‘THE CHILD OF DEATH’: PERSONAL NAMES AND PARENTAL ATTITUDES TOWARDS MORTALITY IN BUNYORO, WESTERN UGANDA, 1900–2005

Abstract

Traditional sources tell us relatively little about how Africans perceived death in the past. In some societies, however, changing attitudes towards mortality can be identified from the names which were given to babies. In Bunyoro almost a third of the names that were given during the colonial period referred to death. The declining frequency of death-related names from the 1940s offers significant insights into the impact of Christianity, education and population growth on the Nyoro's worldview. That death-related names did not re-emerge in the era of AIDS is a significant indication of how the pandemic has been viewed in western Uganda.

Q. Can you please one by one tell me your name and why you were given it? A. Nkiboinehati (I have just realised it [that this is a blessing]): seventeen children were born before me and out of those fourteen died. I was the last born …

A. Birakurataki (it will follow others): twelve children before me died and they thought that I would also die.¹

INTRODUCTION

Africa's modern demographic history is undeniably dramatic. Over the course of the twentieth century the estimated total population of the continent increased sixfold. From the late colonial period a surge in birth rates culminated in Kenyan women achieving a remarkable total fertility rate of 8·1 live births in 1980. Death rates fell just as precipitously. Under-five mortality rates in Kenya, for example, fell from 205 per 1,000 in 1960 to 97 per 1,000 in 1990. Towards the end of the century though, both fertility and mortality trends underwent a major reversal across the continent. In Zimbabwe total fertility rates halved between 1980 and 2004, while AIDS-related mortality caused life expectancy at birth in Botswana, which had risen from 46 years in 1955 to 65 in 1990, to fall back to 34 by 2005.² While recent changes are relatively well understood, the weaknesses of colonial data, especially relating to mortality, have limited the progress achieved by demographic historians in explaining Africa's transition to rapid population growth.³ If a detailed understanding of any kind of quantitative change during the colonial period is difficult to achieve, so too is a sense of individual experience in the African past. Historians of Africa possess relatively few sources which give an insight into the outlook of individual Africans in the colonial period. Documents such as personal diaries and correspondence, autobiographies, or court case transcripts have been used to great effect,⁴ but such records are rare or tend to have an elite bias. These imperfections can be counter-balanced to some extent by oral interviews,⁵ but historians are increasingly aware of how present-day concerns or written sources can affect oral memories of the past.⁶

However, in some societies, there does exist a large-scale source which seems to indicate very clearly what individual Africans' worldview was at a particular point in time. In some places, in the past, babies were given names at birth which reflected the context in which their families found themselves. Many of these names relate to marital strife, conflicts with neighbours, changing religious beliefs, or attempts at ethnic integration. Most interesting of all, though, were those names which give an insight into parental perceptions of their demographic situation. This article will focus on one western Ugandan society, Bunyoro, which experienced high rates of infertility and mortality during the colonial period. As birth registers dating before the 1970s from this region have not survived, the best sources of children's names are church baptismal registers. This paper analyses the names given to over 22,000 babies baptized between 1900 and 2005 in the Roman Catholic parish of Bujumbura-Hoima. Of all names given during the colonial period, almost a third had some direct reference to death. Such names could relate to parental mortality (Kasigwa: the child's father died before it was born), to the deaths of previous children (Kalyongera: this one will also die), to the anticipated death of the recently born child (Byarufu: the child belongs to death) or to parents' future reproductive plans (Kabainura: death forces us to stop bearing children).⁷ The incidence of death-related names reflected what is known of mortality trends in colonial Bunyoro reasonably closely. But this link between naming and perceptions of mortality risk weakened after independence, and did not revive significantly despite the emergence of AIDS-related mortality in Bunyoro in the 1980s.

This article will analyse these trends in naming patterns, and argue that changes in naming reflect both variations in perceptions of mortality, and a wider, if uneven, shift in consciousness. While an interest in hybridity has long dominated analyses of African religious and therapeutic beliefs, it has often proven difficult to develop a sense of how the balance between local and introduced ideas has changed over time.⁸ The names dataset provides a chronologically precise guide to the evolution of attitudes over the past century. Whereas very many Catholic Nyoro parents gave their children a saint's name and a name referring to indigenous Cwezi and embandwa spirits in the early 1900s, the gradual shift towards outward orthodoxy and, in some cases, deeper belief was marked by a new category of ultra-Catholic names. Similarly, ever-widening education helped Western biomedical explanations for disease and death come to rival those which emphasized malevolent spiritual power, in public discussions at least. Intense, sustained HIV-related propaganda by the Ugandan government and a host of NGOs has caused a marked shift not just in sexual behaviour in this region but also, it seems, in the credence given to scientific analyses of the mechanisms of morbidity and mortality. Finally, the article will suggest that local perceptions of AIDS-related mortality in this part of Africa have been relativized by the Nyoro's colonial experience of demographic decline followed by rapid population growth from the 1960s.

