Kin in Daily Routines: Time Use and Childrearing in Rural South Africa

Introduction

It has long been established that models of child rearing function according to cultural norms and contextual needs (Levine, 1974; Harkness & Super, 1983; Ogbu, 1979). A critical component of any model is time use. Many studies on time use and childrearing have focused on biological parents in the US context (Bianchi & Robinson, 1997; Bianchi, Robinson, & Milkie, 2006; Hofferth, 2006). There is, however, a dearth in recent scholarship on time use that has incorporated extended kin in the African context. The role of extended kin in child rearing has been well established by scholars working in Africa (Caldwell & Caldwell, 1987; Fapohunda, 1988; Goody, 1982). Yet we know very little about how time is actually used by the various members who partake in some aspect of routine child rearing (Hewlett's (1991) study of paternal time investment among the Aka is one notable exception). In this paper, we address this gap by examining the interaction between young children and parents, extended kin and non-kin in a rural Black community in South Africa where Black families are responding to and resisting the social and economic transformation that is currently under way in the aftermath of apartheid. Using observational data from an ethnographic study, we examine 1) the composition of children's care networks; 2) the quantity and quality of time invested by members of the care network; and 3) the relationship between time use and children's behavior and hygiene. Our findings lend support to a “socially distributed” model of childrearing that operates in a context marked by distinctive forms of modernity and enduring traditional norms.

Background

Time Use and Models of Child Rearing

Time investment in children is influenced by norms about who should be doing what and for how long. The focus on biological parents in most western contexts is, to a large extent, driven by assumptions about the primacy of biological parenting. This could be justified by an evolutionary perspective that posits that biological parents have the greatest incentive to invest time in their children (Cox, 2008). From a demographic perspective, fertility decline in much of the world has been partly driven by a quantity-quality trade-off in children with time being a critical resource (Becker and Lewis, 1973). With smaller family size, parents have actually increased their time investment in their children (Gauthier et al., 2003). Even though most sociologists who have examined time use in the U.S. are neither biological nor demographic determinists, they, nonetheless, use a nuclear family as the normative model for examining time use patterns. Therefore, it is hardly surprising that all of the extant literature is limited to analysis of time use in child rearing by biological mothers and fathers. It also explains why so much of this literature has been limited to examining the effects of women's labor force participation and changes in family structure on only parental time use (Bianchi, 2000; Bianchi and Robinson, 1997; Hofferth and Sandberg, 2001). Interestingly, neither race nor class appears to play a prominent role in these findings (even though they are included as controls in models).

Ethnographic studies, on the other hand, present a different picture highlighting both class and race differences in models of childrearing. In her classic work on inequality and childrearing Lareau (2002) makes a distinction between a model of “concerted childrearing” that is followed by middle class parents, both Black and White, and a model of “natural growth” which is more characteristic of working class and poor families. A prominent feature of the “natural growth” model is the presence of extended kin in the daily lives of children. The critical role of kin in family life, particularly in poor Black communities, is the basis of Stack and Burton's concept of “kin work” (1993). “Kin work” refers to the distribution of time amongst kin for child rearing in order to keep the kin group functioning over the life course. It includes wage and non-wage work, intergenerational care of children or dependents, and family migration and migratory labor. Extended child-focused networks are maintained by parental figures who perform kin-work tasks “to regenerate families, maintain lifetime continuities, sustain intergenerational responsibilities, and reinforce shared values” (Stack and Burton, 1993:160). It is particularly critical in contexts marked by severe social and economic inequalities where particular sets of values and norms are privileged over others and where myriad structural inequalities greatly limit opportunities for social mobility. As such, the idea of kinship as a bulwark of resistance and a source of strength is consistent with the socially distributed model of childrearing found in many parts of Africa.

Socially Distributed Model of Childrearing in the African Context

The role of extended kin in childrearing has long been established by scholars of African family systems (Caldwell and Caldwell, 1987; Goody, 1982; Lloyd and Blanc, 1996). Riesman, in his classic ethnography of child rearing among the Fulani of Burkina Faso (1992), showed the importance of actively embedding young children into a kin group. Caldwell and Caldwell (1987) attribute high fertility in Africa to a culture of shared responsibility by kin around child rearing. It is also made evident by the high rates of child fosterage found across the continent (Bledsoe, Ewbank and Isiugo-Abanihe, 1988; Isiugo-Abanihe, 1985; Verhoef and Morelli, 2007). Fosterage occurs out of necessity when parents die or are very sick as in the case of South and southern Africa with its high HIV/AIDS prevalence (Ansell and van Blerk, 2004; Madhavan, 2004). It can also be initiated by biological parents who choose to send children away to live with other kin as a way to access better education, provide domestic labor and to become better integrated into the larger kin structure (Bledsoe, 1994; Goody, 1982). The socially distributed model of child rearing applies both to the life course as well as to the use of time in the daily tasks of raising and socializing children.

