The relics are a fundamental source of sacred and material power; control over them has direct ramifications in the traditional sphere (Feeley-Harnik 1991 a) and has been theobject ofconsiderabledispute between royal factions ior most oi this century. The relics are a force in contemporary local, regional, and, some would argue, national politics.

Contributions from Malagasy emigres presently comprise a significant proportion of the funds that support shrines and mediums. I use the words shrine and pilgrimage in place of local terms, not because they are the precise equivalents, but in order to grasp the religious signiiicance of Sakalava activities. My own perception began to change in subtle ways once I started to think in these terms. I have considered calling the main shrine in Mahajanga a temple. Why should the term temple be restricted to Asia, as it so often is? In fact, the main activities at the shrine-the bringing of offerings and prayers in return for blessings, and the annual ceremony in which the relics are bathed and circumambulate the shrine-have parallels in Hindu practice. The ditierences between the two modes are also wide enough that I am not suggesting any historical connections; I am instead noting how the choice oi language affects the kinds oi implicit comparisons we make and hence the terms of reference with which we view our subject.

The term is Bakhtin's (1 9811, and is well applied to spirit possession by Besnier (1996). On the "coincidence" of temporalities see also the profound analysis by Becker (1 979).

Compare the Comaroffs: "consciousness is best understood as the active process-sometimes implicit, sometimes explicit-in which human actors deploy historically salient cultural categories to construct their self-awareness" (1 992:176).

There are some exceptions: for example, the spirits of drowned Ste. Marien sailors. I describe the creative interventions of a sailor spirit speaking about AIDS in Lambek 1997c.

This is a paraphrase from Hayden White's essay (1 987:l 72) on Ricoeur's Time and Narrative(l984).

The term is of course borrowed from Anderson (1 991 ).

Poiesis clearly encompasses both written and nonwritten cultural forms; it is broader than "text" and thereby overcomes some of the insufiiciencies that others have seen in the text model. (These insufficiencies arise in large part because they have mistaken the idea oi iixed texts for textual production.) Herzfeld (1 997:144 iO invokes a "social poetics" ior reasons broadly similar to mine.

None of this is to disparage the marvelous account oi mimesis provided by Taussig (19931, whose emphasis on sensuousness and the tactile as opposed to the distancing property of vision supplements the usual accounts oi representation and unbalances rather than reinforces standard Western notions of alterity.

In fact, Ndramaro's fear and savagery is based not only on when he lived but on how he lived; he was a rustic who spent much oi his time hunting in the forest with only hisdogs for company. His "wildness" is thusoverdetermined. Yet noneof the spirits whom I have seen noted the link with the foreign slave traders.

Levi-Strauss's "science of the concrete" (1966) provides a notable attempt to mediate Plato's dichotomy and is thereby an instance of the very process Levi-Strauss takes to be central to human thought.

Aristotle's mimesis (1 947) is a general term for the representational aim of aesthetics, not opposed to reason. I believe that questions oi identiiication might be better discussed under his category oi moral knowledge (see also Lambek 1993, 1997a, 1997b).

White (1 973) is concerned with narrative understood in the conventional sense as writing; it seems a small step to include various kinds of periormances within his concept of discourse.

White (1 973) draws on the typologies of broadly Aristotelian critics such as Burke (1 945, 1969), Frye (1 957), and Pepper (1 942).

It is one of the virtues of White's structuralism (1973) that it enables us to distinguish dirierent styles or emphases within a given cultural tradition. And we could add cultural forms that come somewhat closer to Sakalava possession, such as the presentation of history at tourist sites like Edinburgh Castle or, as both Wendy Jamesand Maggie Sackville-Hunt have suggested to me, Shakespearean representations oiantiquity composed in Elizabethan idiom and transformed to a 19th-century setting for a 20th-century audience. There is every reason to expect variation among the approaches to history oi Malagasy "experts" as well.

20. It is recognized that he does not ofier his knowledge equally through all his mediums.

Although he covers the whole course oi Western literature, Auerbach (1 968) would single out the 19th-century French novel's depiction of the social milieu as a kind of apogee.

Bakhtin's concept of chronotope must be supplemented by his insistence on heteroglossia and multiplicity; any given social group will have multiple chronotopes at its disposal, issuing productive challenges to one another.

Clearly the chronotope found in possession no longer bears the same relationship todominant social institutions in Madagascar as it did in earlier periods. The chronotopes themselves have undoubtedly also changed, but their history is not one I can address.