THE WEAKNESS OF CONVENTIONAL SOURCES

Data problems are a common feature of African historical demography. As Bruce Fetter has observed, colonial census records are an ‘unsavory witness’ of early twentieth-century population change in Africa.⁹ In Uganda the 1931 census, a vast improvement on earlier censuses, was still merely an extrapolation from enumerations of 40 per cent of villages. Bunyoro's 1948 general census results somehow disappeared before reaching Kampala. The 1959 census not only survived, but was recognizably modern in its nature, yet it was held over a three-day period, reducing the reliability of the figures. Moreover, the 1948 and 1959 sample censuses in Bunyoro were relatively small and so had large error margins.¹⁰ The value of the first post-independence census of 1969, meanwhile, was limited because the published data did not categorize people by ethnicity, so it is impossible to know what proportion of Bunyoro's population growth was the result of immigration. The theft of the 1980 census results, prior to their full analysis, re-emphasizes the inadequacy of Uganda's official population counts in the past.

Despite their failings, Bunyoro's censuses do provide a sense of overall population trends. Most significantly, they show that Bunyoro was the only district in Uganda to experience an absolute demographic decline during the colonial period, its population falling from 130,922 in 1911 to 126,875 in 1959. Only in 1959, by which time Bunyoro's population was well on the way to recovery from its early colonial crisis, do the census results give a clear indication of the causes of Bunyoro's decreasing numbers. This census revealed that, then, Bunyoro was the only district in Uganda still to suffer both low fertility and high mortality. But it also showed that Bunyoro's crisis was over. Compared with the 1948 sample census results, the general fertility rate, for example, had increased by 23 per cent, while infant mortality had fallen by 13 per cent to 180 per 1,000 live births.¹¹

While the published census mortality data were aggregated at the district level, contemporary reports and oral testimonies indicate that in fact the risk of death varied sharply across colonial Bunyoro, depending on individuals' wealth, education, status and location.¹² The best source of information on such indicators comes from colonial researchers' village surveys. In 1953–4, for example, a nutritionist, John McFie, selected two Bunyoro villages as part of a Protectorate-wide investigation into diet and disease in rural Uganda. McFie chose Kibiro because it was a prosperous fishing and trading lakeshore village and so provided an ideal contrast with Kihoko, an upland village with poor communications, no access to fish, and few livestock, considered to be ‘as typical as possible of rural Bunyoro’.¹³ McFie found that in Kibiro, the average woman had had 2·8 live births, and that child mortality was 38 per cent, while in Kihoko, women had only 2·2 live births on average, and 47 per cent of these children subsequently died.¹⁴ These surveys indicate not only that the demographic experience of late colonial Bunyoro was sharply differentiated geographically, but also that even the wealthiest villages experienced severe demographic problems. A further survey in 1963, part of a comprehensive nationwide agricultural census, found that the most prosperous villages in Bunyoro were characterized by extreme inequality, with the wealthiest Nyoro marrying more wives, having more children and consuming more food than their poorer neighbours. Bunyoro's rural economy was exceptional in two ways. The proportion of Nyoro landholders who had a second occupation to farming was double the national average, and landholding in Bunyoro was much more unequal than anywhere else in Uganda. Bunyoro's two distinctive features were connected, with the people who had second jobs being both those who were very poor and those who were very wealthy in land. As Kitching argues for Kenya, additional employment seems to have been the road to real wealth for some, and a necessity for survival for those who had to devote part of their labour to other people's farms.¹⁵

Mortality varied with income in the later colonial period, but we still have little sense of mortality trends in the early twentieth century. For these years we have to rely mainly on the district's vital registration records. From 1910, chiefs in Uganda's ‘advanced’ districts were required to provide annual counts of births, deaths and still-births. These figures are of uncertain unreliability because of a lack of supervision and problems of definition. Some babies who died soon after birth may have been recorded as still-births, while many deaths in infancy were probably never recorded at all.¹⁶ Nonetheless the data indicate that Bunyoro suffered a demographic crisis that was significantly greater than anywhere else in early colonial southern Uganda. Between 1910 and 1924 Bunyoro's mean live birth rate was 19·3 per 1,000, the lowest recorded in any district, and the mean death rate was 27·5 per 1,000, which was significantly higher than average, while 31·9 per cent of all births were still-births, by far the highest rate in Uganda (see Figs. 1–2).¹⁷ These figures do seem to indicate significant population decline in Bunyoro. Unfortunately the colonial government's checking of vital registration fell away in the late 1920s, so that reports of massive, rapid demographic changes passed unchallenged. For example, Bunyoro's infant mortality rates supposedly fell from 535 in 1926 to 62 in 1938, which was a lower level than most European countries enjoyed, and was undoubtedly wrong.¹⁸ The great hope for historical demography in Africa is family reconstitution from parish registers of baptisms, marriages and deaths. In some places such data have greatly advanced understanding of when and how historical demographic change occurred,¹⁹ but in Bunyoro most people died far from mission centres and were buried on their own land with neither the bereaved nor the local clergy prioritizing the official recording of deaths in the parish registers.

