Speech, silence, and slave descent in highland Madagascar


  • Correction Note: This article was first published online on the 7th of August 2013, under a subscription publication licence. The article has since been made OnlineOpen, and the copyright line and licence statement was therefore updated in August 2014.


This article is an examination of the uses and effects of words and silence. It analyses the rhetorical strategies used in connection with a fundamental cleavage in highland Malagasy society: the distinction between people of free and slave descent. A pervasive silence hangs over this topic since it is almost never mentioned between the two groups. This silence, along with the careful words used to play down status differentiation, forms the rhetorical micro-politics of village life. The article takes the view that this wholesale avoidance constitutes a generalized speech act: that is to say, it is constituted of diverse motivations and strategies, and has multiple and contradictory effects. One of these is that while allowing a liveable fiction of equality to be evoked, these rhetorical strategies also entrench the division even more deeply.

Ny vava tsy ambina no ahitan-doza.

An unguarded mouth spells danger.

Malagasy proverb

It takes less than two minutes to walk from the main village of Antanety, westwards across groundnut and manioc fields, to the shabby little group of mud and thatch houses called Tananomby. The distance is meaningful, for the residents of Tananomby are descendants of slaves while the residents of Antanety are not.

The placement of slave-descent hamlets to the west of the main village is common throughout the Malagasy highlands, and the western location of settlements such as Tananomby is a topographical fact with symbolic resonance. This resonance comes from the cosmological significance invested in the cardinal points: in brief, the north and east are associated with the ancestors, and are considered auspicious and sacred, while the west and the south are profane and dangerous (Hébert 1965; Vig 1977).

In my early days of fieldwork in Antanety, armed with a cursory knowledge of such arrangements from my preparatory ethnographic readings, I was able immediately to identify Tananomby as a slave-descent settlement. Without this foreknowledge, it might have taken me much longer, for although the distinction between the two groups was evident in the village's cosmologically influenced topography, it was almost completely absent from discourse: one of the central organizing principles of village life was shrouded in silence. Not only did people hardly ever mention this topic, but when they did it was with immense care and circumspection.

This article takes the view that silence and speech are politically creative forces. It analyses how, in avoiding the topic of slavery, or in carefully using words that mask or reveal status difference, people in Antanety position themselves and others socially, making strategic moves that reflect both personal motivations and larger structural constraints. The fact that slave descent is highly stigmatized makes this micro-politics of identity an extremely sensitive and precarious process.

The social force of words is the central principle of speech act theory as originally proposed by Austin (1962) and developed most notably by Grice (1975) and Searle (1969), and adopted in a diverse range of disciplines, including social psychology, sociolinguistics, and anthropology. This theory has shifted the analytical focus from the abstract content of language to its social effect. In this perspective, the meaning of an utterance can be found in the use to which it is put. The theory privileges the insight that speech acts take place in the context of an awareness of and concern for their interpersonal implications, notably the identities of interlocutors and the relationship between them (Holtgraves 2002: 37). Moreover, speech acts do not simply reflect the static status of those identities and relationships. They also challenge, confirm, modify, and remake them with each interaction.

The analysis of speech acts need not be limited to verbal utterances. In the context of a spoken interaction, silence itself is communicative and functional, and can therefore be analysed for illocutionary force and perlocutionary effect (Saville-Troike 1985: 6). In a detailed analysis of verbal exchanges in Ghana, Kofi Agyekum (2002) shows how the Akan use ‘eloquent’ silence to produce a range of rhetorical effects such as reverence, awe, or contempt. The impact of these silences, he argues, is ‘to organize and regulate the social relationships among members of the Akan speech community with regard to position, status, gender, and age’ (2002: 34). Such is the social force of silence. It is, in effect, a non-spoken speech act.

Silence, then, is neither a communicational void nor a deviation from ‘normal’ communicative practice. It is an integral and meaningful part of that practice. Its meaning depends on the context in which it is produced and the varying intentions of those producing and receiving it. For example, it can work as an assertion of power (e.g. Clair 1998) or an expression of powerlessness (e.g. Agyekum 2002: 33); it can operate as a face-saving mechanism (e.g. Lebra 1987: 347) or a face-threatening act (e.g. Ambuyo, Indede & Karanja 2011: 211); it can be desirable in one speech community (e.g. Lehtonen & Sajavaara 1985) and discomforting in another (e.g. Tannen 1985).

The silence about slavery in Antanety has a different character to the examples given above. Rather than being the eloquent absence of words in particular conversations, it is a systematic avoidance of the topic that is generalized throughout nearly all conversations. Despite these differences, there are productive parallels to be drawn. In an illuminating article on why people opt out of discussing certain topics, Malgorzata Bonikowksa argues that ‘[t]he opting out choice is as much a pragmatic choice as any strategic choice employed in speech-act performance, made through activating the same components of practical knowledge’ (1988: 169). This practical knowledge is comprised of a mixture of personal, social, and cultural factors such as status, power, rights, topic, face, and sentiment. It is engaged as much in deciding whether to refer to something as in deciding how to refer to it (1988: 171).

This article examines both the ‘whether’ and the ‘how’ of talking about the ever-pertinent topic of slave status in Antanety. Given the sharp social divide that cleaves the village, the motivations behind individuals’ strategic use of speech and silence are bound to vary. For the same reason, the effects of their speech acts (spoken or unspoken) will vary according to who is interacting with whom.

