Abstract
Survey data show that most Tanzanian women find wife-beating justifiable. What is the meaning of the violence that enjoys such broad social approval? Does respect for women’s agency invalidate feminist opposition to wife-beating? I explore these questions by analyzing data on hegemonic norms generated through 27 focus group discussions in Arumeru and Kigoma-Vijijini districts, and find that wife-beating was supported for its role in constituting social order. This analysis of how exactly violence can constitute order yielded insights into the interplay between violence and consent that are theoretically relevant to violence against women in other forms and contexts, reminding researchers and practitioners of the role of power and coercion in supposedly agreed-upon community norms.
In his influential “Critique of Humanitarian Reason,” anthropologist Didier Fassin encourages critical social scientists to pay attention to what is lost when we use terms of suffering to speak of injustice, of misfortune to speak of domination, or of ill-health to speak of violence, as we increasingly do in the current “moral economy” (Fassin, 2011, p. 8). While it was as a sociopolitical problem that violence against women (VAW) was placed on the international agenda in 1995 (in the form of the Beijing Platform For Action), 1 it is indeed as a health problem that it has received increasing attention since then (Jakobsen, 2014). This article sheds light on one dimension that is lost when we frame VAW in terms of the suffering it causes, namely how violence contributes to a particular social order.
This article explores the sociopolitical dimension of domestic violence (DV) that emerges from qualitative data on how it is justified in Tanzania. The latest Demographic and Health Survey found high levels of support for the DV known as “wife-beating” in Tanzania, particularly among women (National Bureau of Statistics, 2011). This raises questions the much-cited survey cannot answer. Why do so many people in a society characterized neither by generalized violence nor by tolerance of it support DV? And how do we interpret women’s approval of VAW? How can we approach a practice that most women support as violence against women?
I answer these questions by examining data on the locally embedded meanings of wife-beating for resonance with Sumner’s theory of censure, Gramscian ideas of hegemony, sociological theories on violence and order, and feminist theories of DV. After a brief background on the main concepts and on the Tanzanian context, I describe how I generated and analyzed the data. The findings I present on dominant norms supporting wife-beating in Tanzania show that supposedly agreed-upon community norms actually interlink with coercion, that power is deeply implicated in the choice to conform to a norm, and that violence orders society even when it is not used.
Violence and Social Order
That men’s violence against women maintains a certain social order is no new discovery. This ordering nature of the violence was at the core of the feminist research that first achieved recognition of VAW as a social problem in Tanzania as well as in the United States and the United Kingdom (Brownmiller, 1975; Dobash & Dobash, 1979; Hanmer, 1978; Hanmer & Saunders, 1984; Sheikh, 2002). Also in the international community, the international women’s movement placed VAW on the agenda as a social problem not so much in terms of suffering, ill-health, and misfortune, as in terms of how it orders society. For example, VAW was included in the Beijing Platform for Action as an obstacle to equality, and both a cause and consequence of inequality (§112). In this cornerstone document, violence is a problem not because it causes ill-health, but because it “is one of the crucial social mechanisms by which women are forced into a subordinate position” (§117).
However, as more powerful actors incorporated VAW into their agendas, VAW research was increasingly instrumentalized toward the goals of those agendas, rather than the social scientific goal of understanding VAW as a social and political phenomenon (Bumiller, 2010). Practical consequences of the inattention to the ordering nature of VAW are evident in interventions that, ironically, increase the controlling of women (Bumiller, 2010) and re-enforce traditional gender orders (Stanko, 1996). This article builds on, and seeks to contribute to, efforts to restore the sociopolitical to our understanding of VAW. Evan Stark, for instance, calls for “a more accurate cartography of abuse” which recognizes the sociopolitical harms of DV as surpassing its physical and psychological aspects (Stark, 2009, p. 1510). Conceptualizing DV as part of a pattern of “coercive control” rather than as discrete physical acts, he claims, would be more in tune with how participants themselves perceive it (Stark, 2007).
In Tanzania, the recent mushrooming of research on VAW coincides with its incorporation into international aid agencies’ agendas. The development sector dominates social research in sub-Saharan Africa (Arnfred, 2004; Lewis, 2008; Mama, 2011), and as an “anti-politics machine” (Ferguson, 1990) it exemplifies the depoliticizing “humanitarian reasoning” Fassin (2011) describes. Gender issues have proven particularly prone to depoliticization as they enter aid agencies’ agendas (Arnfred, 2004; Cornwall, Harrison, & Whitehead, 2007; Lewis, 2003; Mama, 2011). While applied research has documented the prevalence of VAW and its correlation to health and development indicators, readers interested in social scientific questions of how VAW contributes to a particular sociopolitical order in Tanzania are left piecing together clues from historians’ accounts of how VAW ordered Tanzanian society in times past (Ivaska, 2002; Lovett, 1996). Real-world consequences of this inattention to the ordering nature of VAW in contemporary Tanzania include interventions where making victims obey perpetrators is considered success (Jakobsen, 2015, p. 16).