Fig. 1. Births and deaths in Bunyoro (1910–1924) according to chiefs' returns.

Fig. 2. Still-births as a percentage of all births.

THE VALUE OF NAMES

As well as providing an inadequate sense of trends in mortality in colonial Bunyoro, existing sources also say relatively little about how Nyoro felt about, and responded to, the extremely high mortality levels they experienced during the colonial period. Precolonial conceptions of death are better evidenced. British military sources emphasized Nyoro soldiers' fearlessness in battle, while missionaries at the start of the twentieth century complained of the resignation with which Nyoro awaited death when suffering from severe disease or famine.²⁰ Mortality levels appear to have been very high for much of the late nineteenth century, if contemporary observers and oral interviewees are to be believed, due to new epidemics, slave raids, civil wars, invasions and famine. But over-familiarity did not mean that Nyoro were unaffected by death before the coming of Christianity. Ethnographic accounts emphasized that the Nyoro cared for the sick, in case they died discontented and then troubled the living from the underworld. While the powerful strove to establish heritable rights to landed property through their burial sites, people feared to speak of their family's health or prosperity lest what missionaries described as ‘malicious spirits’ brought death to punish their hubris. Traditional funerary practices expressed respect for the ritual danger associated with the transition from life to death; most of the Cwezi spirits had some power over disease, especially epidemics, and death; and kubandwa spirit possession focused heavily on fertility and mortality. Belief in an afterlife existed on several levels. Thus the mythical Cwezi dynasty did not die but retired to the otherworld, while at death people's spirits might move on to this parallel universe, okuzimu, but could still intervene for good or ill in the lives of those left behind. These mizimu spirits, if properly commemorated, could protect their families from misfortune, but if neglected after death or mistreated during life, they could afflict the living with illness, poverty and other evils. Being forgotten after death was deeply feared, which partly explains why childlessness and, to a lesser extent, infant death were so tragic in this society.²¹ Colonial-era accounts are less rounded, focusing on the local leadership's fears that the Nyoro were ‘a dying race’, and the unease induced by the competition between aggressive missionization and local perceptions that murder by means of sorcery was on the increase.²² The analysis of naming patterns will add new complexity to our understanding of Nyoro perceptions of mortality, and above all reveal how these changed significantly over time.²³

The value of names as a historical source has not gone unnoticed by historians of Africa. The re-naming of individuals in the era of slavery has been interpreted as an act of domination, through which ‘the previous social persona of the slave [was] obliterated’. Naming can also be an act of resistance – a means, for example, of protecting female multiple identities. A sense of historical change in the pattern of name-giving comes from evidence of Ethiopians' abandonment of monarchist names after the 1974 revolution.²⁴ The significance of names as a window into the character of a community, and its conflicts, also figured prominently in John Iliffe's recent work on honour. Iliffe showed, for example, how late nineteenth-century Creoles in Sierra Leone took on African names as a symbol of their retreat from uncritical Westernization; how black South African gangs gave themselves names of defiance, such as Gestapo; and how the pastoralist elite of the Ugandan kingdom of Nkore gave themselves laudatory names, whereas the names taken by agriculturalist subjects expressed social tension and self-criticism.²⁵

But Africanists have tended not to imitate the use of large datasets of names by historians of Western societies, who have by this means achieved significant advances in understanding of social and familial relations in the past. Studies of naming practices in early modern North America, for example, indicate that sharp differences existed in family structures and relationships between lineage, kin and the wider community from one immigrant group to another, usually reflecting variations in rules of inheritance.²⁶ Students of slave populations have used names as a means of tracing the way in which Africans became African Americans, exploring the shift from naming American-born children bilaterally after African-born grandparents to giving names associated with Old Testament patriarchs to reflect slaves' conversion to Christianity and incorporation into a patrilineal society.²⁷ And recent research into a data base of Californian babies' names has shown how radically different black and white naming practices became from the early 1970s. This source demonstrated though that personal names were an indicator rather than a determinant of a child's life chances, names being more symptom than cause of the evolving racial divide.²⁸