The relationship between the two groups is complex, having historical, social, economic, ritual, and cosmological dimensions which mean that the unspoken status difference is immanent in all social interactions between them. This raises an issue that recurs frequently in the ethnography of Madagascar: how people negotiate their way through the social problems posed by the tension between a common rhetoric of equality and the reality of hierarchy (e.g. Cole 1997; Kottak 1980; Somda 2009; Wooley 2002) and how, in addressing this problem in speech, they re-create, invert, and subvert it (e.g. Bloch 1978; 1981; Gezon 1999). This tension is particularly pronounced in Antanety because the status difference between people of free and slave descent is severely troubling, referencing as it does both past injustice and present differentiation as well as being cosmologically loaded with notions of shame, blame, and retribution. Words on this topic are inherently dangerous, so control must be placed on their expression (cf. Brenneis & Myers 1984).

Theories of face (Goffman 1967), politeness (P. Brown & Levinson 1987), and ‘relational work’ (Locher & Watts 2005) provide useful perspectives from which to examine communication regarding slave descent since they are particularly alert to how actors manage their interactions in view of status difference. The limitation, however, is that they are less able to cope with culturally specific nuances in status or with cross-cutting and contradictory measures of it. This factor is particularly relevant to the study of post-slavery communities, where the new and shifting forms slave identities assume in the years following emancipation are likely to bring new levels of ambiguity to the status relationship inhering between people of free and slave descent. Even studies of pre-emancipation systems of slavery show little consensus on the status of slaves, with slave identity being classified anywhere along a spectrum ranging from alien chattel to potential kin. While this classification depends in part on the system under study, it also depends on which aspect of the status the analyst decides to foreground: the relationship with the master (Kopytoff & Miers 1977), the labour relations (Meillassoux 1992), or the shame attached to the status (Patterson 1982). The key point to note, though, is that the status of a slave or slave descendant need not correspond to a single position on any one spectrum but will likely include various aspects of different positions depending on context. This ambiguity of status is evident in contemporary Antanety and is a complicating factor in terms of how relative hierarchy and equality are handled and produced through speech and silence on the topic of slave descent.

Out of this ambiguity there emerges the micro-politics of relative status. This is not confined to formal political meetings (Bloch 1975; Brenneis & Myers 1984) but is produced in everyday interactions that reveal the dialectical relationship between speech acts and social structure. Like status, speech acts are themselves ambiguous, both hiding and revealing status difference while in dialogue with the structural inequalities that orientate their production.

My hypothesis here is that the systematic avoidance of the topic of slave descent is not just a symptom of the problem of slavery, an embarrassed silence about a difficult past. Like the cautious words people use on the infrequent occasions they refer to the topic, it is also a pragmatic response to that problem, loaded with people's strategies and intentions for dealing with it in their daily interactions as co-villagers. Some of the multiple, contradictory, and even paradoxical consequences of these responses are discussed in the conclusion.

I explore this hypothesis by examining principally two kinds of situation. The first kind involves people of free descent discussing the topic of slave status among themselves. Such conversations are generally rare and tentative. The second kind involves situations in which people from the two groups interact and to which the topic of slave status is relevant. In such situations the topic is rarely mentioned, but its pertinence is evoked both through its very avoidance and through other verbal strategies.

In order to appreciate the complex strategizing that goes into these interactions, it is necessary first to understand what makes the people of Tananomby so different from people of free descent that discussion of their difference needs to be treated with such caution.

Slavery and the village

The village of Antanety, of which Tananomby is a part, sits on a small hillock in the valley of Sahamadio in the northern Betsileo region of Fisakana. The Betsileo are an ethnic group of about one million people based in the southern highlands of Madagascar. While scholars have problematized the ascription of ethnonyms to different Malagasy populations (Astuti 1995; Eggert 1986; Larson 1996), the name Betsileo is nevertheless used by today's Betsileo of themselves. Amongst ethnographers, as well as the Malagasy population in general, the Betsileo are famous for their mastery of two important arts: rice cultivation and rhetoric. The latter of these skills will be discussed in more detail later, for the verbal treatment of the topic of slavery is, I argue, a meticulous rhetorical construction in itself.

The meticulously constructed rice terraces that characterize the landscape of Fisakana provide a large proportion of the population with their subsistence. But as the soil is poor and the land overcrowded, many of Antanety's sons and daughters rely on supplementary means of subsistence. Principal among these are the remittances sent home by relatives working in Antananarivo and other urban centres. Over the last century the northern Betsileo of free descent have prospered as civil servants in the colonial and postcolonial administrations, filling an occupational niche between their northern neighbours the Merina, who occupy the most powerful administrative positions, and the coastal populations, who occupy comparatively few.

The professional migrants’ relationship with their home village is maintained through family – whom they visit and to whom they send money; tombs and tomb ceremonies – which they finance and attend; and land – which they keep under cultivation by employing sharecroppers or day labourers. In Antanety all the sharecroppers are descendants of slaves, whereas the day labourers are drawn from both descendants of slaves and people of free descent. The common factor is that both own insufficient land to meet their subsistence needs.

Madagascar has a long and complex history of involvement with the institution of slavery (André 1899; Campbell 1981; Larson 2001). The expansion of the Sakalava and Merina kingdoms in the eighteenth and nineteenth centuries, respectively, led to both internal slave markets and a lively export trade. Differing economic and political circumstances have created different forms of slavery in different parts of the island, and as its legacy has mutated over time it is unwise to make generalizations. The people of Tananomby are most likely descended from slaves captured in punitive military raids as the Merina empire expanded south- and westward in the mid-1800s (Ellis 1985: 27). These captives were traded in highland slave markets and purchased by landowners for serf labour as irrigated rice farming expanded across the plateau. Some of those landowners would have had administrative or military posts which, as well as absenting them from home, gave them the economic surplus to buy in serf labour. But it was not uncommon for medium-sized landowners with no, or only indirect, outside connections to own a few slaves for agricultural tasks. The fact that slaves were excused from the punishing rigours of government forced labour and military service (fanompoana) so as not to deplete the rice fields of manual labour is an indication of their contribution to agricultural productivity. It may also explain why they largely accepted their slave status and why so many remained with their owners on emancipation (see Campbell 1988; Kottak 1980: 104).