A closer look at how violence relates to social order can sharpen our understanding of social order as well as of violence. By social order, I mean how people in a given society are positioned in relation to one another in terms of power, with hierarchy exemplifying a particular form of social order. An increasing number of sociologists are calling for violence to be integrated into social theory in ways that recognize its significance both in maintaining social order and in the very constitution of the social (Hearn, 2013; Ray, 2011; Walby, 2009, 2013). They challenge the assumption that while state violence may maintain order, individuals’ violence is the opposite of ordering: peripheral or exceptional to the social order (Ray, 2011, p. 2; Walby, 2013, p. 97). They refer to VAW research to illustrate the empirical inaccuracy of this assumption, as do criminologists challenging the long-assumed association of violent crime with disorder, disruption, and deviance from social norms (DeKeseredy & Schwartz, 2009). I build on and contribute to these challenges.
How can we study the role of violence in social order? Walby’s (2009, pp. 193-198; 2013, pp. 103-104) ontology of violence casts violence as a form of power in its own right: a distinctive practice interconnected with other practices. In her ontology, violence may tell us something about who already has power in a given society, but violence also contributes to those power relations. Thus, we need to examine not only how violence reflects social order, but also how violence constitutes social order (Walby, 2009, pp. 191-193), making and reproducing structures of inequality (Walby, 2009, p. 193; 2013, p. 104). Examining how violence constitutes social order requires us to look not merely at how violence is deployed, but also how it is regulated. Similarly, Ray (2011) recommends that investigators of the violence–order nexus look to the discourses that authorize violence (p. 14).
Examining how people discuss the rights and wrongs of violence, then, seems a fitting inroad into understanding how that violence orders society. Research on VAW in particular suggests that the discursive dimension of this violence is key to understanding the violence–order nexus. Empirical studies of the relation between social order and VAW have focused mainly on the chicken-and-egg conundrum of what-causes-what between VAW and gender inequality (Hunnicutt, 2009; Mogford, 2011). Is violence facilitated by women’s subordination? Or is it provoked by their insubordination? Does a more equal gender order lead to more violence against women, or less? Statistical analyses of the relationship between variables representing women’s status and VAW have yielded inconclusive and conflicting results. In her review of these studies, Hunnicutt (2009) suggests the conundrum will remain unsolved unless the ideological dimension is taken into account. This study examines the violence–order nexus on this discursive or ideological level, rather than investigating it as a cause-or-effect question.
Deviance, Control, and Censure
I will present Tanzanian data on the ordering dimension of VAW using Sumner’s (2012) reconceptualization of the twin concepts “social control” and “deviance.” His reworking of both as elements of “censure” breaks down the presumed division between the formal practices of punishment associated with state actors’ “social control” and individuals’ practices of violence often studied as “deviance” (Sumner, 1994). “There is a latent violence in censure and a latent censure in violence,” he claims, as violence is inherently the censure of perceived offense (Sumner, 1997, p. 3). Data on which and whose violence is deemed legitimate (such as the data which are presented here) reflect “certain cultural ideologies of domination” (Sumner, 1997, p. 4). Studying deviance as Sumner proposes, as no more than “deviation from a dominant moral code” (Amatrudo, 2009, p. 87), allows me to analyze talk about deviance for such norms of domination.
While discredited approaches to deviance and social control failed to acknowledge the power dimension implicit in the concepts, Sumner’s point that “who censures, wins” has led researchers to use these twin concepts to throw light on “who rules”(Marshall, Douglas, & McDonnell, 2007). I use the terms “deviance” and “social control” as elements within Sumner’s sociology of censure to reveal the power relations implicated in conforming to the norm. Rather than assume that rules are agreed upon with full and free consent, I use them to examine how supposedly agreed-upon community norms actually interlink with coercion.
Consent and Coercion in Hegemony
The case of a practice being approved of by the same women it is seen as abusing is not unique to wife-beating in Tanzania. It is contested in relation to feminist engagement with the Global South in general, as well as with other practices, such as prostitution and genital mutilation. Can a practice women approve of constitute violence against women? Is it coercion when the “victims” themselves approve? Such questions posit consent as counter-evidence of coercion. He who consents is not coerced, it is assumed, and vice versa: It is she who fails to procure consent who proceeds by force. This idea that coercion and consent are mutually exclusive, like victimhood and agency, colors the scholarly debates of women’s support for practices that undermine women’s collective interests. The dichotomy dictates that women’s consent be interpreted as invalidating feminist understandings of that practice as coercive. 2 I will analyze the data for resonance with Gramsci’s understanding of consent and coercion (Gramsci et al., 1971). He perceives them as not only indivisible but mutually porous, combining to form hegemony. According to his “dual perspective,” coercive apparati play a consensual role in that the threat of coercion helps to achieve consent (Jones, 2007).