Names then can be used to link social actors, reproduce social relationships or create change. They establish order, by both categorizing and differentiating people.²⁹ The founding myths of Bunyoro describe the period before the institution of monarchical government as a time when people, like animals, lacked personal names, a time of chaos.³⁰ People, through naming, can express power over the child, within the household, or in the community.³¹ In Bunyoro a child was named usually by the father, though grandparents often exerted influence over the process. Sources disagree on whether the formal naming ceremony occurred three to four days or months after the birth of a child, but names were in any case usually decided before that date.³² Bunyoro had an individualizing pattern of naming, where children typically had a name that was invented for them, and would disappear with them. In the past, parents had few qualms, it seems, about giving a child a name with negative connotations. Often this was regarded as merely formalizing a situation which was already well known to the local community. For some people, though, the naming of a child was a means of demonstrating their awareness of neighbours' or family members' masked antipathy. There is a large repertoire of Nyoro names which refer to hidden enemies, such as Barungindoho (they are nice to my face). The naming of a child was an opportunity for fighting battles publicly while avoiding direct confrontation. Individuals, as they grew up, might adopt new names, but the name which they were given as babies would remain their official title. In the past, shedding an unpleasant birth-name was seen as a mark of disrespect to parental intentions, though in recent decades it has become common for an unpleasantly named child, on reaching adolescence, to seek to adopt a more positive identity.³³ These issues of mutability and evolving identities are of little relevance to this discussion. A baptismal register provides a pathway back to the moment of infancy, before the recipient of a name had a chance to change it.

THE HOIMA DATASET

The main dataset consists of a record of all of the details of baptisms of children that occurred within the Roman Catholic parish of Bujumbura-Hoima between 1900 and 1980, the period on which my demographic research focuses. In addition the details of children baptized in every subsequent fifth year (1985, 1990, 1995, 2000 and 2005) have been entered, in order to permit an analysis of naming in the era of AIDS. The register almost always provides the Christian and Nyoro names of the child, parents and godparents, and the date of baptism.³⁴ It also usually detailed the child's date of birth or age at baptism, but the date of confirmation, and especially the dates of marriage and death, were given less frequently. The recording of death is insufficiently comprehensive for any meaningful mortality rates to be calculated or for full family reconstitution to be achieved. These inadequacies are compensated for, to some extent, by the size of the dataset. Bujumbura parish, when it was established in the district capital, Hoima, in 1900 covered the entire western half of the kingdom of Bunyoro, an area of some 2,200 square miles.³⁵ The number of Catholics within the parish's total population (around 60,000 people during the colonial period) increased rapidly, so that by 1924 around 37 per cent of all live births in western Bunyoro were recorded in the Bujumbura baptism register.³⁶

A total of 36,839 children's baptismal details have been recorded, of which 62 per cent have had their Nyoro names translated with the help of a lecturer in Bunyoro's vernacular language at Makerere University, Kampala and a series of focus group discussions. Of those names which remain untranslated, most belong to non-Nyoro ethnic groups, while the outstanding Nyoro names are obscure, each occurring only a handful of times. The names were then placed in a category, such as ‘fertility problems’, and grouped by birth year into five-year periods. Two types of names were excluded from analysis, those that had several potential meanings, and those where an individual's name had been inherited, as this made it impossible to identify changes in attitudes (other than those regarding the nature of the family) over time. Family reconstitution, though, has thrown up very few cases of the transfer of names across generations.³⁷ These refinements have resulted in a final data base which contains 22,767 individuals who possess 1,655 different names.³⁸

Perceptions of death

Despite its size, a degree of caution is still necessary in considering this sample. These babies' parents were almost always baptized Roman Catholics, and so would have differed to some extent from followers of other religions in terms of their worldview, education and perhaps level of income and social class. Catholics, though, fell somewhere in the middle of Nyoro society, poorer and less educated than Anglicans, but enjoying a superior status to Muslims and those who retained their beliefs in indigenous spirits. By the 1920s Catholics were the largest religious grouping in Bunyoro.³⁹ Just how Catholic the members of Bujumbura's congregation were is more difficult to assess. Missionaries strove for orthodoxy, but complained vigorously about converts' reluctance to marry in church, devotion to polygamy and adherence to other un-Christian beliefs and practices.⁴⁰ The names in the baptism register, though, suggest that missionaries' struggle for purity did have some success, at least in terms of parishioners' public behaviour.⁴¹ Whereas many early converts gave their children names such as Kembandwa and Mugasa, indicating that the birth of the child was the result of prayers to traditional spirits, from the 1930s such names begin to decline in frequency, being replaced by names which thank the Christian God, like Mbabazi (God's grace) (see Fig. 3).⁴² From the early 1950s, Nyoro names which are expressly Christian constitute the largest category in the dataset. Yet despite this impressive nominal show of zeal, we still have little sense of how deeply Catholic visions of heaven, purgatory and hell changed most parishioners' attitudes towards death.

Fig. 3. The changing frequency of religious names.