The practice of holding slaves was abolished by the French colonial government when it took control of Madagascar in 1896. The effect of this manumission on the social, economic, and ritual situation of former slaves and their descendants varied according to local political and ecological conditions. In northeastern Madagascar, where land and labour opportunities were relatively abundant, former slaves were able to build independent new lives on a footing almost equal to people of free descent (M. Brown 2004). By contrast, in the overcrowded northern Betsileo region of Fisakana, where status, wealth, and ritual power are closely linked to ownership of the scarce agricultural land, the situation of former slaves has changed little since emancipation.

Today, Tananomby is made up of about ten houses (though some are unoccupied and crumbling) and thirty residents (many of them children), all of whom are of slave descent. The lack of land and opportunity for labour has led to many people leaving for the mining towns or wide fertile plains of the north and west. The comparative poverty of the hamlet is immediately noticeable, and in my early days in the village I was a little nervous of Tananomby, perhaps because I knew that the inhabitants’ poverty was linked to them being a different ‘kind’ of people whose geographical estrangement from the village was calculated and meaningful. As time went on, though, I made some very good friends there and felt the warmth of their welcome, as I also did in other nearby slave-descent settlements.

The people of Tananomby: descent and difference

The geographical marginalization of Tananomby from the main settlement of Antanety is part of a wider social and ritual marginalization of its inhabitants: they are of a different ‘kind’. The word ‘kind’ (karazana) denotes an identity based on ‘intrinsic rather than acquired qualities’ (Astuti 1995: 467). A ‘kind’ is an identity group that, in theory at least, cannot be joined through learning or change: it is prescribed by birth. It is significant that the word shares a root with the word for ancestor (razana), since ancestors and descent groups provide the basis of one of the most salient means of social categorization in rural highland Madagascar.

People of free descent in Antanety usually belong to one of two named descent groups, both of which have a tendency to endogamy, although the two groups have a history of intermarriage meaning that people may trace their descent to either group's apical ancestors. There are also other free descent groups in the valley with whom they may marry and some whom ancestral taboo prevents them from marrying. The corporate unity of these descent groups is maintained and celebrated in lavish tomb ceremonies (lanonana) that occur throughout the dry season, sponsored mainly by educated urban-based emigrants. One key element of these ceremonies is the ritual passage in which the names of key ancestors are invoked and their living descendants assemble, usually huddled together on a mat on the ground, to receive a blessing from descent group elders.

The people of Tananomby, however, do not have named descent groups. The tomb they use, hidden in a forgotten fold of uncultivable land overhung by eucalyptus trees, is shared with people from other slave-descent hamlets in the valley, people with whom there are no necessary ties of kinship, filiation, or marriage. Unlike the bold and visible tombs of the free, the Tananomby tomb is not identified by the name of an apical ancestor. Consequently, their tomb rituals involve no calling together of an ancestor's living descendants. In life, as in death, they are brought together at the tomb not by common descent, but by lack of it.

The other clear line of demarcation between people of free descent and people of slave descent is that of marriage. Intermarriage between the two groups is extremely rare. I know of only two cases in living memory (though there may be more) of intermarriage occurring in the vicinity of Antanety. One case was a free-descent man with a slave-descent woman; the other was a free-descent woman with a slave-descent man. I have been informed by members of the two free-descent families that their families have decided not to allow the offspring of the marriage to be buried in their tombs (see Kottak 1980: 103). This suggests that the offspring are considered by their free-descent relatives to have inherited the apparently natural, irremovable attributes of the slave ‘kind’: they can never become the ‘kind’ of people who are buried in free-descent tombs. This contrasts with the fate of the offspring of a couple from ancestrally tabooed free-descent groups. In such cases a rite called ala fady (the lifting of the taboo) is performed to appease the ancestors and allow the marriage. There is no problem about burial. No such rite exists for marriages joining people of mixed slave-free descent, since it is not considered a taboo that can be lifted.

The prohibition on the burial of children of intermarriages in free-descent tombs suggests the presence of an inherited attribute considered to confer ritual impurity. Slave descendants in the southern Betsileo region are said to have ‘dirty blood’ (Evers 2002: 70; Regnier 2012: 180), an idea reflected in a pair of adjectives used by free-descent people in Antanety in rare and controlled circumstances. These are madio (clean), in reference to themselves, and – much more rarely – maloto (dirty), in reference to slave descendants. I have never heard the latter used, though it is clearly implied when a person utters the former. Nor have I ever heard slave descendants use either of these terms. There are various explanations of this association with pollution. One common explanation is that slaves were physically polluted through dealing with excrement in the course of their labours (Regnier 2012: 180). Some accounts of slavery among the Merina suggest that slaves were deliberately defiled by their masters as a means of effacing any residual free-descent ritual efficacy (Graeber 1997: 374-5). Another explanation is that some noble Merina, ancestrally tabooed from touching the dead, used descendants of slaves to prepare corpses for tomb ceremonies (Vig 1977). Thus they are sometimes euphemistically known as ‘grasshoppers that guard the tomb’ (valala mpiandry fasana). To my knowledge, Betsileo slaves and their descendants did not carry out this funereal service, and I heard no stories of ritual defilement. The common factor in all these explanations, though, is the idea that some kind of permanent contamination – the thing that makes slave descendants a different ‘kind’ – is transmitted across generations.