Serpell (1993) developed the idea of a “socially distributed model for socialization” to characterize the African system in which parents and other kin encourage children to be cooperative and interdependent rather than focus on individual achievement. This process of socialization teaches a child his/her place within a larger kin system with respect to gender differences, norms and values, boundaries for behavior, etc. It is a way to accrue social and cultural capital for the child and draw in adults other than biological parents to be actively engaged in child rearing. A socially distributed model emphasizes the need for children to become socialized in the appropriate cultural competencies that are necessary for successful development into adulthood (Harkness and Super, 1983; Levine, Levine, Dixon, and Richman, 1996; Ogbu, 1979). In the African context, this includes an emphasis on collective or communal well-being rather than a focus on individuals that is often emphasized in Western contexts. Nsamenang, for example, has shown that among the Nso' of Cameroon, it is the collective, not the individual, who are essential to a child's sense of identity, security, well-being (1992). He uses an agriculture metaphor to describe the “cultivation of a plant (child) in a field of kinsfolk” (1992). A notable recent addition to this literature is Parker Shipton's (2007) work on “entrustments” among the Luo of Kenya. He demonstrates that patterns of reciprocity and obligation extend to both material goods as well children.

Time investment in children is an important component of a socially distributed model of child rearing in many African contexts. Who does what for children and for how long not only quantifies the ideas that are discussed above but also brings to light both adaptation patterns to existing (often challenging) circumstances as well as the promulgation of cultural norms and values. In this analysis, we examine time use in child rearing in rural South Africa which, admittedly, is unique in a number of ways. One, the legacy of apartheid and high unemployment continues to relegate most Black people to the most disadvantaged position in the richest country on the continent; and two, there are large minorities of White, Indian, Colored and increasingly international migrants in the population. However, we believe that it is precisely these features that necessitate and enable rural Black families to follow the socially distributed model described above. In the section that follows, we discuss the conditions that have brought about both an economic imperative as well as an active resistance to hegemonic models of family functioning (Siqwana-Ndulo, 1998).

The South African Context

The legacy of apartheid has resulted in some unique employment and family formation patterns in both urban and rural Black communities. This, in turn, has conditioned how families use their time and their interactions with children. Black families in rural South Africa lived in “stretched”, dispersed and fluid forms (Spiegel, 1996) as a result of draconian apartheid era laws, such as the Group Areas Act, that circumscribed where Black people could live and forced men to migrate far away from families in order to work (Ramphele and Richter, 2006). As a result, Black children had very limited or sometimes no contact with fathers and other men (Murray, 1981). Moreover, men's inability to meet their obligations as breadwinners affected their interactions with their families, in particular, their children (Hunter, 2007). Given low marriage rates and high rates of out-of-wedlock childbearing for Black women, children grew up primarily in female centered families without marriage (Preston-Whyte, 1978).

While some of the most pernicious aspects of the apartheid system were dismantled in 1990, the process of transformation continues with, perhaps, the greatest challenge seen in employment and poverty alleviation. The post-apartheid context is marked by very high unemployment rates particularly for the Black population. Black unemployment rose from 19.6% in 1995 to 36.8% in 2002, compared to an increase from 3.3% to 6.2% for whites. More than half of the unemployed in the country are under the age of 30 (Altaian and Woolard, 2004). Female labor migration has been steadily increasing as women from rural areas, many of whom are mothers, move to small towns and cities in search of employment (Posel and Casale, 2003). Their children are more likely to stay behind most commonly in the mother's natal home when other women, sisters and grandmothers, assume the tasks of child rearing (authors blinded for review, forthcoming). In such a context, various kin members have always taken on particular parenting roles ranging from formal education to skills training to moral guidance (Mkhize, 2006). Therefore, it is common for Black children in rural areas to have regular, and sometimes, intensive daily contact with adults who are not their parents.

There is a large body of research on children's well-being in South Africa in the apartheid and post-apartheid context much of it, not surprisingly, focused on social inequalities and child well-being (Barbarin and Richter, 2001; Burgard, 2002; Carter and Maluccio, 2003; DuFlo, 2000). Additionally, the HTV/AIDS epidemic has produced a substantial literature on the well-being of children orphaned by AIDS (Case and Ardington, 2006; Cluver, Gardener, and Operario, 2007; Richter, 2004). The topic of time use, by comparison, has received very little attention. We located only one study that included South Africa in a cross national comparison of parental time investments. The authors used standardized time use surveys to compare time use patterns across industrialized countries and concluded that parents in South Africa spent the least amount of time with their children (Guryan, Hurst and Kearney, 2008). In the analysis presented here, we shift the lens to one community in rural South Africa and include care givers other than biological parents to observe the mundane routines that occur over the course of a day and week. It is this level of interaction, we argue, which provides the most proximate context in which to understand the conditions under which children thrive or are put at risk.