Some of the names in the baptism register, however, do give a more rounded sense of how Nyoro regarded death after 1900. A number of names show that some Nyoro at least conceived of death not only as a state of being or something impersonal which happened to people, but also as an active agent, a character with a personality. Death was malevolent and relentless, but it could be fooled by the wily, who had learnt its ways.⁴³ Of all children in the dataset, 0·5 per cent were given names which aimed at distracting or deterring death. As death was believed to be malicious, parents might pretend indifference to the newborn, or unhappiness with an excessive number of children, so that death, unable to disappoint them, would move on. Thus names such as Kunobere (I hate this child) or Kabaingi (so many children) were quite common. Similarly, some babies were called after unpleasant or insignificant objects, such as rats, goats, insects or monkeys.⁴⁴

Overall, of all the names given during the colonial period, those which related to mortality totalled 30 per cent.⁴⁵ The incidence was highest in the early colonial period, peaking at 40·1 per cent during the five-year period 1915–19, which vital registration records and medical reports indicate was indeed a time of extremely high mortality, as the First World War deprived Bunyoro of its doctors and many of its cultivators, and unusual population movements accelerated the spread of epidemics. In 1918 alone, smallpox and influenza caused 2,759 deaths. Interestingly, though, average mortality levels were higher still in the 1910–14 period. This lag between mortality trends and naming practices is not in fact so surprising, since so many names referred back to previous deaths.⁴⁶ The frequency of death-related names dropped to around 30 per cent in the 1920s and 1930s, fell again in the 1940s to around 25 per cent, and then declined further through the 1950s, reaching a colonial-era low of 15·4 per cent in 1955–9. This change in the pattern of naming follows the general trend of mortality levels in colonial Bunyoro.⁴⁷

Categories of death

This large body of mortality-related names can be broken down into different groups (see Fig. 4). Names like Mulekwa (the child is an orphan), which referred to the death of one or both parents, were relatively consistent through the colonial period, averaging 0·6 per cent of all names given. It is likely that these names underestimate parental mortality. In 1921 the medical department reported that 2·04 per cent of women who gave birth that year in Bunyoro died in labour.⁴⁸ More babies were given names which referred to the death of siblings (1·45 per cent of all names from the colonial period). Nyamayarwo, for example, observes that ‘children are meat for it [death]’, Nkafiika means ‘I am the only survivor’, while Nzarabaki asks ‘why did I labour to produce these children?’ In addition, a number of other names, such as Kaijabahoire (the child was born when people are finished) and Galimaka (it was once a family), refer to non-specific family death. Such names made up 3·5 per cent of the total.

Fig. 4. The frequency of mortality-related names in colonial Bunyoro.

Previous experience of child loss influenced many colonial-era parents' decision to give names belonging to the two most common mortality-related categories, those which meant either the child will die (10·56 per cent of all names) or the child might die (12·72 per cent). This context of high mortality is reflected in names such as Kabwijukya (it is a sad reminder), Rwesemereza (death pretends to be good), Ndoleriire (I am waiting to see what happens), Bagada (what a waste of energy) and Byakutaga (it is a hopeless attempt). Names which anticipated the imminent death of a baby, though, also emanated from parents' perception of the particular vulnerability of the newborn. Thus Katebalirwa, for example, means ‘this child cannot yet be counted’, Nsimireki asks ‘why should I be grateful?’, Byeitaka states ‘the child belongs to soil’, and Kaijakwamya predicts ‘the child has come for a short while’. Karafa dispenses with euphemisms, meaning simply ‘the child will die’. It may be significant that names predicting ‘the child will die’ were more common than those suggesting ‘the child might die’ only in the period 1920–34, the time which followed the years of highest mortality. It seems likely that improvements in healthcare provision in Bunyoro began to increase children's survival chances from the late 1920s. By 1927 Bunyoro had 2 government hospitals, 2 mission maternity hospitals and 5 rural dispensaries. The number of government dispensaries increased gradually to 18 by 1946. Bunyoro by the 1930s was as well served by colonial medicine as anywhere in East Africa.⁴⁹ It seems reasonable to assume that a gradual decline in child mortality levels lay behind the shift back to the less pessimistic ‘child might die’ names from 1935.

It is theoretically possible that the prominence of death-related names in the baptismal register simply results from problems of selection. Children with names of death might have appeared more frequently in the registers than in the population as a whole. It may be that parents whose newborn babies seemed in danger of imminent death would have had their child baptized earlier than the norm.⁵⁰ In fact, though, if we consider the mean delay between birth and baptism during the colonial period, we see the opposite of the expected outcome, with the parents of babies whose names indicated concerns about high mortality waiting 31 days longer than average before baptizing them. But these averages conceal a striking variation in the distribution of the age of baptism. The period of greatest risk for babies was the first four weeks of their lives, which was the time when health scares would have been most likely to prompt an early baptism. Of babies with names which meant either ‘the child might die’ or ‘the child will die’, 16·9 per cent were baptized within this neonatal period. The rate of early baptism for names which did not directly predict death was much lower than this, at 11·6 per cent.

