Recognition

John Comaroff identifies ‘public recognition’ as the ‘final element in the creation of a legitimate union’ (Reference Comaroff1980: 167) among the Tshidi, one subgroup of the Tswana. While recognition is implicit in the four other elements he identifies – patlo negotiations, the prestation of gifts, cohabitation, and bogadi – he distinguishes it from these because it is not linked to a specific event. I propose to build on this observation by foregrounding recognition as a frame of analysis that draws together the cumulative social processes and events that mark contemporary Tswana marriage and pregnancy, with special relevance in a time of AIDS.Footnote ³

While ‘recognition’ is an etic framing, I use it to capture a range of emic terms and ideas: specifically, go bona (to see), go bua (to speak), go utlwa (to hear/feel) and go itse (to know). These terms appear regularly – often interchangeably – in Tswana conversation, as exclamations and challenges. O a bona (you see) is frequently appended to the end of sentences, as is o a itse (you know). O a utlwa (you hear) is affixed to instructions or requests. Such injunctions may indicate the clarification of ambiguity, an invitation to agree, an attempt to convince, or an implicit insistence on being heeded; and responses cast in the same terms may mark either willingness or refusal.Footnote ⁴ Recognition, in this sense, is perpetually sought, but frequently evaded and contested. And it takes on special relevance in the context of both self-making and relationships. Among the Tswana, love, care, understanding and so on involve not simply sentiment but action, demonstration and performance, so that they can be seen, heard and felt (Alverson Reference Alverson1978: 138; Klaits Reference Klaits2010: 6). In being seen, heard and felt, they create intersubjective effects: health, strengthened relationships, prosperity, and the capacity to give and evoke love and care in turn. At the same time, refusals or misinterpretations of such demonstrations can produce jealousy and scorn, which also take the form of sentimental action, with potentially deleterious repercussions for the well-being of others – including the threat of witchcraft (Klaits Reference Klaits2010: 4–5). In this sense, recognition is both a key dimension of intersubjectivity and sociality, and a key source of social risk (Durham Reference Durham2002; Durham and Klaits Reference Durham and Klaits2002). This duplicity underpins the Tswana understanding of intersubjective personhood as well – and the imperative to maintain a self that is partible, fragmented and concealed in order to protect it from danger (Comaroff and Comaroff Reference Comaroff and Comaroff2001). The management of recognition, then, involves the management of selves and relationships, and, as such, it also structures power. The licence to hear, know and speak in the resolution of disputes, for example – whether at home or in the kgotla (customary court) – is held customarily by older men and is instrumental in conveying their authority (van Dijk Reference van Dijk, Dekker and van Dijk2010: 290).

These dynamics of recognition, of course, take on a new significance in a time of AIDS. The recognition of those living with HIV has alternately mediated or foreclosed access to treatment, precipitated alienation from community and kin, or granted ‘therapeutic citizenship’ (Henderson Reference Henderson2011: 24; LeMarcis Reference LeMarcis2012; Nguyen Reference Nguyen2010: 89–110). In Botswana, governmental and non-governmental responses to the epidemic have produced new, formalized modes of recognition, emphasizing the need to know one's status and speak about it with sexual partners, while promulgating ‘confessional technologies’ and ‘a market for testimonials’ seen elsewhere in the management of the epidemic (Nguyen Reference Nguyen2010: 21, 35–60). Tellingly, Botswana's nationwide voluntary HIV testing programme is called Tebelopele, or ‘vision’. Secrecy, concealment and silence, on the other hand, are linked to the spread of the virus – originally cast in Botswana, as elsewhere, as a dangerously ‘silent’ or ‘invisible’ epidemic (ibid.: 2) – and thereby pathologized. These shifting, heightened stakes around recognition suggest one possible link between the parallel ‘crises’ of AIDS and marriage; and the work that marriage does in the management of recognition suggests one reason why churches and other intervening agencies might present marriage as a panacea to the epidemic (van Dijk Reference van Dijk, Dekker and van Dijk2010: 287).

Framing conjugal relationships in terms of recognition, I suggest, avoids the limitations of considering them in terms of either exchange or love, as either collective processes of social reproduction or strictly personal projects – framings that have predominated in the anthropology of marriage and intimacy in Africa especially (Smith Reference Smith, Cole and Thomas2009: 159). Recognition makes room for both affect and economy, sociality and self-making, capturing their mutual entanglements and the tensions between them while underscoring the social creativity of the conflicts that inevitably emerge. It creates a space to draw filial and affinal relationships into the same analytical frame, marking a key point of articulation between the two. And it makes room for ambiguity, partiality and reversibility, incorporating – for example – practices of secrecy and concealment, where relationships may be known but not spoken (Hirsch et al. Reference Hirsch, Wardlow, Smith, Phinney, Parikh and Nathanson2009). Moreover, it accommodates the jural, processual and ritual dimensions of conjugality. In this sense, recognition captures both the historical sensibilities that inform Tswana marriage and emergent practices that may be changing it (Comaroff Reference Comaroff1980; Comaroff and Roberts Reference Comaroff and Roberts1977; Solway Reference Solway2016; van Dijk Reference van Dijk2016).