Data and Methods

Research Context

The study was carried out in 2002 in the Agincourt sub-district, located 500 km northeast of the capital, Johannesburg, in South Africa's Mpumalanga province. The Agincourt sub-region is typical of much of southern Africa in three important respects: 1) the land is insufficient to support the population through subsistence agriculture or other local activities; 2) there are very few local employment opportunities; and 3) the population has high levels of migration and mobility. This rural area was, under apartheid, part of the ‘homeland’ system that aimed to concentrate the Black population in areas with little infrastructure and poor land. The population of about 80,000 lives in 21 villages established through forced resettlement between 1920 and 1970. All villages have water provided through neighborhood taps and at least one primary school and most have electricity and a secondary school. Employment opportunities within the Agincourt villages are extremely limited for both men and women and usually include jobs that offer insufficient income. The most common types of formal sector employment are teaching, nursing and retail. The informal sector includes small-scale commerce of food stuffs and domestic work. The lack of viable employment in the area pushes people into temporary or permanent labor migration to the nearby game park, small towns and Johannesburg.

Child Rearing Practices of Tsonga

The patrilineal Tsonga live in multigenerational, extended family arrangements, known as stands, in which adult siblings live close to one another. People and food flow constantly between them, and labor, including child care and supervision is commonly shared or exchanged. In this sense, the socially distributed model of childrearing is appropriate to characterize child rearing practices of the Tsonga. It is also supported by historical and contemporary studies of Tsonga social life. Junod (1962) described how young boys would often spend the first years of life with grandparents who are portrayed as permissive and loving. Even when children were under the care of their own fathers, rarely did active disciplining begin until age 6 when children started assuming responsibilities such as herding. Junod's description suggests a carefree, happy early life for children under the careful watch of a range of kin. However, he does observe that maternal grandfathers were more lenient with grandsons from their daughters (compared to progeny from their sons) because their daughters were now married into other families. The role of grandparents continues to be important in child rearing as noted by Niehaus (2001) who describes the relationship between grandparents and grandchildren as free and loving with very little conflict because there is no financial obligation or any sexual competition.

While biological parents are expected to undertake the bulk of financial responsibility for children, older siblings, uncles and aunts play important role in child rearing. For example, fathers are not supposed to provide sexual counsel to their children, a role reserved for maternal uncles who also are also active in disciplining. This is not surprising given that the tie between uncles and nephews/nieces is often treated as a joking relationship with far less formality than parent-child relationships. Uncles also have substantial rights over their nephews and nieces and play a role later in life in bridewealth negotiations. If uncles have no biological children of their own, their nieces and nephews can inherit from them. Paternal aunts are expected to provide counsel to their brothers' children on sexual behavior, marriage and assist their nieces during childbirth. Sibling relationships tend to be very close with older siblings playing a role in children's lives uiroughout their lives. Niehaus (2001) documents how, at the time of marriage, elderly women used to give the husband's youngest sibling to the newly married couple with the expectation that for the rest of their married life, they would care for the child. The wife would wash the child's clothes while the husband would pay for school fees.

While such normative patterns continue, it is important to recognize the role of apartheid and more contemporary phenomena in altering traditional patterns of child care. Labor migrants, particularly female, often leave their children behind in the care of their families and make periodic visits making mothers' roles in their children's lives somewhat more tenuous in some cases. Paid child care is still more of an urban phenomenon in South Africa as described in Cock's (1980) classic study of domestic labor under apartheid and in Aly's (2010) recent study of Black domestic workers but pre-schools are increasingly in demand in our site as an alternative to kin-based child care. Young women (and some men) are exposed to other models of child care through education and media which often value nuclear over extended family in child rearing. The HIV/AIDS epidemic has necessitated families to develop coping mechanisms in response to illness and the death of young parents. In short, Tsonga communities are undergoing substantial change in child care practices. In this analysis, we examine how traditional patterns of child care and more contemporary forms (some forced) are reflected in time use patterns, something that has not been systematically examined in the extant ethnographic literature.