These findings can be interpreted in two radically different ways. They might indicate that babies with mortality-related names are over-represented in the registers, because a proportion of babies whose early deaths were not anticipated would in fact have died during the typically lengthy period between birth and baptism. The risk of mortality in this region was high right through until age five, and parents could not always have predicted the survival chances of their babies accurately. Alternatively, it can be argued that early baptism was not a realistic option for the great majority of children who were born at home, many miles from the nearest priest. Parents might not have wanted to expose a sick child to the rigours of a long journey purely for the sake of accelerated baptism. Instead many would presumably have hoped the child would survive until the next visit by a priest on tour, when, like most babies in rural communities, it would have been baptized with dozens of other local children. This interpretation is supported by the fact that babies with a name which meant ‘the child might die’ or ‘the child will die’ were baptized 23 per cent earlier than average if they were born in Bujumbura itself, and 14–17 per cent later than average if they were born in large outlying villages like Mparo and Butema.⁵¹ The other way of considering the distribution of baptism ages is to assume that those children whose names indicated an imminent risk of death did indeed experience a higher wastage rate than children who seemed less vulnerable early in their lives. If this assumption is correct, and it seems logical, then the baptism registers may actually underestimate the proportion of babies who were given death-related names. Is it more likely that Fig. 5 shows different rates of mortality rather than distinct age distributions for baptism?

Fig. 5. Distribution of baptismal age by name category.

Postcolonial naming

The frequency of all mortality-related names, which was in rapid decline through the 1950s, fell further to around 9 per cent of all names given in the 1960s, 7 per cent in the 1970s, 5 per cent between 1980 and 2000, before reaching a low of 4·4 per cent in 2005 (see Fig. 6). This change in naming patterns outstrips the reduction in actual mortality in Bunyoro over the past century. The highest annual death rate recorded in Bunyoro was 46 per 1,000 in 1913. Data on the crude death rate in contemporary Bunyoro is not available, but it is reasonable to assume that it is not far from the national average, which in 2005 was 15 per 1,000.⁵² These figures indicate that mortality levels fell by roughly 67 per cent between 1913 and 2005, whereas the incidence of mortality-related names reduced by 89 per cent between 1915–19 and 2005.

Fig. 6. The frequency of mortality-related names in postcolonial Bunyoro.

In group interviews in 2006, elders were asked why names which referred to death had become less common in recent decades. Their responses indicated that both the way in which death was perceived and individuals' relationship with the community had changed. It seems that, in the second half of the twentieth century, a new sense of privacy, part of a self-conscious modernity, was valued: ‘Such names were given in the past because people had the spirit of saying let me expose to the whole world what is going on here … Today people want to cover up.’⁵³ In addition, the negativism which had so impressed Bunyoro's administrators and visitors before the 1950s became less prominent as the sense of political discrimination receded, prosperity gradually increased and a desire to escape secular and religious marginalization grew.⁵⁴

Interviewees also repeatedly emphasized that a deeper belief in Christianity had altered the naming of children:

Many people have become devoted to their religion. I cannot give my child a name such as Kabwimukya (grief will come again), Bonabana (all are children [the child was born with a defect]), Gafabusa (waste of energy) which are bad. Instead I would give a name such as Asiimwe (thanks be to God), Kyomuhendo (it is of significance), Kunihira (hope), Kugonza (God's love), Birungi (good). Those are the new names which we have today and are based on religion.⁵⁵

Catholicism had responded to the rise of born-again Protestantism in the 1930s by demanding a greater display of evangelical devotion and an acceptance of God's will:⁵⁶ ‘We are now changed people and we cannot go back to such names … We do everything in the name of Jesus. Why should we give names which reflect sadness?’⁵⁷ Early converts' perception of the deity was little changed from that of the indigenous religion, seeing God as a distant, non-interventionist Creator. But local priests claimed that as ‘people witnessed cures through prayer’, and parishioners were encouraged by catechists and clergy to see their God as a personal saviour, so belief in the deity's nearness and activism increased. The bereaved were pressured to see death as a positive transition and a temporary parting, with children achieving immediate respite in an afterlife which was very different from the unquiet other world of indigenous belief. The Christian afterlife had a more enduring and democratic quality, which particularly benefited the young. A short time on earth was no longer a spiritual disadvantage. Formerly, infants were marginalized in ancestor veneration, lacking a personality to be remembered, and never having had the chance to reproduce. But for Nyoro Catholics the innocence of baptized infants who died guaranteed direct access to heaven.

Yet the old beliefs, even where they were rejected, were not dismissed. The inheritance of what were perceived as negative names was strongly discouraged by catechists who still accepted the power of the living dead, who could ‘claim you, bring misfortune to you’ though the power of the name. The ancestors who insisted on being remembered and propitiated were even redefined by the most ardent Catholics as being of the devil.⁵⁸ It is important not to exaggerate the extent of this shift in consciousness. Many Nyoro Catholics still believe in the power of witchcraft and the ancestors, even while distancing themselves from them; the clustered nature of AIDS-related mortality caused some Catholics to see their God as not only activist but vengeful; and the new intimacy with the deity which has been demanded since the 1960s pushed some Catholics away from the church. But orthodoxy, at least in outward behaviour and rhetoric, has increased in Bunyoro, driven by catechists' demand for an exclusively Catholic identity.⁵⁹