Recognizing pregnancy

When Boipelo's pregnancy began showing, at about four months, her mother Khumo hastened halfway across the village to her own mother's – Mmapula's – home. Boipelo, not yet twenty, was the eldest of Khumo's five children. Khumo's report to her mother was at turns angry and upset: ‘Who could the boy be, in this village? They're useless! Unemployed, no money. How will we look after a baby?’ Khumo and her children lived in a cramped, two-room lean-to, and they struggled to make ends meet. Boipelo had just finished school, and her mother had hoped she would find work and help build a house. Instead, there was a baby on the way.

Lorato, Boipelo's older cousin, fell pregnant at roughly the same time. Lorato knew about Boipelo's pregnancy from the beginning, but told no one at home about her own. Knowing that it would put enormous pressure on the family to have two babies at once, Lorato and her boyfriend considered an illicit abortion; but he had a good job and was building a house in the city. Perhaps, she thought, they could manage to raise a baby on their own. They decided to keep the child.

Lorato's pregnancy started showing shortly after Boipelo's. When her grandmother noticed, she sent her daughters to call the girl and confront her. Having had her suspicions confirmed, the old woman hastened down the street to confer with trusted neighbours, angry and upset.

For many of my friends in Botswana, as well as for Boipelo and Lorato, pregnancy marked a major watershed in relationships with lovers, in family relationships, and in life trajectories. In most cases, it preceded – but seldom precipitated – marriage (this is a long-standing trend; see Schapera Reference Schapera1933; Comaroff and Roberts Reference Comaroff and Roberts1977: 99). Pregnancy was often, though not always, the point at which a courtship became unavoidably apparent. It brought sexual relationships, otherwise carefully kept secret, into the sphere of the seen and the spoken, the known and the negotiable. It made them recognizable. And this shift was part of what gave pregnancy an aspect of crisis, both for the soon-to-be parents and for their families. It was a risky shift: pregnancy rendered the existence of an intimate relationship recognizable, but not its critical details. There was no incontrovertible means of identifying the father and no certainty that his partner would name him, nor that he or his family would consent to be recognized. If he were willing to be identified but he and his family had few resources, the mother's family had little hope of laying charges or claiming financial support for the coming child and might wish that he had stayed hidden. On the other hand, if he was well off, charges might be laid (a colonial-era invention; see Schapera Reference Schapera1933: 84) but might not be honoured, and might thereby undermine the relationship itself. The recognition of pregnancy was, in other words, a source of numerous dikgang, which needed careful negotiation both between couples and within and between families. The success or failure of reproducing family lay in the success or failure of these negotiations, as much as or more than in the pregnancy itself. Success, in this context, meant leaving these dikgang at least partially unresolved. Such a suspension did not necessarily stabilize the relationship, but it left open the eventual possibility of marriage.

After her distraught visit to the neighbours, Mmapula set the mechanisms of pregnancy negotiation in motion on two fronts. She asked her sons – Lorato and Boipelo's uncles, Kagiso and Moagi – to talk to the girls individually and to find out who the fathers of the children were. They learned that Lorato's boyfriend was older, and employed, although he was from far away. This information gave Mmapula hope: if the negotiations were handled properly, he would be in a good position to support the child and might ultimately prove to be suitable marriage material. In the meantime, she could fine him. She dispatched the uncles to summon him to the yard. Boipelo's boyfriend, by contrast, was a former neighbour, young and sporadically employed, and his family was not well off. His family's proximity meant that they could easily have been called or visited, but the matter was not pursued. In fact, the boy's family was not officially notified about the pregnancy until after the child had been born, although he and Boipelo remained involved.

Lorato's uncles sought out her boyfriend, but he evaded his summons. On a couple of occasions Lorato was visiting him when her uncles tried to call him, and she identified the callers. When he still refused to answer, she began to doubt his willingness to take responsibility for the child he had fathered. ‘He said, “I haven't done anything wrong, why should I be called?”’ she explained, still hurt by the refusal. ‘I told him he couldn't refuse to speak to my uncles. I asked him if he was refusing the child. He didn't say anything.’ In her mind, his rejection of the summons suggested a rejection of the potential for kinship that her pregnancy had initiated.

Eventually, Mmapula herself acquired the telephone number of the man's family from Lorato and phoned his mother to report the pregnancy and assert a charge of 5,000 pula (roughly £425, enough for a couple of cows or a good bull) for ‘making our daughter's breasts fall’ (for a description of the ‘fence-jumping’ fine, tlaga legora, see van Dijk Reference van Dijk2016: 32). The man's mother agreed to report the charge to her son, but promised little more. The matter was left there.