Data Description

The data for this analysis come from an ethnographic study on Children's Weil-Being and Social Connections (CWSC). The initial sample for the CWSC study was drawn from the Agincourt Health and Demographic Surveillance System (AHDSS) which includes annual censuses in 21 villages. Because the CWSC was designed to investigate, in detail, the wide range of social connections that link members of different residential households and their impact on children's well-being, the initial sample was limited to two villages, one above and one below the median level of access and services for the area. We selected two households that had at least one child aged 10 or 11 from each of three wealth stratum. Working with eight trained fieldworkers from the area over a period of 4 months, we mapped out the totality of connections between the initial household and members of all connected households. This was done through a two step process. First, we determined the “universe” of kin by developing detailed kinship diagrams with a range of respondents. While this group is, in theory, not closed, we were able to determine the point at which, there was a noticeable drop-off of knowledge about and contact with kin. Using these diagrams as the starting point, we followed an intensive, iterative process of identifying those households that were actually part of the functional network of kin. We arrived at these judgments through close and repeated observations of visitation patterns, movements of people and resources, in-depth interviews with members of various households, and field-worker judgment. Moreover, this process also enabled us to include non-kin connections such as neighbors who do not appear on kinship diagram but may feature prominently in functional support networks. The resulting twelve contact groups contained, on average, 6-8 households with a total of 349 children under the age of 22 in 2001 and 69 under the age of 6.

In the course of fieldwork, we collected a variety of data on social connection including kinship diagrams, residential histories, interview data and detailed daily observation records. Each instrument was designed to provide a different dimension of social connectivity in different scales of time ranging from the life course to daily activities. In this analysis, we draw on the observation records of children under the age of 6. We observed their daily activities and their interactions with parents, kin and non-kin over the course of a week. We limited this phase of data collection to pre-school age children in order to focus on home based and non-school activities. We conducted intensive training of fieldworkers for this phase of the project in order to standardize the recording of activities and interactions. We selected, at random, 2 children aged 3-5 from each contact group and had fieldworkers observe each child in 15 minute intervals from 7 AM until 8 PM on different days of the week. Due to scheduling issues, children were observed for different periods of time with a mean observation duration of 24 hours. We also recorded information on each person who came in contact with the child at any point during the observation period including age, gender, relationship to child and type of interaction coded as direct or passive.

Methods

The sample for this analysis is 24 focal children, of which 14 were girls and 10 were boys, and 335 individuals over the age of 10 who were observed at some point during the observation period. In this community, children over the age of 10 are expected to perform tasks related to childrearing. Our analysis proceeds in four steps. First, we describe the characteristics of the people who interact with the child and comprise what we term the “socially distributed care network.” Two, we use kinship diagrams to compare the composition of the care network and the total number of kin available to the child. Third, we calculate the quantity and quality of time spent with the focal children over the course of a day and the observation period by relationship to the person. Finally, we examine the relationship between selected features of care networks and time use and two outcomes of child rearing.

Conventional methods for studying time use rely on time diaries that contain a record of all activities undertaken over the past day or interviews that include a 24 hour recall (Lloyd, Grant and Ritchie, 2008). These methods have the distinct advantage of having standardized time periods. Because we wanted to allow the fieldworkers and families maximum flexibility in scheduling the observations and because our focus was on very young children, our strategy yielded variation in the time periods of observation. However, each child was observed at least once at every time period over the course of the day. Additionally, our method addresses the problem of social desirability bias that is common with time diaries (Sandberg and Hofferth, 2001). Even though our sample is small, we report our results in a conventional way—showing distribution of time spent on particular activities and for the people involved—in order to be comparable with extant studies of time use. However, inferential statistics would be inappropriate for the sample size; rather we use a “modified grounded theory” (Glaser and Strauss, 1977) approach to suggest possible relationships between kin involvement and children's outcomes which we give considerable attention to in our concluding remarks. We also include observations from field notes where appropriate to lend greater richness to the results.

Results

Who Interacts?

Table 1 presents the sample characteristics of the 24 groups of care networks (pertaining to the 24 focal children) who were observed at some point during the observation period. We separate grandparents from kin to consider the distinctive role of the elderly in care networks.

Table 1. Characteristics of Care Networks of Focal Children (N = 24).