It is striking that mortality-related names did not revive significantly when AIDS began to increase death rates again from the 1980s. Only names which referred to sibling mortality increased noticeably in the age of AIDS, from 0·76 per cent in 1980–4 to 2·46 per cent in 2000. Names referring to parental mortality or which predicted the death of the newborn declined in frequency. Bunyoro has never been regarded as one of the areas of Uganda most severely affected by HIV, but the disease was responsible nonetheless for a sharp rise in death rates.⁶⁰ Research in Hoima district in 1993 found that families focused most on the practical problems posed by AIDS, such as financial costs and loss of labour. But still for a tenth of households the most significant impact of the epidemic was the ‘social depression’ it had caused within the home.⁶¹ So why did AIDS not cause families to return to the naming patterns of the colonial period?

To some extent AIDS itself would have curtailed the fertility of infected parents,⁶² but this factor alone seems an insufficient explanation for this trend. Interviewees indicated that cultural change and longer-term demographic expansion had altered the understanding of mortality. Schooling, faith and the success of Uganda's HIV-awareness programme had reduced the sense that death was either inexplicable or the result of malice:⁶³

In those days people did not know the cause of death … That is why they could give such names as Baruzalire (this child is for death). AIDS is known and the cause is known, so people cannot give such names … Can you give your child a name related to AIDS! No. You cannot.⁶⁴

People are now enlightened. We know it is God who gives and it is God who takes away. We know that death is not a strange thing. Anybody can die at any age whether young or old, boy or girl; time comes and one has to die.⁶⁵

The permanence of the change in naming patterns also reflected a sense that the trauma of death varied with a family's demographic context. Interviewees were well aware that in the past Bunyoro had suffered higher death rates and much lower fertility rates compared to the present:

It was common to find infertile people in those days … today if they say so and so is barren you would think that the person is just not interested in having children therefore she is taking contraceptives.

In the past we were few. For example a lineage would have only like two men and one woman. But today you find a lineage with as many as ten men – so how do you name a child Kaijabahoire (the child was born when people are finished)? … The name would not carry meaning because the brothers and sisters of the late father are still there.⁶⁶

These trends in postcolonial naming illustrate why the incidence of mortality-related personal names cannot be taken as a proxy for death rates. Naming is too subjective and susceptible to external pressures to be used directly as a demographic source. It seems that many children were still named in a way which reflected ongoing problems of mortality, but social and perhaps clerical influence demanded that parents refer positively to the Creator rather than negatively to Death. The apparent decline in death-related names, in other words, in fact conceals a persisting, but increasingly indirect, concern with the vulnerability of the newborn. It is possible also that parents have come to realize that a name can affect a child's life chances, now that education and employment might take people away from the local community which would be likely to understand sympathetically the context of a descriptive name.⁶⁷

Names, mortality and fertility

The dataset of names then provides a unique insight into changing parental perceptions of death since 1900. It may be, however, that demographic historians will be most interested in what the dataset says about the relationship between mortality and fertility. Names which indicate fertility problems are particularly interesting because traditional demographic sources tend to focus only on average female fertility levels and infertility rates. Names such as Kyamusalire (finally we have a child) and Kanyama (the child was born after many fruitless attempts) give more of a sense of some couples' long struggle to conceive and give birth successfully. The fertility problems category constituted 1 per cent of all names during the colonial period, and only 0·3 per cent after independence, which fits well with census data which indicate that Nyoro fertility levels rose rapidly from the 1950s.⁶⁸ A handful of names, moreover, indicate that Nyoro in the past had some level of control over their fertility, or at least believed that they had. Bategeka, for example, means ‘the child was planned’, while names such as Libainura (Death forces us to stop having children) and Kasigirenda (this child has left the womb empty) suggest that a small number of parents intended to cease their efforts to reproduce.

Libainura is a very rare name in itself, but it may be an explicit, extreme expression of a broader relationship between parental perceptions of the likelihood of a child dying and their reproductive decision-making. Life histories in fact indicate that very many people responded to a history of miscarriages or infant deaths not by abandoning their attempts to reproduce but by persisting. One interviewee's parents, for example, produced seven stillborn babies before being cured of syphilis and having him, Byamaani (from strength).⁶⁹ But we should not assume that this was mere blind perseverance. Nyoro in the past were aware that inadequate birth spacing increased the risk of infant and child mortality, and that they could lengthen birth intervals through post-partum sexual abstinence and breast-feeding (which delays the resumption of menstruation). In interviews elderly Nyoro women reported that their birth intervals had ranged from 24 to 48 months, the longer intervals being associated with refraining from resuming sexual activity for up to 5 months after childbirth and breastfeeding for up to 3 years, unusually long periods compared to many other Ugandan societies.⁷⁰

It is possible that decisions about birth spacing were affected by a changing sense of the risk of infant and child death. Fig. 7 compares the trends in birth intervals with those in the group of names referring specifically to child mortality (incorporating the categories ‘child might die’, ‘child will die’ and ‘sibling mortality’) during the colonial period. There is not a perfect relationship, particularly from the 1940s when religious names increased dramatically in popularity. Nonetheless, the similarity between the two sets of data is suggestive. The graph seems to indicate that not only did fertility in Bunyoro begin to increase gradually during the Second World War, but that this increase may in part have been stimulated by growing parental confidence in the likelihood of child survival.