After that point, the man was sufficiently ‘known’ to Lorato's family that they would ask after him, talk or joke about him as a potential husband, and allow Lorato to visit him for a few days at a time. As the pregnancy progressed, he supplied Lorato with ample food, clothes, lotions, magazines and supplies for the child, reassuring her that he recognized the child as his own. But this mutual recognition remained tentative and tenuous; the man had refused his summons, had never officially visited the yard, and had yet to pay the fine levied on him. If he came to visit Lorato, he generally stayed in his car down the lane and avoided entering the lelwapa (courtyard). When Lorato went off to see him, Mmapula occasionally asked, ‘And when is he coming to greet us? Tell him we are still waiting to see him. One of these days if something happens to you, we won't even know where to look for you.’ Boipelo's boyfriend was similarly circumspect, although he had been a frequent visitor to the yard before her pregnancy. He, too, was tentatively recognized as the father of Boipelo's child, and Mmapula occasionally asked after him in private; but he was unable to cater to Boipelo's needs as well as Lorato's boyfriend, and there were few jokes about Boipelo marrying him.

The sorts of recognition conveyed by pregnancy, then, produce multiple dikgang, which are addressed in ways that perpetuate ambiguity rather than eliminating it. This ambiguity produces further dikgang in turn – but also leaves open the possibility of kin-making. Gulbrandsen (Reference Gulbrandsen1986: 22) noted a reluctance to take disputes around pregnancy fines to kgotla (customary court) for formal resolution, in spite of a tendency to favour the woman's cause. He explains this paucity of prosecution in terms of guardians’ wariness about their daughters gaining reputations for being quick to sue (ibid.). I suggest that something simpler is at work: having failed to draw another family into mutual recognition and into the joint negotiation of dikgang, the would-be complainant's family has already failed to make the would-be defendant's family into kin. Drawing the family into formal negotiation at the kgotla may produce a final resolution – usually in the form of a payment awarded – but neither the formal process nor the final decision will produce a husband, nor the community of shared risk and continuous dikgang management that makes kin. Indeed, formal resolution would ultimately foreclose those possibilities. Where fines and agreements are left ambiguous, both recognition and resolution are suspended but can still be pursued – leaving the opportunity of kin acquisition as open as possible, on as many levels as possible, for as long as possible. This open-endedness creates a cycle of conflict and irresolution – potentially extending, as we will see, over the course of generations – and this cycle, I suggest, underpins the production and reproduction of Tswana kinship.Footnote ⁵

After birth

Her grandmother and aunt swaddled the baby boy and took him away before Lorato even knew of his death. At seven months, Lorato had gone into hospital, short of breath and with high blood pressure. The doctors performed an emergency caesarean, but the child's lungs had begun to bleed, and by the time Lorato woke he was gone.

Lorato's cousin Boipelo had been delivered of a baby girl shortly afterwards. The cousins were taken to be motsetse – a term for new mothers in confinement – and both stayed with the baby in a room they shared in their maternal grandmother's yard. Neither of them was meant to move out of the house or yard for a month. Neither could have male guests, and neither could visit her boyfriend nor receive him at home. There were no special constraints on the girls’ movement outside the village, but while they were in the village they could not set foot beyond the gate. Lorato was uncertain of the reasoning for this edict, but connected it loosely to the prevention of drought and harm to cattle, and to the avoidance of risk to people who might cross her path – as well as avoiding risks to herself, her cousin, and the child with whom they were confined. It was also intended to protect against witchcraft and illness, which were especially marked risks given the loss of Lorato's baby (see Lambek and Solway Reference Lambek and Solway2001 on dikgaba, illnesses that afflict children and are linked to jealousy and witchcraft among relatives; see also Schapera Reference Schapera1940: 233–4).

In this sense, a woman's movement out of the yard and around the village after the birth – or loss – of a child presents a further series of dangers, or dikgang, to be contained; and family have a special responsibility in containing them, especially where family-linked witchcraft is implicated. Confinement helps contain these risks in part by reversing the recognition that a woman's pregnancy brings upon her and the relationship that produced it; it renders her and her child temporarily invisible, inaccessible, and their status unknown. The re-emergence of new mothers and babies into public spaces after their confinement is also a carefully managed, gradual process of controlling what can be seen, heard, spoken or known, by whom and how. When Boipelo's baby was first allowed out into the yard, her six-year-old uncle remarked to her, indulgently, ‘Ga re go itse, akere!’ – We don't know you, do we? – as if to introduce himself, while distancing her from the risks that relational recognition might create. Parties are often held for children when they turn one year old, although only family and friends attend instead of the large public attendance expected at most other domestic celebrations. At the end of her confinement, Lorato's maternal grandfather, Dipuo, instructed her to wash her feet, and then led her around the village silently, well before anyone was awake and might see them. He sprinkled her washing water before her, as if to contain the traces she might leave, enabling her emergence by concealing it. Containing recognition cannot eliminate dikgang, but it carefully circumscribes the relational sphere in which they may emerge.