The above account of ritual marginalization could be criticized for taking as its starting-point a rather normative view of free descent and then presenting slave descendants’ cosmology in terms of everything they lack: ancestors, tombs, descent groups, purity, and so on. But this lack cannot be ignored, since slavery itself, despite regional and cultural variation, is essentially a system of deprivation, an ‘institution of marginality’ (Kopytoff & Miers 1977) which separates its subjects from the very foundations of identity and autonomy. In highland Madagascar, these foundations are land and tombs belonging to named descent groups built on that land: exactly the things the people of Tananomby do not have. They are, then, ‘lost people’ (Graeber 2007), deprived of economic autonomy and historical identity by their forebears’ misfortune of being uprooted from their ancestral land, separated from kin, and scattered far and wide by the slave trade.

The population of Tananomby possess almost no land of their own, not even the plot on which their houses stand. This is why they work as day labourers and sharecroppers in the fields of free-descent patrons, many of whom are absentee urban migrants who either administer the harvest and workforce themselves or use a village-based relative to do so. The two parties are thus mutually dependent: the sharecroppers rely on the landowners for cultivable land, and the landowners rely on the sharecroppers to keep it under cultivation.

The situation in Antanety is very different from that described by David Graeber (2007) in a village about eighty miles away in the western highlands, where a remnant population of descendants of slaves has successfully challenged the economic, political, and social power of the free-descent population. It seems the most likely reason this has not occurred in Antanety is that, unlike Graeber's case, the absentee free-descent population maintain strong connections with the ancestral village, not least of which is their co-dependent agricultural relationship with the people of Tananomby and other nearby slave-descent settlements.

The Antanety case also differs significantly from ethnographic accounts of slave descendants in the southern Betsileo region. Sandra Evers (2002) describes how incomers hoping to settle in a migrant frontier zone who are not able to prove free ancestry are ascribed the status of slave descent and permitted to cultivate only the worst land. Thus the discourse of slavery reproduces itself as a way of administering access to land. The major difference with Antanety is that the two populations have no common history and, perhaps as a consequence, the discrimination is perniciously verbalized. Denis Regnier (2012) describes the situation amongst slave descendants, who, as they own their land, are not economically dependent on their free-descent neighbours. Unlike their counterparts in Antanety, this relative autonomy has enabled them to resist many aspects of marginality, for example through group histories that define their identity on their own terms (Regnier 2012: 243). Conrad Kottak (1980: 104) describes how slave-descent people in the 1970s were reminded ‘in a thousand encounters in everyday life’ of their inferior status. For example, in free-descent houses they were obliged to sit along the south wall with the free-descent juniors, and at ceremonies they were given the cuts of meat normally destined for minors. The descendants of slaves in contemporary Antanety do not suffer these indignities. In fact, it is striking how much effort free-descent people make to avoid allusion to slave status.

The dangerous topic of slave status

So far I have not mentioned the Malagasy word that denotes slaves and their descendants. The word is andevo. It is a word I rarely heard used in Antanety, a word that, to anyone acquainted with highland society, is heavily loaded and – owing to the tensions it can provoke – socially dangerous. I still feel its weight as I write this now. When I presented an early version of this article to a mixed Malagasy and European audience in France, it was only with difficulty that I brought myself to pronounce it.

I was perhaps being over-sensitive. As words take their import from their context, andevo is obviously much more dangerous in the context of a Betsileo village than in a Parisian lecture theatre. Nevertheless, some of its weight remained, as my Malagasy colleagues acknowledged at the time through their reticence to use the word in the question-and-answer session.

Andevo literally means ‘slave’ and is also used nowadays to designate people of slave descent. Another word used is mainty, which means ‘black’. The politics and usage of these words have changed over space and time, and are extremely complex and context-bound. In today's use, mainty is a reference to slave descendants’ generally more negroid physical characteristics (Bloch 1971: 4), although it once denoted ritual impurity and not necessarily slave descent (Campbell 2008: 1279). I shall not enter into various contextual nuances of such terms here as this article is about their avoidance rather than their use. On the whole, in the rare cases where they are applied in a Betsileo village context they are generally interchangeable. It is their import, rather than their strict meaning, that matters.

In fact, the word andevo is so socially charged that it needs to be carefully regulated even in cases where it is applied in a non-Malagasy context. For example, one Sunday morning after church I was standing in a group of men talking, for some forgotten reason, about public executions in Saudi Arabia. My adopted brother Solo was telling the group that the executioners were slaves (andevo) of the king, but as the heavy word came into view he started to glance around over his shoulder and lowered his voice before muttering the term and then carrying on normally. None of the people present were of slave descent, yet they all understood the reason for Solo's circumspection. His fear was probably that the word might be overheard by somebody (and not necessarily a person of slave descent) who would think it was being applied in a local context.

The other way of policing the danger inherent in the topic and vocabulary of slavery is carefully to control the circumstances in which the matter is mentioned. This is exactly what happened the first time somebody talked to me openly about it. This was one afternoon when Ramama (my adopted mother) and I were alone in the kitchen, having just paid off a small group of Tananomby day labourers. ‘You know those people?’ she said in low tones, ‘They're not like us’. I knew what she was getting at but I adopted the old ethnographic strategy of ignorance:

‘Not like us? How?’