The situation thereby supports Connerton's argument (1 989) concerning the significance of embod- ied and performative memory.

This is in contrast, for example, to the ancient Greek adventure story, which is a chain of random contingencies in which "the nature oi a given place does not figure as a component in the event" (Bakhtin 1981:100), and in which spaces are interchangeable just as the individual escapades could be rearranged in a difierent temporal order. These distinctions are prefigured in Aristotle's comparison of tragedy and epic poetry (1 947:sect. 1449b).

26. Present-day conilicts play their own part in ritual exclusions.

124 american ethnologist

All this is quite different irom Mayotte (Lambek 1981 ), where the arrival of particular spirits is much more haphazard and spirits routinely move between mediums during a single periormance. The ability to ensure the presence of particular spirits lies in the iact that Mahajanga is a large city with a relatively high number oi mediums to draw irom.

Moreover, the material described in this paragraph serves as a reminder of the artificiality of any firm boundary between poiesis and practice.

Space prohibits discussion of these topics here. Posthumous praise names summarize some striking aspects of deceased monarchs' characters or events in their lives. Often the meanings of the names or the stories that lie behind them (that is, the reasons for their selection) are forgotten. Nicknames tend to single out a kinship or age status that epitomizes the deceased-mother's brother, small child, and so on.

The quotation concerns the characters found in Apuleius 1962. As Geertz observed (19731, temporalization and personalization are closely linked. Bakhtin (1981) refines the point.

Aristotle (1947) distinguished among six parts in every tragedy: plot, character, diction, thought, spectacle, and melody. He acknowledged that pity and fear could be aroused by the spectacle, but ielt that the superior way was by means of the plot.

The juxtaposition of death with life and the striking nexus of vitality, boisterous exuberance, and power are found in other Malagasy societies, notably in contexts of funerals, reburials, and circumcisions. Their relationships have exercised a number of interpreters (Astuti 1995; Bloch 1982, 1986; Graeber 1995; Huntington 1973; and Middleton 1995a, 199513); unfortunately this direction cannot be pursued here.

Drawing on the root metaphors of Pepper (1 9421, White proposes iour alternate forms of historical argument: formist, organic, mechanistic, and contextualist (White 1973:l 3).

In past work I have been concerned with the ironic qualities oi possession, hence the title Human Spirits (Lambek 1981). But as Sakalava are saying something positive about the past and about society, I now emphasize synecdoche. Irony is negational whereas synecdoche is integrational (White 1973:34).

35. New experiences are incorporated by the spirits oi successive generations.

The portrait of Ndramandaming as the precedence-demanding governor suggests Taussig's argu- ments about the colonial mirror and that "what's being mimicked is mimickry itself" (1993:241; ci. Kramer 1993; Stoller 1995).

Yet these spirits may actually derive from an earlier event that has since been collapsed with the Merina invasion (Walsh 1998). Ironically, this may have taken place between two branches of Sakalava, the iormer aggressors subsequently appropriating the role of victim (Mellis 1938; Emmanuel Tehin- drazanarivelo, personal communication, 1996).

38. The branch is Sakalava Bemihisatra, identified with the royal tombs at Betsioko.

White distinguishes among four modes oi emplotment as set out by Frye (1 957): romance, comedy, tragedy, and satire.

It is worth quoting the preceding remarks on the comic mode asepitomized in the historiographical style of Ranke:

Those"forms" which Rankediscerned in the historical iield werethoughttoexist in the kind oi harmonious condition which conventionally appears at the end of a Comedy. The reader is left to contemplate the coherence oi the historical iield, considered as a completed structure of "Ideas" (i.e., institutions and values), and with the kind oifeeling engendered in the audience of a drama that has achieved a deiinitive Comic resolution oi all the apparently tragic conflicts within it. The tone of voice is accommodationist, the mood is optimistic, and the ideological implications are Conservative. [White 1973:28]
The space that brings the nobles of various eras together is the common space oi their dying; as in Bloch's argument (1 9861, death overcomes time. See also note 32.

Similarly, it fits a Ricoeurian interpretation, having "as its 'ultimate referent' nothing other than temporality itself" (White 1987:52).

Aristotle himself distinguished history from poetry with characteristic succinctness: "one describes the thing that has been and the other the kind of thing that might be" (1 947:sect. 1451 b).