I noted several changes in Lorato after the loss of her child and her confinement. Most notable was her attitude towards her younger cousins. Where she had always been friendly, playful and at ease with them, like siblings, she now scolded them and spoke sharply, gruffly sending them on errands or putting them to work. Indeed, her aunts, uncles and grandparents chastised her if she was too familiar with them. When I mentioned it, she replied with conviction: ‘Ke motsadi [I am a parent]; I can't just play with children anymore.’ Boipelo, too, took on a new tone of authority; she was preoccupied with finding paid work and left her sister with most of the childcare responsibilities. Both women spoke, dressed and behaved differently, and related differently to those with whom they had been most familiar. They had come to be recognized as parents, and as women.Footnote ⁶

Thus, while pregnancy and birth may leave considerable ambiguity in relationships between new parents, and between their families, in one respect they are unambiguous: they reorganize a woman's relationship to her natal family. This reorganization begins in pregnancy negotiations, but is perhaps most marked in the management of dikgang after birth. Neither the father nor his family has any formal part to play in taking on or ameliorating these dikgang, and there is little negotiation involved. If anything, he and his kin are conscientiously excluded. This is the case even for married couples: with their first child, women will generally return to their natal homestead for confinement after the birth (which is increasingly conducted in clinics or hospitals). I suggest that this unilateral responsibility for the risks of birth and their containment works primarily to produce and reproduce kinship between the woman, her child (if there is one) and her natal family, who will be important figures in her child's life whether she has married and moved away from them or not – especially her brothers, but also her sisters and parents.

Notably, the Legaes spoke of neither Boipelo's nor Lorato's boyfriends as batsadi (parents) or banna (men) due to them having fathered offspring. Only Lorato's boyfriend was identified as monna (man), with explicit reference to his potential marriageability. Rather than pregnancy – in which men are only indeterminately recognizable, and from the dikgang of which they are excluded – it is marriage that confers on men a degree of recognition and the ability to reproduce and realign kin relations. But, as the experience of Boipelo and Lorato's uncle Kagiso indicates, reproducing kinship through marriage is also a fraught and uncertain process.

Recognizing marriage

‘Ah, it's not going to work out,’ Kagiso admitted with resignation as he stood under the backyard acacia, absentmindedly pulling leaves from its thorny branches. It had been two months since Kagiso, his parents and uncles had formally visited his girlfriend's house with the hope of asking for her in marriage. The foray had not gone well: to their collective astonishment and dismay, the girl's father had refused even to receive the delegation, much less to negotiate terms with them. When I spoke to Kagiso on his return, he was disappointed and angry, but already strategizing for workable alternatives. His parents were less hopeful. Dipuo had simply shaken his head and left for the farm lands promptly after taking tea. Mmapula, uncharacteristically, spent the entire following day lying on the stoep, alternately sleeping, pondering and talking through the previous day's disappointment with her daughters.

Given his original determination, Kagiso's resignation came as a surprise to me. ‘Are you just going to give up, then?’ I asked. ‘What can I do?’ he countered, smiling. ‘You know, he refused even to come out to greet us,’ he said, describing his girlfriend's father's odd recalcitrance. ‘He just hid in the house. The wife [his girlfriend's stepmother] kept telling us he was coming, but he didn't come.’

Kagiso had been seeing the young woman for two years by then, and he was keen to marry. He had been working assiduously for years to set aside the money needed to pay bogadi, and had since become a respected preacher in a local church; he knew he was a good catch. But Kagiso had had an inkling for some time that his girlfriend's father would prove evasive. The man avoided him and refused to greet him when they passed each other in the street. After some ‘research’, as he called it, Kagiso concluded that there was an unresolved conflict with the girl's mother's family – likely related to the custody of the girl herself. ‘Maybe he took the child when he wasn't supposed to, and they are still disputing it,’ he ventured. Whether the girl's parents had been married was unclear, and her mother had met a strange and untimely death (which, like the death of Lorato's baby, rendered it subject to the suspicion of witchcraft). Kagiso's girlfriend, it seemed, was unable to explain what the problem was.

‘He could have come out at least to reject us,’ Kagiso mused, after a pause. ‘He refused because he knew he had no right. Her cousins on the mother's side told her that man has no say in your marriage. Why is that? The stepmother even said, “You know him – this thing, you have to do for yourself.”’ ‘How do you get married by yourself?’ I asked, perplexed. Kagiso shrugged. ‘Gakeitse!’ he answered – I don't know. ‘But I'll keep trying,’ he added, flashing a confident smile.