Ramama listened for other movement in the house. ‘They're not like us. They're a different kind. They're … ’. She paused again, lowered her voice. ‘They're andevo. You see they are poor. And black. And their hair is frizzy. We don't marry with them, us clean people (olona madio). You know how all the girls are always chasing after Ragry [her third son]. That's all very well, but I've warned him not to go near the Tananomby girls.’

She looked at me knowingly, watching the lesson sink in. Then finally she said, ‘You mustn't talk to them about it. Nothing at all. It makes them too ashamed. So don't talk about it’. She gave me a long, slow look and left it at that. That was the only time we ever did talk about it.

There are two points that seem very salient about this interaction. First, Ramama carefully controlled the context of this conversation. Not only did she make sure that the day labourers had left – that much could be expected; but she also chose a moment when no other family members were present or likely to appear. This topic is what the Malagasy call ‘difficult’ (sarotra), so the fewer people present the smaller the chance of embarrassment or complications. For this reason it was a conversation very much held on Ramama's terms: she chose the moment, she started it, and she finished it. This clear demarcation contrasts with the normal free-flowing and open-ended conversations that characterize most social interactions.

Second, it is significant that it was Ramama rather than any other family member or acquaintance who conveyed this difficult information. This was a topic for an elder to handle, but it would have been unimaginable for Radada (my father) to do this work. As Keenan (1974) has noted, women are permitted (or permit themselves) greater latitude in speaking directly than men. The corollary of this is that they can speak on important subjects that men might avoid or to which men might lend too much ambiguity. This was a case that required a degree of plain speech for the benefit of immature foreign ears.

One reason for the circumspection with which free-descent people speak of this difficult topic may be that slave-descent families have the right to claim compensation of one zebu from any person publicly calling them andevo. This theoretical right has never been exercised in living memory in Antanety. This could be either because people never apply the prohibited term for fear of sanction or because in cases where they have applied it the slave-descent population have lacked the influence to enforce their legitimate claim. Although the threat of the cost of a zebu may act as a significant deterrent, I think monetary considerations are not the sole or even primary reason for the topic's avoidance. After all, Ramama never mentioned this to me. She was much more concerned with the shame that such talk could cause.

The related concepts of shame (henatra) and guilt (tsiny) are fundamental to the way the topic of slave status is discussed and avoided. If injudiciously handled, a conversation about slavery could easily bring guilt or blame to the speaker and shame on the subjects (see Keenan 1976: 76). The notion of ‘guilt’ is linked to its metaphysical counterpart tody (punishment or retribution), and the pair is extensively represented in highland Malagasy proverbs (Andriamanjato 1957; Navone 1977). One well-known proverb runs, Ny tody tsy misy fa ny atao no miverina, which roughly translates as, ‘There is no such thing as retribution, but what you do comes back to you’.

The same could be said of words, for ideas about shame, guilt, and retribution are particularly articulated through verbal action and performance. As Ramama showed when she warned me not to mention the topic of slave descent, it is speech which causes shame and blame, and speech which must be managed, because once words are uttered, you cannot control where they will go or what they will do. For the same reason, formal speeches begin with a fialantsiny, an elaborate set of excuses which pre-empt any offence the speaker may inadvertently go on to cause, and absolve him (it is usually a man) from any guilt or blame. Social differentiation, status inequalities, and slave descent are incontrovertible facts of village life. Since these uncomfortable facts cannot be unmade, the priority is to avoid the trouble caused by referring to them. Such precautionary strategies, in recognizing the social force of words in politics and interpersonal relations, try to rein in the danger inherent in the act of speaking. No topic is more dangerous than slavery, for it is speaking of slavery that unleashes its power for shame and blame.

Playing down status difference

The dangerous potential of spoken words pertains in all interactions where differences in hierarchy can be evoked or exploited. In a society in which there are many gradations of status difference (wealth, educational achievement, profession, tenure of public office, etc.), great care is generally taken – particularly by the ‘superior’ person – to play down and disguise that difference. This gains an added importance when the status difference involves free versus slave descent.

The rhetorical strategies employed to achieve such effects have been extensively analysed in studies of politeness, notably by Penelope Brown and Stephen Levinson (1987). These approaches use speech acts such as requests and commands as a primary source of data since these are heavily inflected with considerations of status. This is true of the dialect of Malagasy spoken by the northern Betsileo, which employs various rhetorical strategies to minimize the power differential implied in making requests or giving orders. This is also particularly pertinent to the topic of this article since one recurring element of the relationship between free-descent and slave-descent people is that of employer and employee.

One way this is achieved is through the use of the passive voice, which can be deployed without an agent so as to request the performance of a particular action while only indirectly implying who is to carry it out. For example, a typical way for an employer to ask a worker to bring some manioc would be to say, ‘Could the manioc be brought?’ (Azo aferina kely ny mangahazo?). The advantage of this agent-less passive form is to allow the hierarchical relationship of requester-requested to be effaced in the temporary fiction of things happening without anybody actually doing anything. The manioc will be brought in, but no individual has been explicitly identified as the recipient of the request. In fact, the request barely resembles a request at all, thus disguising the status difference already existing in the employer-worker relationship.

Another way of disguising the directive intent inherent in a request is to use the inclusive first-person plural form as the agent of a passive or active construction. I have heard this mode used by a landowner instructing a team of slave-descent day labourers to weed a rice field. The way she presented the instruction was, ‘Today we [inclusive] will weed that rice field down to the west’ (Iry tanimbary ambany andrefana iry no havaintsika androany, literally: ‘That rice field down to the west will be weeded by us [inclusive] today’). The landowner did not take part in the weeding, but used her inclusion as an agent to attenuate the directive intent.