44. For a profound reading oi possession as allegory see Boddy 1989.

45. This is not to deny the presence oiskepticism (cf. Lambek 1993) or the iact that the authority oi the spirits in key mediums is tested (Feeley-Harnik 1978:412).

references cited
Anderson, Benedict R. O'G. 1991 Imagined Communities: Reilections on the Origin and Spread of Nationalism. Revised edition. London: Verso. Apuleius 1962 The Golden Ass. Harry C. Schnur, ed. William Adlington, trans. New York: Collier.

Aristotle 1947 Poetics. Richard McKeon, ed. New York: Modern Library. 1976 The Ethics oi Aristotle: The Nichomachean Ethics. Hugh Tredennick, ed. j. A. K. Thomson, trans.

Harmondsworth, UK: Penguin.

Sakalava poiesis of history 125

Astuti, Rita 1995 People of the Sea: Identity and Descent among the Vezo oi Madagascar. Cambridge: Cambridge University Press. Auerbach, Erich 1968 Mimesis: The Representation of Reality in Western Literature. Willard R. Trask, trans. Princeton, NJ: Princeton University Press. Bakhtin, Mikhail M. 1981 The Dialogic Imagination. Michael Holquist, ed. Caryl Emerson and Michael Holquist, trans. Austin: University oi Texas Press. Bare, J-F. 1980 Sable rouge: Une monarchie du nord-ouest malgache dans I'historie. Paris: L'Harmattan. Becker, A. L.

1979 Text-Building, Epistemology, and Aesthetics in Javanese Shadow Theatre. In The Imagination oi Reality: Essays in Southeast Asian Coherence Systems. A. L. Becker and Aram A. Yengoyan, eds. Pp. 21 1-243. Norwood, NJ: Ablex.

Besnier, Niko 1996 Heteroglossic Discourses on Nukulaelae Spirits. In Spirits in Culture, History, and Mind. Jeannette Marie Mageo and Alan Howard, eds. Pp. 75-97. New York: Routledge. Bloch, Maurice 1971 Placing the Dead: Tombs, Ancestral Villages, and Kinship Organization in Madagascar. London: Seminar Press. 1982 Death, Women, and Power. In Death and the Regeneration of Life. Maurice Bloch and Jonathan Parry, eds. Pp. 21 1-230. Cambridge: Cambridge University Press. 1986 From Blessing to Violence: History and Ideology in the Circumcision Ritual oi the Merina oi Madagascar. Cambridge: Cambridge University Press. 1989[19771 The Disconnection between Rank and Power as a Process. In Ritual, History, and Power: Selected Papers in Anthropology. Pp. 46-88. London: Athlone Press. 1996 Internal and External Memory: Different Ways of Being in History. In Tense Past: Cultural Essays in Trauma and Memory. Paul Antze and Michael Lambek, eds. Pp. 21 5-233. New York: Routledge. Boddy, Janice 1989 Wombs and Alien Spirits. Madison: University oi Wisconsin Press.

Burke, Kenneth 1945 A Grammar oi Motives. New York: Prentice-Hall. 1969 A Rhetoric of Motives. Berkeley: University oi Caliiornia Press.

Comaroff, John, and Jean Comarofi 1992 Ethnography and the Historical Imagination. Boulder, CO: Westview. Connerton, Paul 1989 How Societies Remember. Cambridge: Cambridge University Press. Feeley-Harnik, Gillian 1978 Divine Kingship and the Meaning oi History among the Sakalava (Madagascar). Man (n.s.1

13:402417. 1982 The King's Men in Madagascar: Slavery, Citizenship, and Sakalava Monarchy. Airica 52(2):31-49. 1984 The Political Economy of Death: Communication and Change in Malagasy Colonial History.

American Ethnologist 11 :1-19. 1991 a A Green Estate. Washington, DC: Smithsonian Institution Press. 1991 b Finding Memories in Madagascar. In Images oi Memory: On Remembering and Representation.

Susanne Kiichler and Walter Melion,eds. Pp. 121 -1 40. Washington, DC: Smithsonian Institution Press.

Frye, Northrop 1957 Anatomy oi Criticism. Princeton, NJ: Princeton University Press. 1991 The Double Vision: Language and Meaning in Religion. Toronto: University of Toronto Press.