Marriage stands at the heart of the unique structural ambiguities and flexibilities of Tswana kinship. Historically, Tswana marriage preferences accommodated both cross-cousins and parallel cousins, a renowned anomaly among Southern African kinship systems (Kuper Reference Kuper2017; Radcliffe-Brown Reference Radcliffe-Brown, Forde and Radcliffe-Brown1950; Schapera Reference Schapera, Forde and Radcliffe-Brown1950). Over time, these preferences created an overlapping and indeterminate field of kin relations, in which any given kin tie might be ‘at once agnatic, matrilateral, and affinal’ (Comaroff and Comaroff Reference Comaroff and Comaroff1991: 138, emphasis in the original) – meaning that, in practice, kin relationships were susceptible to constant contestation and renegotiation, oriented around relative wealth, power, and so on (ibid.; see also Comaroff Reference Comaroff, Comaroff and Krige1981). While nobles married kin more frequently than commoners did (Schapera Reference Schapera1957), parallel cousin marriage – and the principles of ambiguity, flexibility and pragmatic responsiveness to social variables it generated – nonetheless marked an ideal form (an ideal that persists; see the description of ‘Formula One’ marriages among the Kgalagadi in Solway Reference Solway2016: 317). These ideals were markedly insular: rather than simply prioritizing the extension of kinship to other, unrelated households, marriage was in many ways preoccupied with containing, reproducing and reorganizing existing kin relationships (perhaps especially following the decline of polygyny; see Solway Reference Solway1990).

Tswana marriage has long been characterized as a drawn-out, indeterminate, often incomplete and potentially reversible process – rather than a definitive event or state of being – which reproduces and compounds the structural ambiguities described above (Comaroff Reference Comaroff1980; Comaroff and Roberts Reference Comaroff and Roberts1977). By contrast, contemporary marriage seems increasingly geared towards foreclosing indeterminacy. Where the stages of marriage once unfolded over years, it is now concluded rapidly and all at once, with bogadi paid, vows made and spectacular celebrations oriented around one extended event (Solway Reference Solway2016; van Dijk Reference van Dijk, Dekker and van Dijk2010; Reference van Dijk2016). Where it was once an explicitly intergenerational undertaking – a father paid bogadi for his son's bride; a sister's bridewealth enabled her brother's marriage and established her claim on his daughter in marriage for her son (Kuper Reference Kuper2017: 274) – now intergenerational kin involvement seems to be waning (Solway Reference Solway2016). At the same time, marriage itself has been in sharp decline. As the conversation at the beginning of this article suggests, it remains a highly desirable goal for men and women, and for their families, but marriage has become an elusive ideal, increasingly restricted to the middle classes (Pauli and van Dijk Reference Pauli and van Dijk2016). Approaching marriage in terms of recognition and the dikgang that accompanies it, I suggest that this elusiveness remains linked to ambiguity. Rather than being eliminated, ambiguity seems to have been relocated from marriage as such to the pre-wedding phase – and beyond that, into familial histories. Aside from the question of financial strategies and resources, the failure to marry may also be a question of the costs of seeking definitive clarity in intergenerational relationships, which rely on a degree of ambiguity for their continuity. In this sense, I suggest that contemporary Tswana marriage remains preoccupied with the management of existing kin links, showing an uncanny resonance with the orientation of pregnancy towards reorganizing women's natal kin relationships (as described above).Footnote ⁷

Ideally, marriage negotiations involve a step-by-step process of seeking formal recognition for a conjugal relationship: a man seeks recognition from his parents, who seek it in turn from his uncles, so that they might be called in to negotiate on his behalf; his kin seek recognition from his girlfriend's parents and uncles in initial negotiations; her parents seek it from her extended family, and so on – until formal recognition is sought from the state (in marriage ceremonies at the district administrator's office), from God (if the couple marry in church), and from the extended kin and communities of both spouses (in feasts and celebrations). These stages are often preceded by other opportunities for recognition: if the couple has cohabited, for example, or had a child together. At every stage, acts of seeing/showing, speaking, hearing and knowing are explicitly foregrounded, requiring other acts of recognition in turn. And their potential interpretations must be carefully managed, especially given the historical tendency towards indeterminacy and dispute (Comaroff and Roberts Reference Comaroff and Roberts1977). For example, after conducting the relationship itself with great secrecy, Kagiso had to tell his parents of his intentions, demonstrate his ability to pay bogadi, and ask them to call his uncles and speak to his potential in-laws. His parents, having heard his request, had to call and speak to their brothers (his uncles), to demonstrate the viability of his proposal to them, and then ask them to assist in repeating the process of speaking, making known and asking with Kagiso's potential in-laws. The cycle continues right through wedding-related rituals: as Solway (Reference Solway2016: 316) notes, ‘seeing’ and showing bogadi cattle has long been a crucial aspect of conferring recognition on a marriage and on the networks of relationships that enable the achievement of that wedding (which the cattle make evident – although they, too, are subject to multiple interpretations; ibid.; see also Comaroff Reference Comaroff, Comaroff and Krige1981: 172). Today, showy white weddings, photographs and videos seek similar recognition in novel ways, extending the recognizability of the couple's success, and that of their kin, in time and space (Solway Reference Solway2016: 313; see also Pauli and Dawids Reference Pauli and Dawids2017: 23). At stake for Kagiso was not only a ‘form of adulthood … related to authority, to the license to speak’ (van Dijk Reference van Dijk, Dekker and van Dijk2010: 290), but a new role in the family, in which he could ‘tak[e] decisions in family affairs, inheritance and the ownership of property’ as well as negotiating the marriages and disputes of others (ibid.).