It might seem surprising that free-descent employers should go to such lengths to play down the hierarchical elements of their relationship. After all, theories of politeness (P. Brown & Levinson 1987; Goguen & Linde 1983) suggest that people of higher status do less face-saving work in their requests to low-status people than vice versa. In the Betsileo case this is true up to a point. For example, requests made by adults to children lack the high degrees of indirectness noted above. And, as Maurice Bloch (1998) points out, when highland farmers direct the movements of their cattle, they do not use Malagasy but French, known for its associations with colonial bossiness as ‘the language of command’ (ny teny baiko).

It might initially appear that the gap in status between people of free and slave descent is so wide that it would not require a high degree of diplomacy from the former. Nevertheless, they make that effort, as witnessed in both the general silence around the topic of slavery and the face-saving nature of directives given to the day labourers. The need for this is due to two key features of the relationship. The first of these is the stigma of shame that attaches to slave descent identity. This makes it very different from the status difference inhering, for example, between adults and children or farmers and cows: there is no stigma attached to being a child or a cow. This means that drawing attention to either of these statuses will not bring shame on the child or the cow and will therefore not bring guilt upon the speaker. But as there is stigma attached to slave descent, there is a risk that a free-descent employer who does not disguise the status difference implicit in her commands to a slave-descent employee may be interpreted as referring to that ‘shameful’ difference.

The second feature of the status relationship that requires such careful face-saving is that there are cross-cutting elements of status identity that inhere simultaneously and therefore compete. In any status relationship this creates ambiguity. Although in ritual and economic respects the status difference is clear, in other respects it is far from being so, for there are other contexts in which slave-descent people can and do act as equals or may exercise a degree of authority. These include contexts like membership of a community organization, for example a women's group or church, holding positions of responsibility such as a schoolteacher or village official (the president of the Catholic Church in Antanety was for many years a man of slave descent), or simply the fact of belonging to the wider village community or the moral community evoked in the term fihavanana. It is the status ambiguity which these contrasting measures produce, as much as the inherent status difference itself, that creates the need for the cautious, diplomatic speech acts described here.

It is the combination of these two factors – shame and blame, on the one hand, and ambiguity of status difference, on the other – that accounts for free-descent people using such a high degree of verbal caution in matters pertaining to the relationship between the two groups. The foremost mode of verbal caution in view of the latent political force of speech is simply not to talk about status difference at all, particularly not in interactions between the two groups. This is equally true for people of slave descent but for different reasons.

Why slave descendants remain silent on the matter of their status

I have never spoken to a person of slave descent about his or her status. Schooled by Ramama in the free-descent diplomacy of not mentioning it because ‘it makes them too ashamed’, I never dared raise the topic. And of course I don't know if they discuss it in my absence. I have only once heard a person of slave descent refer to it, and this was in reply to a clumsy question I put to her about land ownership in Tananomby. She answered by saying, ‘We are second people’ (Olona faharoa izahay), which might roughly be translated as, ‘We are second-class citizens’. Her tone seemed to be one of resentment mixed with resignation. I moved the conversation swiftly on, but have wondered ever since whether she was offering me an opening.

According to Graeber, it is the horror and shame of the painful memories of the slave past that make slave descendants unable to speak of slavery: ‘Admitting to such a past deprived one of the authority with which to speak. It was inherently shameful’ (1997: 375). But while this might explain why slave descendants don't speak of the past, it does little to explain why they don't speak openly about their current subordinate, marginalized, and stigmatized position.

Identifying the reasons behind this silence is difficult since, unlike in Bonikowska's (1988) analysis of topic avoidance, it was not possible to ask informants why they remain silent on the matter in hand since I did not wish to cause offence (or be fined a zebu). Besides, it is unlikely that people would have been able precisely to pinpoint their motivations or to speak on behalf of others. My approach here, then, is tentatively to outline some of the cultural logic and pragmatic choices at play when slave-descent people choose to say nothing.

One possible reason why slave descendants maintain the silence about their status is that they fear that complaining about their stigmatization would put their means of economic subsistence in jeopardy since their landlords and employers could easily deny them access to land and employment. I think this was the case with one group of sharecroppers, whose alcoholic landlord would occasionally, when severely inebriated, abuse them verbally with reference to their slave-descent status. Other villagers attribute their failure to press charges against the landlord to their dependence on his land and their fear of his status as a member of a wealthy and influential free-descent family.

This case of verbal abuse stands out for its rarity, and was widely attributed to the landlord's desperate alcoholism. Most landlords and employers maintain close, even affectionate, relations with their slave-descent labourers. The fact that labour relations may generally be unfavourable to slave descendants does not militate against a disinterested desire to maintain good social relations and avoid conflict.

Another possible impediment to airing a grievance publicly is the pervasive silence itself. The fact that people of free descent are very careful – out of fear of being fined or of casting shame and accruing blame – to avoid publicly mentioning the stigma makes it hard for slave descendants to complain that they are being discriminated against. Paradoxically, the very mechanisms that protect them also restrict their options for resisting.

This raises a fourth possible explanation, which is linked to norms of communication in a society in which measured and diplomatic control of words is both evidence and instrument of political authority. In this view of political and rhetorical authority, which is largely held by senior, free-descent men, raising the subject of slavery is the kind of dangerous speech act expected of women (who, as we have seen, have a degree of licence in this) and children, who, knowing no better, are prone to ‘say too much’ (miteniteny foana). The problem for slave descendants who might wish to speak publicly about slavery is that, confronted by this rhetorical hegemony, to raise this dangerous topic would be to confirm themselves as socially and politically inferior.