Characteristic	N
Number of People in Care Network	12
Gender
Parity	3
Female Dominant	18
Male Dominant	3
Age
11-18 Dominant	. 11
19+ Dominant	13
Relationship to Focal Child
Mother	23
Father	9
Older Siblings
0	11
1-3	10
4+	3
Grandparent
0	6
1	13
2	5
Maternal kin other than grandparent
0	8
1-5	6
6+	10
Paternal kin other than grandparent
0	16
1+	8
Neighbor/Friend/Other
0	4
1-3	11
4+	9

Care networks are fairly large with a median of 12 people who interact with the child over the course of the observation period. The majority of focal children (18) have female dominant care networks and 11 of the 24 focal children have groups dominated by people aged 11-18. The distribution of kin relationships within networks is affected by survival and physical presence in the child's village. Not surprisingly, we find 23 out of the 24 mothers present in the observation period; the one absent mother works away from home. However, only 9 out of 24 fathers were observed. This is a result of labor migration (3), living elsewhere (7) and being out of contact with their children (5). More than half of the biological parents in this sample were not in a formal union at the time of observation.

Sixteen out of the 24 children have at least one maternal kin who was observed with 10 children having more than 6 different maternal kin in their groups. The 8 children who have no maternal kin observed are those whose mothers married into the village leaving their natal families elsewhere. This is further confirmed by the distribution of paternal kin where we find the same 8 children having at least 1 paternal kin interact with them over the observation period. Seventeen out of the 24 focal children have at least one grandparent who was observed and most of them are maternal grandparents. Out of the 7 children who have no grandparents included, one has no surviving grandparents, 3 have grandparents living in another village and 3 who have at least one grandparent living in the village but who did not interact with the children during the observation period. All 11 focal children who have no older siblings in their care networks have no older siblings over the age of 10. Finally, most care networks contain at least one non-kin member.

At this first step in the analysis, there is evidence of a wide range of kin and to a lesser extent, non-kin, who appear in the lives of young children over the course of a week. Many of those who did not appear during out observation period, namely fathers, are in contact with their children and even provide support for them (Madhavan et al., 2008). The validity of these patterns is strengthened by the fact that they are consistent with what we know about residence, employment and union formation in this community. Despite the presence of a large number of kin in care networks, it is not clear whether they represent all or most of the available kin who reside in the child's village. In the next section, we examine the extent of overlap between the total number of available kin (potential) and the number of kin who interact with the child during the course of our observation (functional).

Do All Kin Participate in Child Rearing?

Table 2 shows the relationship between the size of the total kin network residing in the child's village and the composition of the care network that was actually observed during our fieldwork. Data on kin networks come from kinship diagrams that we collected for each child. We only include kin older than 10 years of age in the total kin network to be consistent with the criteria applied to care networks.

Table 2. Relationship Between Size of Total Kin Network and Composition of Care Network (Mean).

Size of Kin Network in Child's village	Size of Total Care Network	Proportion of Care Network made up of Kin	Proportion of Care Network made up of Non-Kin	Proportion of Kin Network observed in Care Network
Under 10	12	.63	.37	1
10–20	11.6	.69	.31	.43
21–30	14.2	.78	.22	.45
31+	16.7	.78	.22	.30

In general, the size of the total care network increases as the size of the kin network increases. However, the total care network is made up of both kin and non-kin. The second column shows that the proportion of the care network made up of kin increases (and the proportion of non-kin decreases) as the size of the kin network increases. However, there appears to be a threshold kin network size after which the proportion of kin in care networks does not further increase. This suggests that a core group of kin take on daily child rearing responsibilities. The last column shows the extent of overlap between the total kin network and the care network. The smaller the kin network, the more likely that all of them will appear in the care network but there is a notable drop off in the proportion of kin network who also participate in the care network when the size of the kin network is large. Less than half of the care network is made up of kin when the kin network size is greater than 10 and further decreases to .3 when the kin group size is greater than 30. In sum, this table tells us that when there is only a small number of available kin, most or all of them take part in the care network but additional resources are there in the form of non-kin. Conversely, when there are a large number of available kin, less than half of them participate in any child rearing activities even though they constitute a larger proportion of the care network itself. This is best illustrated in the case of Nico, a 4 year old boy in our observation sample. Even though he is connected to six households in close proximity (all through paternal kin links), it became clear that only two of them, his grandmother's and his elder paternal uncle's, served as functional kin households. Fieldworker observations recorded people from these two households repeatedly providing services (disciplining) and resources (food) to Nico. To further understand the role of functional kin, we examine the quantity and quality of time investment in the next section.

Quantity and Quality of Time that People Spend with Children

Table 3 shows the mean number of hours that kin are physically in the presence of the child and the division of time spent on direct and passive interaction over the course of a day and over the total observation week. These averages are pooled estimates over all 24 focal children and across all days of observation. Observation could take place for any number of hours from 7 AM to 8 PM but averaged 6 hours per day. The daily average provides a sense of how much continuous time people spend with children whereas the weekly average tells us something about the frequency of interaction over a week. Direct interaction pertains to talking, playing, bathing, sleeping with, carrying, or any other form of close interaction whereas passive interaction covers situations in which the person is aware of the child's presence but is not directly engaging.