But at each stage, negotiations face the risk of increasingly public disagreement, refusal or failure – dikgang that may adversely affect the relationships of the partners and of their negotiating kin, whose contributions to the process remain key. When Kagiso's would-be father-in-law refused to see the delegation or hear their request for his daughter's hand, he not only refused to recognize the relationship but showed Kagiso's parents and uncles that refusal. The refusal undermined Kagiso's hopes for marriage and his claims to adulthood, but it also cast doubt on his family's ability to secure marriage for him and on their status relative to that of their potential in-laws. Even where couples seek to avoid these difficulties by ‘marrying themselves’, as van Dijk (Reference van Dijk2016: 36) notes, potentially fraught disclosures of the marrying couple's resources to their respective kin run comparable risks of inviting jealousy or refusals to assist. To the extent that the recognitions involved in marriage negotiations demand other disclosures and recognitions in turn, such refusals are also often explicit concealments: not only of relative resources, but, as in the case of Kagiso's partner's father, of the unresolved – or unresolvable – dikgang of the past. The would-be father-in-law's refusal to receive Kagiso was a way of keeping the fraught, ambiguous history of his relationships with his dead partner and her family hidden.

Marriage negotiations therefore risk forcing long-standing, unresolved familial issues out of suspension and back into play: whether between a potential spouse's own parents, or between the parents’ respective siblings and extended kin. They risk rendering the ambiguities of those relationships recognizable, often uncomfortably so, to a generation among whom they were previously unknown and for whom they might pose further problems. Fresh marriage negotiations offer a rare means of resolving such long-standing dikgang, but in practice they often exacerbate them. Thus, for example, were Boipelo to get married – as Dipuo reminded her at the beginning of this article – the payment of bogadi from her marriage would go to Dipuo, her mother's father, effectively formalizing her parents’ marriage and thereby resolving the suspended questions of their status, their respective responsibilities, their children's inheritances and so on. Ideally, the distribution of Boipelo's bogadi among her mother's family would strengthen and reinforce their relationships to one another, and reconcile them to her father and his family. A certain separation or independence in Boipelo's family would likewise receive recognition. (The Tswana term kgaoganya – both ‘sharing’ and ‘separating’ – also connotes ‘resolving’.) At the same time, should delays or disputes about the payment of that bogadi emerge between Boipelo's future husband's family, her parents and her mother's parents, the confusion of stakeholders and proliferation of claims could well destabilize relationships even further and derail a potential marriage altogether. Certainly, the inability of Boipelo's father and his kin to successfully negotiate the dikgang of his own ‘marriage’ without the help of his daughter's marriage would also render his capacity to cope with dikgang suspect, thereby further undermining his position.

Similarly, had Kagiso insisted on negotiating his marriage with his girlfriend's maternal kin, responsibility for her custody might have been more clearly defined, and the causes of animosity between her maternal and paternal kin articulated and potentially addressed. At the same time, if – as seems likely – the issues at the heart of Kagiso's would-be father-in-law's evasiveness were deeply irreconcilable, pushing his case could have risked irreparable ruptures in the young woman's family and might have foreclosed the possibility of marriage. In the end, her father having refused to recognize Kagiso's overtures, Kagiso's girlfriend moved north to visit her maternal kin. Her relationship with Kagiso faded into obscurity not long afterwards. Having failed to negotiate the dikgang of recognition, Kagiso found himself back at square one, his role and relationships within his own family unchanged.

Beyond the oft-cited pressures of expense, then – whether for bogadi or weddings – it is perhaps the difficulty of addressing long-standing, suspended dikgang within families, as well as managing the dikgang that emerge between families, that introduces ‘new forms of slowness’ (Solway Reference Solway2016: 318) in the negotiating stage, making marriage so difficult to achieve in contemporary Botswana. Even more than pregnancy, marriage is a deeply fraught but critical means of reorganizing and reproducing families. And this fraught creativity affects not only prospective spouses or their children, but the generations that precede them as well. The tension I have described attaches not simply to questions of exchange or love, affinity or procreation, but to the dikgang generated by recognition. At the same time, marriage is perhaps the only process that offers the structural possibility of resolving the suspended dikgang of the past, while enabling the reproduction of kinship into the future – leaving Tswana families, and particularly their men, in something of a quandary.