Finally, it could be that some slave descendants avoid the topic because they have actually internalized the discourse of their inferiority and so, consumed by shame, balk at exposing this publicly. This seems to be the view held by people of free descent, as witnessed by their worries about shame and blame: they clearly expect slave-descent people to see themselves in the same light as they do. Evers (2002: 52) and Rasoamampionona 2000: 374) both argue that the Betsileo slave-descent communities they studied have internalized the stigma attached to their ‘kind’. Regnier (2012: 230), however, concludes that the slave descendants he lived amongst actively resist the stigma, notably through the creation of their own descent histories. Clearly, the conditions for internalization of slave status depend on the social, political, and economic context of each case as well as on the personal characteristics of each individual.

The reasons behind people's verbal strategies are never unitary. The generalized silence about slavery is not the product of one idea, but is produced by a range of motivations and strategies that create the politics of interaction. The following ethnographic example is a scene from village life involving free and slave descendants in which the subject of slavery is central to the relationship between the actors. The interaction reveals the contradictions and ambiguities at work: aspects of hierarchy and equality, closeness and division, resistance and acceptance, labour relations and social relations.

Co-dependence and the evocation of equality

Rakoto and his wife Rafara were from the slave-descent hamlet of Ambohikely, a mile to the west of Antanety. They had no kin at Tananomby, but they buried in the same damp and hidden tomb, which was used by people from several slave-descent settlements in the valley. I got to know them well as they worked for the family with whom I live in Antanety and with whom I have a very close, long-standing relationship of fictive kinship.

Rasoa, my adopted eldest sister, employed Rakoto and Rafara regularly to work the fields she shared with her brother. She paid them with money he sent back from his business in Antananarivo and from the wages she earned as a schoolteacher. Rasoa told me that her paternal grandfather used to employ Rakoto's grandfather to work in his fields. This would have been in about the 1930s, but I suspect that the relationship between the two families may have been older still, and could even have originated as an owner-slave relationship. Rasoa, however, maintained that her family had never owned slaves.

Whatever the facts of the past, the relationship between the two families was close and personal, not merely utilitarian. Rasoa offered consistent and generous patronage, finding work for Rakoto and his family even in the slack dry season, hiring his sister-in-law even though she was a slow and inefficient worker, and using his boys to mind the cattle. In turn, Rasoa was invited to housewarmings and festivals at Ambohikely, and received gifts and visits if a family member was sick. As a leaving present after my first long period of fieldwork, Rafara wove for me a small raffia mat. I used it daily for years until, to my sadness, it eventually fell apart.

Rasoa paid the wages at the end of every working day. Standing in the yard, she would approach Rakoto and, holding the cash folded into her two hands, would pass it over subtly while uttering thanks. Rakoto would take the money in his two hands and slip it into his ragged trouser pocket without looking at it. Then he would embark upon a mannered little speech, humbly asking that both parties be blessed with health and prosperity, and stressing the ties that bound them by the repeated and conventional use of the inclusive first-person plural (isika). This pronoun, which contrasts with the exclusive form (izahay), is commonly employed in political and ritual contexts as a rhetorical strategy by speakers wishing to underline or create alliances. It contains immediate illocutionary force since its utterance automatically creates unity between the speaker and the audience.

This unity was further evoked by reference to the value of fihavanana. This word, frequently used rather sentimentally in speeches and proverbs, has a range of related meanings all connoting positive values of closeness, loyalty, and group solidarity (see Jackson 2008: 218-19). It can refer to blood ties, fictive kinship, friendship, or membership of a wider co-operative moral community. When I first heard Rakoto use this word I was shocked because I mistakenly thought he intended it in the first sense, which would obviously have been inappropriate in the context. In this instance I think he meant it to evoke an idea somewhat less specific than fictive kinship but more personal than the general solidarity of the wider moral community.

The effect of Rakoto's short oral performance was rhetorically to call up the moral, personal, and sentimental elements of his family's relationship with Rasoa's at the moment that it was at its most pecuniary. His words served to distract from his family being the recipient of wage payment and implicitly therefore structurally inferior to Rasoa's. Her actions also served to divert attention from these aspects of the transaction. In this way the two parties could create an image of the present that was not about one of them being the subject – the historical as well as the present subject – of the other's patronage and domination. Rakoto's blessing, ‘May the Lord bless us all’ (Samia ho tahin’ ny Andriamanitra isika rehetra) disguised the long-standing, socially embedded inequality underpinning the relationship and re-presented it in the vocabulary of equality before God.

The paradoxes of silence

The preceding interaction exhibits many facets of the slave-free relationship and the different ways it is dealt with rhetorically. In fact, everything Rakoto says is a negotiation of his status, an oblique and strategic reference to what is not being mentioned: the fact he is a slave-descent wage labourer receiving payment from a free-descent landowner. The same is true of the role played by Rasoa, his employer. In according him this moment of transient rhetorical authority, she contributes to the temporary renegotiation of status, to the ephemeral fiction of equality.