Geertz, Clifiord 1973 Person, Time, and Conduct in Bali. In The Interpretation of Cultures. New York: Basic Books. Graeber, David 1995 Dancing with Corpses Reconsidered: An Interpretation oi Famadihana [in Arivonimamo, Mada- gascar). American Ethnologist 22:258-278. Grant, Damian 1970 Realism. London: Methuen. Havelock, Eric A. 1963 Preface to Plato. Oxiord: Basil Blackwell. Herzfeld, Michael 1997 Cultural Intimacy: Social Poetics in the Nation-State. New York: Routledge. Holquist, Michael 1981 Introduction. In The Dialogic Imagination. Austin: University oi Texas Press. Huntington, Richard 1973 Death and the Social Order: Bara Funeral Customs. African Studies 32:65-84. Kramer, Fritz 1993 The Red Fez: Art and Spirit Possession in Airica. M.R. Green, trans. London: Verso.

126 american ethnologist

Kus, Susan, and Victor Raharijaona

1990 Domestic Space and the Tenacity of Tradition among Some Betsileo oi Madagascar. In Domestic Architecture and the Use of Space: An Interdisciplinary, Cross-Cultural Study. Susan Kent, ed. Pp. 21-33. Cambridge: Cambridge University Press.

Lambek, Michael 1981 Human Spirits: A Cultural Account of Trance in Mayotte. New York: Cambridge University Press. 1993 Knowledge and Practice in Mayotte: Local Discourses of Islam, Sorcery, and Spirit Possession.

Toronto: University of Toronto Press. 1997a Knowledge and Practice in Mayotte: An Overview. Cultural Dynamics 9(2):131-148. 1997b Pinching the Crocodile's Tongue: Affinity and the Anxieties of Influence in Fieldwork. In

Fieldwork Revisited: Changing Contexts of Ethnographic Practice in the Era of Globalization. Theme issue. Joel Robbins and Sandra Bamiord, eds. Anthropology and Humanism 22(1):31-53.

1997c Body and Mind in Mind, Body and Mind in Body. In Bodies and Persons: Comparative Perspectives from Africa and Melanesia. Andrew Strathern and Michael Lambek, eds. Pp. 103-1 23. Cambridge: Cambridge University Press.

Lambek, Michael, and Andrew Walsh 1997 The Imagined Community of the Antankarafia: Identity, History and Ritual in Madagascar. In Special Madagascar issue. Karen Middleton, ed. Journal of Religion in Africa 27(3):308-333. Leiris, Michel 1980[1958] La possession et ses aspects theitraux chez les Ethiopiens de Gondar. Paris: Le Sycamore. Levi-Strauss, Claude 1966 The Savage Mind. Chicago: University oi Chicago Press. Mellis, J.V. 1938 Volamena et Volafotsy. Tananarive, Madagascar: Pitot de la Beaujardiere. Middleton, Karen

1995a Tombs, Umbilical Cords, and the Syllable Fo. In Cultures oi Madagascar: Ebb and Flow of Influences. Working Papers Series 2. Sandra Evers and Marc Spindler, eds. Pp. 223-235. Leiden: International Institute ior Asian Studies.

199513 Tomb-Work, Body-Work: Body as Leib in Karembola Birth and Death Ritual. Paper presented to the 1995 Satterthwaite Colloquium on Airican Ritual and Religion, Satterthwaite, United Kingdom, April 22-25.

Pepper, Stephen 1942 World Hypotheses. Berkeley: University oi California Press. Ricoeur, Paul 1984 Time and Narrative, 1. Kathleen McLaughlin and David Pellauer, trans. Chicago: University of Chicago Press. Sahlins, Marshall 1985 Islands of History. Chicago: University of Chicago Press. Sharp, Lesley 1993 The Possessed and the Dispossessed: Spirits, Identity and Power in a Madagascar Migrant Town. Berkeley: University oi Caliiornia Press. 1995 Playboy Princely Spirits of Madagascar. Anthropological Quarterly 68(2):75-88. Stoller, Paul 1995 Embodying Colonial Memories. New York: Routledge. Taussig, Michael 1993 Mimesis and Alterity: A Particular History of the Senses. New York: Routledge. Walsh, Andrew 1998 Constructing "Antankarafia": History, Ritual, and ldentity in Northern Madagascar. Ph.D. disser- tation, University of Toronto. White, Hayden 1973 Metahistory: The Historical Imagination in 19th-Century Europe. Baltimore, MD: Johns Hopkins University Press. 1987 Thecontent oithe Form: Narrative Discourseand Historical Representation. Baltimore, MD: Johns Hopkins University Press.

submitted September 13, 1996 accepted December 25, 1996

Sakalava poiesis of history 127

  • Recommend Us