This double bind is perhaps most marked in the context of the AIDS epidemic, where the recognition of relationships has taken on new risks, and where associated dikgang threaten to take on new forms, while working in ways familiar from the discussion above. It is to the dynamics of recognition in the epidemic, and the dikgang that result, that I turn last.

Reproducing kinship in a time of AIDS

‘And … she's pregnant.’ Lesedi and I sat in shock for a few moments. It had taken some time to eke this information out of her; she had refused to tell me anything on the phone, other than that her cousin Tumi was in hospital. But gradually, as we sat on the long benches lining the small courtyard of the maternity ward, the story emerged.

Lesedi had found Tumi in the middle of the night, collapsed in the hallway of the house they shared with two other cousins and Lesedi's daughter in the capital, Gaborone. Tumi had been weak and sick for some time, and had lost weight. She had had episodes when she talked nonsense. The signs were straightforward enough and saved articulating the painfully obvious: apparently Tumi herself had known for some time that she was HIV-positive, although it was only the routine test at the hospital that had brought the fact to the attention of her cousin. The pregnancy was an added surprise to everyone, Tumi included.

The last time I had seen Tumi had been at a family wedding some months before. She had come home with a new boyfriend, although she was reluctant to bring him into the yard. A long-term relationship with another man had ended dramatically not long before, upon her discovery of photograph albums stashed under his bed recording his marriage to another woman in his home village. Tumi had met the new man at the clinic where she worked and where he was a client; shortly thereafter, they had begun seeing each other. He talked of the untimely loss of his first wife, and his desire to remarry. When the clinic doctor sent Tumi's workmate a text message, asking her to warn Tumi that the man was HIV-positive – and likely carried a particularly virulent strain of the virus – she was too much in love to care. ‘Or maybe the workmate didn't tell her?’ I suggested. ‘People can be jealous.’ Lesedi shrugged. ‘Who knows?’ she said. ‘I think she just loved the idea of getting married. You know, what girl doesn't want that?’

Three months later, Tumi had discovered that she, too, was HIV-positive. She mentioned it to her new boyfriend, who quickly began to withdraw. Lesedi surmised that the stress of his abandonment had taken its toll on Tumi, making it impossible for her to cope with the combined effects of the virus and – as was now apparent – a pregnancy to boot. I asked Lesedi whether she planned to tell their grandmother, knowing that in such a situation the elderly woman would be certain to come to help. Lesedi hung her head and shook it slowly. ‘I don't think so,’ she said. ‘I'll just tell them about the pregnancy; it's bad enough. We'll look after her here.’

Tumi's tale resonated with many others I heard in Botswana. Whenever I became naively exasperated with friends for putting themselves in pronounced danger of contracting HIV, as I saw it, I was met with similar explanations: a shrug and an assertion that love, the promise of marriage or the desire for a child made sense of the risk (see the description of AIDS as a problem of love in Klaits Reference Klaits2010: 3; Hunter Reference Hunter2010). The dikgang that surround the goals of pregnancy or marriage in usual circumstances, with far-reaching consequences of their own, put this reaction in context. HIV is rendered one of many risks to be borne in the project of making the family and the self, one of many potential crises to be faced in that process. It is a risk people are willing to take in order to build conjugal relationships, which open up opportunities to self-make and to refigure kin relations. In this sense, it is a risk of the same order as others I have described above, many of which also present the threat of illness or death. Indeed, Batswana actively absorb HIV and AIDS into the range of dikgang associated with conjugal intimacy as a crucial means of living with the epidemic.

Even practices that seem to offer little more than an egregious danger of infection – like maintaining multiple partners, as Batswana often do – might be understood to ameliorate the other risks inherent in intimate relationships. Before antiretroviral (ARV) treatment was made widely available, Klaits notes that men in the Apostolic church he studied kept multiple partners ‘in order to “protect themselves” (go itshireletsa), ironically the same phrase used in health campaigns to promote condoms’ (Reference Klaits2010: 131). Klaits links this ‘protection’ to a distribution of love that ensures emotional well-being and the improved chance of return on one's investments in others. Such protection is no less necessary in a time of treatment. Indeed, the imperative to maintain a relational self that is partible, fragmented and concealed, in order to protect against witchcraft, predates and outstrips the particularities of the pandemic (see Comaroff and Comaroff Reference Comaroff and Comaroff2001). It is a sort of protection decisively linked to managing recognizability, to controlling who can see, speak about or know a person, and to what extent. And it suggests that protection against relational indeterminacies is as important as – or more important than – protection against the virus (Hirsch et al. Reference Hirsch, Wardlow, Smith, Phinney, Parikh and Nathanson2009: 19).