It is arguable that this formulaic rhetoric of equality and fihavanana in which Rasoa and Rakoto collude at the end of each working day not only masks the true hierarchical nature of the landowner-labourer relationship but also serves to perpetuate it and further cement Rakoto's subjugation as an exploited member of the rural proletariat. While this is true to a degree, there are two important qualifications. Firstly, neither the egalitarian rhetoric nor the general silence prevents slave descendants’ recognition of the material bases of their domination. They recognize it and employ speech acts that temper it, acting in the full knowledge that lifting the veil of silence would neither remove their social stigmatization nor reduce their economic dependence. I think this is what Rakoto is doing here. Secondly, the relationship is not simply founded upon labour relations. Between Rakoto's and Rasoa's families, for example, there exists a long-standing exchange of gifts, care, and affection, as witnessed in a small way by the mat Rafara wove for me. This kind of relationship is reminiscent of the lohateny relationships maintained in northeastern Madagascar by former slaves and their former masters following emancipation (M. Brown 2004). Not only did these exchanges of goods, services, food, and money serve a practical purpose, they also probably served to disguise the power inequalities in which they had their source (M. Brown 2004: 634). It would be too cynical to identify the empathy and respect of such relationships as nothing more than the delusional product of a mystifying ideology. It would make of Rakoto's optimistic and affectionate blessing a tool of misrecognized oppression.

I think that the cautious and circumspect way that free-descent people approach the topic of slavery is a genuine attempt to preserve liveable relations between co-villagers of different status. They do not, as do the southern Betsileo described by Evers (2002: 52), deliberately, openly, and vindictively pour out a debasing rhetoric of pollution and inferiority upon their slave-descent co-villagers. Instead, they try conscientiously to hush this up for fear of offending them. In so doing they fail to recognize that the shameful stigma they attribute to slave descent is not, as they think it is, a natural and inevitable attribute of that ‘kind’ of person, but an ideological construct of their own making. It is the descendants of the free, not the descendants of slaves, who misrecognize the mystifying character of the rhetorical strategies that attempt, with the best of intentions, to promote a semblance of equality in this fundamentally hierarchical society.

The paradox of this diplomatic silence is that while it shields slave descendants from verbal disparagement, it also closes off any space in which the injustice could be verbally addressed. Research on race in North American high schools identifies the process of ‘silencing’ by which well-intentioned teachers prevent discussion of race in the classroom on the grounds that race talk is impolite and that one should not speak about the unspeakable (Castagno 2008: 325; see also Fine 1991). This silence exists as a structural fear of naming the great social divide that is race, with the result that ‘critical conversation’ addressing social and economic iniquities never takes place (Fine 1991: 34). The silence, which is encouraged and enforced mainly by white teachers, acts as part of the hegemonic apparatus of ‘Whiteness’, a dominating ideology that justifies the oppression of non-whites by white people while allowing the latter to ignore their implication in that process (Castagno 2008: 320). The result is that ‘[s]tudents are being schooled in both the ideological and institutional aspects of Whiteness even when teachers don't say a word’ (Castagno 2008: 324).

It is far from clear whether, either in American schools or northern Betsileo villages, breaking the silence would actually solve the problem. In the case of Antanety this can only ever be answered hypothetically since there is no dialogue – nor any prospect of dialogue – between the two groups on the subject. Besides, however deeply slave descendants might resent their stigmatization, they do not appear to believe that ‘critical conversations’ on the topic will improve their situation – in fact, they might simply aggravate it. The small acts of status renegotiation that do take place are of the kind employed by Rakoto: adopting the rhetorical style of free elders to appeal to principles of commonality and equality while avoiding the very topic that motivates such claims in the first place.

It could well be that the silence and verbal caution maintained by people of free descent only serve to perpetuate their discriminatory vision, by denying a rhetorical space in which the de-stigmatization of slave descent could be openly and conclusively resisted. If this is true, it is not deliberate. The avoidance of the topic is motivated by a genuine desire not to cast shame upon a people whose stigma they see not as a creation of their own free-descent discourse but as inevitably and inexorably fixed by accidents of birth and history. The paradox of this is that the veil of silence intended to protect slave descendants from exposure to the ‘shame’ believed to be natural to their kind turns out to be instrumental in protecting and even propagating the very ideology that created it. Similarly, threatening to fine people a zebu for calling someone a slave only entrenches the ideological stigma of slavery. The effect is cumulative: the more the stigma of slavery is avoided, the more ‘unspeakable’ it becomes.

Such are the ironies and contradictions created by the various strategies villagers employ to handle the difficult reality of extreme differentiation. They are bound together by long-standing cohabitation and collaboration, yet divided by an ineradicable difference. The generalized silence that exists about slavery is a mutually constructed speech act, built from a range of different motivations and rhetorical strategies, and reflecting a variety of pragmatic choices. The silence is not just the product of a difficult past; it is an active force that shapes the present, creating a fleeting equality while reinforcing status difference. This draws our attention to the fact that the political power of speech lies not only in what people say, but, just as importantly, in what they avoid saying. Silence is potent.


  1. The author expresses his gratitude for their helpful comments to members of the UCL Anthropology departmental seminar and to the Editor of the JRAI and three anonymous reviewers.

Parole, silence et descendance d'esclaves dans les montagnes de Madagascar


Cet article examine les usages et les effets des mots et du silence, par l'analyse des stratégies rhétoriques employées en relation avec un clivage fondamental de la société des hautes terres de Madagascar : celui qui divise les descendants d'hommes libres et d'esclaves. Ce sujet n'est jamais évoqué entre les deux groupes et le silence qui l'entoure, comme les mots prudents utilisés pour minimiser les différences de statut, constituent une micropolitique rhétorique de la vie villageoise. L'auteur adopte le point de vue que cet évitement total constitue un acte de parole généralisé, autrement dit qu'il est constitué suivant diverses motivations et stratégies et a des effets multiples et contradictoires. L'un de ces effets est que tout en permettant de susciter une fiction viable d'égalité, ces stratégies rhétoriques ancrent aussi plus profondément encore la division.


  • Luke Freeman is Lecturer in the Department of Anthropology at University College London. He is author of several articles on highland Madagascar on the topics of education, speech, power, and the production of educated elites.