Table 3. Quantity and Quality of Time Spent in the Presence of Focal Children (Mean Hours).

	Total Time Spent		Time Spent on Direct Interaction		Time Spent on Passive Interaction
Relationship to Focal Child	Daily	Weekly	Daily	Weekly	Daily	Weekly
Mother	5 7	22 3	1 5	60	42	16 3
Father	4.7	15.8	.4	1.4	4.3	14.4
Older Sibling	5.4	19	.3	.9	5.1	18.0
Grandparent	5.5	17.9	.7	2.3	4.1	15.6
Maternal Kin	4.8	12.9	.4	1.1	4.3	11.8
Paternal Kin	3.8	8.9	.3	.7	3.5	8.2
Friend/Neighbor/Other	1.7	2.6	.3	.3	1.4	2.3

Mothers are a constant presence both during the day as well as the week given that both the daily and weekly means are close to the mean observation durations of 6 and 24 hours respectively. Fathers were observed to be in the physical presence of the child for about 4.7 hours during a day and almost 16 hours over the course of the observation period. There are not only fewer fathers around (Table 1) but those who are there are not in the presence of their children as long as mothers. This finding is consistent with other scholarship that has examined father-infant interactions (Hewlett 1991). Older siblings and grandparents spend the same amount of time over the course of a day but differ on the weekly average. Maternal kin spend nearly 5 hours on average over the course of a day and about 13 hours total over the course of the observation week whereas paternal kin spend 4 hours over the course of a day and only 9 hours over the week. Non-kin are physically present for 1.7 hours over the course of a day and 2.6 hours over the observation week.

Turning to quality of time use, much more time is spent on passive interaction across all relationships and within a day and week. In fact, only mothers spend more than 1 hour a day in direct interaction with children. Grandparents rank second followed by fathers and maternal kin. While children have a wide array of young and older adults physically present, most of this interaction occurs in a passive way. Put another way, it appears that children are constantly monitored by many eyes but little of it is in the form of direct interaction. The following vignettes describe a case comprised almost entirely of passive interaction (Fortunate) and one characterized by direct interaction (Promise).

Fortunate was 3 years old when we met him in 2002 and living in an extended family household with his biological parents, three older siblings, paternal grandparents, his father's sister (aunt), the aunt's son, the son's girlfriend and their 3 year old daughter. From early on, it was clear that his unemployed father makes only rare appearances during the day time preferring to spend his time elsewhere. There was also notable tension between Fortunate's mother, grandmother and his aunt. The fieldworker who observed him wrote the following in her notes during one of her observation periods:

Fortune ate soft porridge which was a left over from yesterday. Fortunate loves to be in the company of his brother, always wanting to be with him when he is around. He follows him wherever he is going, and sometimes he cries a lot when he is not home. There is hardly any interaction with his aunt's son or his wife or their daughter. Fortunate and their daughter are the same age but most of the time they are around, they do not play together.

Promise, 4 years old in 2002, lived with both her parents (though her father had worked in Johannesburg for many years until recently when he was laid off), and 2 older siblings. Although her household is a nuclear arrangement, she is only a short walk away from her three paternal uncles and paternal grandmother, each of whom has a separate house next to one another. In addition, there is a steady stream of neighbors and people from the village who come to the house to use the telephone service that her mother runs. Recorded in the fieldwprker's notes is the following:

She looks like a spoiled child. “Every time she demands something, she cries and they [her mother most often] gives it to her. During the afternoon she demanded Coke and her older brother was forced to run all over the village looking for it. When eating food, she doesn't eat porridge but eats chicken only and will always wash her hands after eating each and every piece of chicken. Her older brother is a hard working boy who does all the household chores including cooking and child care for Promise. While at the church Promise concentrated during the first hour while the members of the church were engaged in a choir because she liked to sing and dance. But'when the service started, she began demanding something and her mother was forced to always talk to her.

On the face of it, it appears that kids who live in large extended family arrangements (Fortunate) do not receive the concentrated attention given to children such as Promise who appear to have direct interaction with everyone she lives with, in particular, her mother and older brother. Indeed fieldworkers were troubled by the fact that Fortunate gets no attention from anyone. However, there are other factors that should be considered. One, Promise is in a markedly higher socioeconomic status than Fortunate, allowing those around her to engage with her in a way that is not possible in the first case. Two, family tension exists in both cases, the first within the co-residential household and the second, among adult siblings living in separate households who have differences over the sharing of cattle, the importance of church attendance and alcohol consumption.