Fig. 7. The relationship between parental perceptions of mortality risk and birth spacing.⁷¹

One major weakness of the data on birth intervals is that parents waited so long on average between producing a child and baptizing it.⁷² This meant that many children would have died before they appeared in the baptismal register, thus artificially lengthening the recorded interval between two successive children. The decline in child mortality over time should have improved the accuracy of the baptismal registers, meaning that the apparent fall in birth intervals in this dataset might be linked to decreasing mortality levels, rather than parents' growing optimism about child survival. The names dataset, however, does not support this hypothesis. Parents who gave their babies names referring back to previous child loss had much shorter birth intervals than average, presumably because the death of an infant curtailed the period of breast-feeding and post-partum amenorrhea.⁷³ The relationship between parental perceptions of mortality risk and birth spacing decisions seems therefore to be of real significance.

CONCLUSION

David Schoenbrun's historical linguistic research has demonstrated the power of language as a pathway to the distant past.⁷⁴ Words are loaded with meaning, and perhaps never more so than at the moment when parents name the person they have created. In Bunyoro the giving of a name is a heavily ritualized time for reflection and anticipation, an opportunity to express deeply felt emotions. It would be interesting for this record of the changing trends in the selection of names to be compared with other African societies' experience, to see, for example, whether Bunyoro was unusual in giving so many children – 1 in 12 after 1900 – names that gave permanent voice to intimate conflicts with neighbours or family members. Comparative research might also consider why names associated with traditional spirits were displaced rapidly by those identifying the child with the Christian God in some societies but not others. For example, naming in Buganda did not experience a mid-twentieth-century Christianization and has maintained its predominantly clan-based nature and frequent reference to indigenous spirits, in marked contrast to western Uganda, Rwanda and northwest Tanzania. But then names in precolonial Buganda were not characterized by such extreme negativism as was expressed to the west and south.⁷⁵ It may be that this difference can simply be explained by Buganda's carefully maintained cultural nationalism, but several directions for future research can be suggested. Was the pacification of fearful spirits more prominent in the precolonial religion of the Cwezi–kubandwa complex than in that of Buganda's lubaale? Did the greater confidence of Ganda names reflect the absence of entrenched social tension here between pastoralists and cultivators? Was there less conflict between Christian and ethnic identities in Buganda than in other societies in the region?

New research might also return to the notion that bereavement is less harrowing for the poor in underdeveloped countries than for people in Western cultures or local elites, an idea stated most powerfully in Nancy Scheper-Hughes's Death without Weeping.⁷⁶ The frequency with which interviewees in Bunyoro became deeply upset on remembering the deaths of their children challenged this thesis, but the remarkable frequency with which Nyoro names referred to death during the colonial era gives a more comprehensive sense of the traumatic nature of population decline in the past. The case of Bunyoro, though, also shows that a society's perception of mortality can evolve quite rapidly, partly due to variations in fertility levels, partly due to larger cultural changes. In much of eastern and southern Africa the AIDS pandemic has brought unprecedented levels of mortality. For Nyoro, however, the current tragedy is limited by the knowledge that their society has suffered higher death rates in the past – at a time when the tragedy of loss was compounded by the additional trauma of sub-fertility – and also by a new confidence that death can be explained and, through a belief in a new kind of afterlife, more easily accepted.

Finally, the story of Nyoro naming should also cause some reconsideration of the causes of Africa's rapid modern increase in population. Early formulations of the demographic transition theory posited that “normal” non-contracepting populations were characterized by both high mortality and high fertility. When mortality fell due to improving healthcare and nutrition, then persisting high levels of fertility would bring about rapid population growth.⁷⁷ But societies like Bunyoro, which experienced both high mortality and low fertility, did not fit this model. It is now clear that in many modern populations fertility rose, initially at least, in response to declining mortality, in part due to shortened abstinence and breast-feeding.⁷⁸ This rise in fertility may indicate that African societies were more sensitive to changing levels of mortality than some recent researchers have suggested.⁷⁹ In Bunyoro, perceptions of mortality risk, as indicated by naming patterns, tracked the actual decline in mortality with a lag of five to ten years, while changes in fertility patterns took several more years to respond. Bunyoro may provide the earliest statistical evidence of a rise in fertility levels in tropical Africa, but overall these data support the arguments of those who believe that Africa's transition to rapid population growth in the interwar period was driven by an increase in life expectancy.⁸⁰