Of course, to say that the risk of contracting HIV or developing AIDS is of the same order as other dikgang in intimate relationships is not to say that the stakes remain the same. Public health discourse has actively sought to heighten and shift the stakes of HIV infection, as have behaviour change campaigns run by government, NGOs and international agencies countrywide. In many ways, these responses to the epidemic explicitly pathologize the dikgang I have described, turning the ambiguities associated with intimacy, care, love, sex, marriage, pregnancy and birth into clearer-cut questions of life and death. These renditions seek to raise the stakes around HIV and AIDS by requiring and attempting to refigure its recognition (Henderson Reference Henderson2011: 24; LeMarcis Reference LeMarcis2012; Nguyen Reference Nguyen2010), foregrounding the visibility of the disease over the people and relationships it affects.

HIV became recognizable in Tumi's body in many of the same ways her pregnancy did. Its symptoms became visible gradually, over a period of several months. And, as Lesedi's reflections indicate, it provoked some of the same responses and repercussions. It signalled the existence of a relationship without incontrovertibly identifying the man involved, and it fell to the woman's natal family above all to negotiate the crisis, reasserting her connection to them. I knew young women who returned home to their natal yards to be nursed at advanced stages of illness, much as they might return to give birth and be confined. And nursing – or continuous, intimate care – was a primary means through which the family could address the kgang of illness and seek to contain it. Friends often noted that death after a long illness at home was preferable to sudden death, because it offered family the opportunity to discharge responsibilities still owed to their stricken relative by contributing to her care. Like pregnancy and birth, the recognition of AIDS might thus be seen primarily as reproducing a woman's relationships to her natal kin.

But differences emerge in what is recognized, in the options available for managing the dikgang that arise, and in the repercussions of those negotiations. In Tumi's story, it is recognition of the disease itself that threatens to dominate. The relationship through which it was transmitted, and the people involved, recede from view by comparison. And, in the overdetermined representational context of AIDS interventions, this differential recognition works to change what is made recognizable: that is, mortality and the threat of death, instead of relational personhood and the potential of life. The conceptual distance between recognizing AIDS and recognizing relationships or persons is underscored by Tumi's willingness to accept and overlook her boyfriend's HIV-positive status, which she had many ways of knowing. At the same time, the dominance of the disease in the way in which the clinic staff perceived not only the boyfriend but also his past marriage and Tumi's relationship with him underscores the violent priority of recognition claimed by the virus in contexts where biomedical knowledge and public health discourse hold sway.

The recognizability of AIDS therefore seems to produce dikgang that differ markedly from those that emerge when conjugal relationships are recognized. It throws into question the capacity both of the individual to care for herself and of her family to care for her, without themselves falling ill and dying. As Klaits argues convincingly, AIDS is hard to talk about because it enhances scrutiny around and ‘frequently amounts to critical commentaries on caregiving relationships’ (Reference Klaits2010: 33). In a similar vein, Livingston notes that the care required for debility renders differences among kin problematically visible, as ‘relationships undergo both public and private scrutiny’ (Reference Livingston2005: 3). But the same might be said of marriage and pregnancy. Concern around the Legae pregnancies focused on the family's ability to look after both their own daughters and their daughters’ children, as well as the fathers’ willingness and ability to do so. And Kagiso's abortive proposal meant that his and his girlfriend's relationship, and the full range of kin relationships in which they were embedded, came under deeply problematic scrutiny, highlighting the differences both among and between their families. Dikgang routinely destabilize relationships and call them into question; and, when those questions are addressed, even if they are never fully resolved, they create potential for those relationships to be reconfigured (with mixed results). The difficulty arises only when recognition is shifted away from those relationships, and the dikgang they involve, to AIDS as a terminal disease – adequate responses for which lie exclusively in the hands of biomedicine and public health.

AIDS qua AIDS cannot be reported to a partner's kin the way pregnancy can; fines cannot be levied; uncles cannot be informed and sent to make claims. AIDS cannot be demonstrated to extended kin as a proof of readiness to marry, nor can it be negotiated between two families; and while it may throw the failings of intergenerational relationships into relief, it cannot help address them. But AIDS as a kgang of conjugality, a crisis in the making of selves and of families, can be addressed on those terms. The kgang of AIDS overlaps enough with other conjugal dikgang that it can be absorbed into them, and, where it can't, Batswana actively recast it in terms of dikgang that can be addressed. More than simply ‘hiding’ the disease, this work involves shifting recognition from the disease itself back to people and relationships, shifting the stakes from life and death back to kin-making and self-making. Small wonder, then, that Lesedi would choose to notify her grandmother about Tumi's pregnancy – a kgang about which something could be done, and around which kin-making and self-making could proceed – but not about her HIV-positive status. She worked not so much to conceal her cousin's diagnosis as to subsume it and the crisis it represented in a way that allowed Tumi's self-making and kin-making projects to continue.