Network Size, Time Use and Selected Outcomes

In order to assess the possible impact of the socially distributed model of childrearing on actual outcomes, we examined the relationship between compositional features of the care network and time investment by the different members with overall behavior and cleanliness. Fieldworkers were asked to make subjective assessments of various aspects of the child's personality and behavior using a series of questions with a ranking scale from 1 (low) to 5 (high). We created one composite index for behavior based on the ratings for “respectful”, “obedient”, “uses appropriate language” and “behave appropriately” and another for appearance based on ratings for “clean clothing” and “clean appearance.” Across the 24 children, the mean scores for behavior and cleanliness were 2.75 and 3.22 respectively. Table 4 presents the comparison in network size and time use across the extremes of the distribution, i.e., children who scored very high on behavior with those who scored poorly and those who were very clean compared to those who were very dirty.

Table 4. Comparison of Care Network Features and Time Use Between Children With Highest and Lowest Scores on Behavior and Appearance (Mean).

	Best Behavior(3.75+)	Worst Behavior (1.5 and below)	Best Appearance (4+)	Worst Appearance (2 and below)
Number of members in Care Network	9.8	12.7	14.8	12
Hours spent over a day by:
Mother	5.3	6.2	6.1	5.8
Father	3.7	4.2	6.0	5.3
Grandparents	5.0	6.3	5.4	5.8
Maternal Kin	3.4	5.3	6.4	4.5
Paternal Kin	0	5.4	4.5	4.8
Older Sibling	3.4	6.8	3.7	5.4
Non-kin	1.9	2.9	1.9	1.7
Proportion of Care Network made up of kin	.89	.81	.72	.88

Perhaps surprisingly, the children with the best behavior had fewer members in their care networks than those who were deemed to have poor behavior. Moreover, the average number of hours spent with the best behaved children is actually lower than with the worst behaved children across all relationship types. One explanation could be reverse causality such that difficult children require more kin as well as more time. The following excerpt from field notes about a 3 year old girl, Nelly, suggests a relationship between poor behavior and limited interaction with adults:

While her mother was busy cooking Nelly was grabbing things from the table and drawers, a bottle filled with salt fallen down and her mother tried to call her into order but she seemed not to hear anything. When she opened a drawer, her mother trapped her fingers which made her cry. Her mother did not help her remove her fingers telling her she wants her to feel the pain so that she will listen next time.” The fieldworker proceeded to note, “Once again, Nelly was there alone with her mother. There really are hardly any visitors, which means that Nelly ends up spending a lot of time only with her mother.

The children with the best behavior had, on average, a higher proportion of kin in their kin networks suggesting that kin may be able to exert greater social control than non-kin.

We find the opposite pattern with appearance. Children who were categorized as extremely clean and well-dressed had larger care networks than those who were deemed dirty and poorly dressed. The average hours spent daily by relationship type varies for these two sub groups of children. The high scoring children spend more time with their mothers, fathers, maternal kin and non kin compared to their low scoring counterparts. Conversely, low scoring children spend more time with grandparents, paternal kin and older siblings than high scoring children. Some of this can be explained by the allocation of responsibility for various child rearing activities. Parents, particularly mothers, and maternal kin, particularly female, are the ones who are expected to attend to hygiene and clothing needs of children. Therefore, it makes sense that those who scored the highest in this category spent more time with them than those who registered the lowest scores. Finally, we find that the children with the best appearance had a smaller proportion of their care networks made up of kin compared to their poorly dressed counterparts.

Discussion

Our goal in this paper was to demonstrate the critical role of extended kin and to a lesser extent, non-kin, in child rearing in a rural South African context by examining daily and weekly patterns of time use. Unlike time use studies conducted in the US and other Western contexts, this analysis is theoretically motivated by a model of childrearing not anchored by biological parenting. Our findings provide support for the socially distributed model of childbearing (Serpell 1993) in five ways. One, most young children in this context are in daily contact with a wide range of kin and non-kin but many do not have fathers as part of their regular care network. Two, the larger the size of the available kin network, the larger the proportion of kin in the care network. However, not all available kin take part in care networks suggesting that a core network of kin actively takes up child rearing responsibilities. Three, while mothers make the greatest investment of time in young children as they have been shown to do in time use studies in the US (Bianchi 2000), they are supported in this effort first by grandparents and maternal kin followed by older siblings and paternal kin. Non-kin do not appear to play a very significant role in daily child care. Four, much of the time investment occurs through passive rather than direct interaction. Finally, there appears to be a relationship between the composition of care networks and certain child outcomes though not in the expected direction.

The value of this analysis can be appreciated for the South African context but also for other contexts marked by social inequalities including the US. Whereas a substantial amount of scholarship has developed on various aspects of South African social life, i.e., HIV/AIDS, teenage pregnancy, education, crime, etc., far less attention has been paid to the practice of everyday life which offers, arguably, the most useful lens through which to understand human thought and behavior (Caughey, 1982). An analysis of the patterns of time use by young and older adults in child rearing improves our understanding of rural Black families in South Africa in at least three ways. One, it gives us an idea of how people are coping with high unemployment which disproportionately affects the parents of the focal children in our analysis. Those who were not observed in our study – particularly fathers – were either working or looking for employment away from their homes. The separation of the sites of production and reproduction, which was institutionalized under apartheid and continues in the present, conditions who spends how much time with children. Two, the continuing importance of kin, particularly grandparents, uncles and older siblings, in child rearing can be seen as a form of resistance to hegemonic norms of family functioning based on white, European models. Under apartheid, Black families, because of their preference and reliance on extended kin, were seen as dysfunctional and disorganized compared to White nuclear families or having deficient culture (Siqwana-Ndulo 1998). In the post-apartheid climate, Black families are attempting to create conditions for family life that retain the strengths of' extended family involvement to ensure culturally appropriate socialization of children but also offer space for young people to seek new opportunities in education, employment and union formation. Three, although we have mostly talked about the negative consequences of separating families during the apartheid and post-apartheid periods, it is possible to interpret the composition of child care networks and time use patterns noted in this analysis in a more liberating narrative. It may be that men, women, young and old are responding to a new freedom of movement that allows people to live wherever they want in order to take advantage of new educational and employment opportunities. This would, in turn, necessitate that the involvement of a range of kin in different ways to ensure the well-being children through patterns of care substitution.

Beyond its value as an indicator of coping, adaptation and resistance, time use patterns also have an impact on children's well-being. Based on the exploratory analysis that we conducted on two outcomes (behavior and appearance), there are several useful insights that could be further developed into formal hypothesis testing with larger, more representative samples. Our analysis suggests that poorly behaved children have larger care networks than their well behaved counterparts. More rigorous testing is needed to determine whether this is true and why. One might hypothesize that children who have more people who spend time with them are the ones who need more “concerted” child rearing. This is clearly a different use of Lareau's (2002) original conceptualization but one which might better reflect the conditions under which low-income families function. Alternatively, it might be more important to have sustained time with certain people such as grandparents, whose presence has been shown to have a positive influence on children (Mkhize 2006; Parker and Short 2009; Townsend et al., 2002). The very different patterns for behavior and cleanliness underscore the need to examine effects on a varied set of outcome measures.

We also showed that mothers invest the most time in children. One important question to ask would be whether there is a minimum amount of time needed from mothers for various outcomes. Related to this is the extent of substitution that occurs by other kin. In other words, do children do better when several people are attending to them at the same time or when particular people are substituting for mothers? In terms of quality of interaction, we found that much of the time investment even by mothers is in the form of passive interaction. Whereas there exists a copious quantity of both scholarly research as well as parenting advice in western contexts and in South Africa that emphasizes the need for direct engagement with young children, it is not clear how much (or little) is needed for particular outcomes nor is it evident whether a socially distributed model offers the benefits found in Lareau's “concerted childrearing” model which is implicitly premised on focused parental engagement. We have argued that time use in a rural South African context follows a different model of childrearing which, in itself, may be beneficial to children. However, there is little doubt that particular patterns of time investment by kin could also lead to negative outcomes just as can happen with nuclear family models.

These next steps would require data that are sufficiently large and representative. Much of the research in the US has been based on data from time use diaries collected over various periods of time, the American Time Use Survey (ATUS), and modules attached to the Panel Study of Income Dynamics and the Time Use Longitudinal Panel Study. The only existing data for South Africa that we are aware of come from the Multinational Time Use Study (MTUS) which examined child care trends in 16 countries from 1960-2000. However, these data are restricted to parental time investment in the household. There are several large scale data collection efforts currently under way in both rural and urban contexts in South Africa that could accommodate a detailed time use module incorporating parents, kin and non-kin. Therefore, the additional cost of including a detailed time use component, while not inconsequential, should be relatively feasible. We would advocate for a mixed data collection approach that would include both time use diaries as well as observation of interactions to improve validity. The resulting data would have important pay offs for both research and policy driven agendas related to the larger issue of child well-being. Finally, we would also support the expansion of time use data collection in the US and in other comparative efforts from its current focus on households and nuclear families to incorporate extended kin and if appropriate, non-kin. This appears particularly useful during an economic recession when reciprocity and living arrangements between family members is likely to be changing globally.