Wife-Beating and Social Order in Tanzania
Economists Warner and Campbell (2000) describe Tanzania as “an agrarian economy with non-symmetric gender relations,” pointing out that husbands contribute less labor to the household economy than wives, and take more out of the household economy for private consumption. The majority of Tanzanian women (88%) qualify as unpaid family workers in “implicit intra-household rent agreements” by which “men extract the fruits of the women’s labour time” (Warner & Campbell, 2000, p. 1330).
Such domestic subservience is still the norm for wives (Fischer, 2014), even in the capital (Krishnan et al., 2012). Three in four Tanzanians live off small-scale, rain-fed farming and informal business in rural areas. Almost all adult women participate in the formal and informal labor force, mainly in agriculture, while also contributing most of the unpaid household labor (National Bureau of Statistics, 2011).
“Wife-beating” is neither new nor extraordinary in Tanzania (Ivaska, 2002; Lovett, 1996; Sheikh, 2002). The World Health Organization (WHO) estimates 41% of women in the capital aged 15-49 to have experienced partner violence in their lifetime, and 56% in their only other location, Mbeya. More than half of these women experienced it less than a year ago (Garcia-Moreno, Jansen, Ellsberg, Heise, & Watts, 2005). Tanzania’s National Bureau of Statistics (2011) estimates that 40% of Tanzanian women of this age group have experienced physical partner violence.
Gender inequality and VAW rank high on the official government agenda. For example, Tanzania’s Development Vision 2025 aims to attain “gender equality and the empowerment of women in all socioeconomic and political relations and cultures” (Planning Commission, 1999), and the Poverty Reduction Strategic Plan lists violence against women as a poverty indicator (Betron, 2008). Tanzanian women’s organizations, however, have formally alerted the international community that these policies have not translated into positive outcomes in women’s lived realities, and that the official policy on gender equality and gender violence is not implemented (Women’s Legal Aid Centre, 2008). Formal support services for GBV are near non-existent outside the capital (McCleary-Sills et al., 2013), and the government rejects demands for violence shelters in the name of family harmony (Women’s Legal Aid Centre, 2008, pp. 23-24). There is no specific law protecting women from DV, and the review of the legal system to remedy this has been pending for a decade (McCleary-Sills et al., 2013). Less than 3% of DV cases are reported to the police, and even fewer to health services (Palermo, Bleck, & Peterman, 2014).
A major reason why Tanzanian legal advocates want a specific law on DV is to make a political statement that wife-beating is wrong (Betron, 2008). This reminds us that VAW is widely condoned, accepted, and expected as part of intimate relations in Tanzania (Krishnan et al., 2012; McCleary-Sills et al., 2013; National Bureau of Statistics, 2011). The aim of this study goes beyond establishing this fact, to better understanding it. The overarching research question is What is the meaning of the violence that enjoys such broad social approval in Tanzania? The answers the data give to this question will then be interrogated with two further questions: What does this tell us about the relationship between VAW and social order? What does this tell us about the meaning of women’s approval of VAW?
Method
Generating Data on Hegemonic Norms 3
I generated data through 27 focus group discussions (FGDs) over the course of 6 months in two disparate districts in Tanzania: Arumeru and Kigoma Rural. Arumeru district forms part of the region typically researched by social scientists, and Kigoma the least. The two regions also find themselves at opposite ends of the national spectrum regarding socioeconomic indicators and exposure to “modernizing” influences such as tourism, agribusiness, and NGOs.
I chose the FGD method because I wanted to generate data on what community residents could say to one another. FGDs do not generate individual opinions or reliable factual descriptions, because what people say in them is influenced by what others in the group may think, especially more powerful others. Precisely because these dynamics shape what is said in FGDs, the method is ideal for generating data on what participants believe to be shared and dominant community norms.
I tweaked the method to ensure it did achieve rigorous data on dominant social norms in the Tanzanian context. The quality of my data, like that of most data generated in lower-income Africa, was threatened by the steep power gradient between researcher and researched. To address this, as well as ethical concerns about positionality and alterity, I spent 13 FGDs experimenting with different ways of facilitating the discussion before arriving at the facilitation style that produced the discussions presented here (Jakobsen, 2012).
Each group consisted of either men or women, of ages either above or below 40. The youngest was 19, and the oldest around 70. In addition, some groups were also homogenous in terms of either religion or ethnicity or socioeconomic status. I presented a conversation starter to the group in Swahili, and then left the participants to discuss with one another. They did so, in Swahili interspersed with Kiha and Kimeru, knowing their voices were being recorded, and when they were done with one conversation starter, they called me back for the next. Native speakers assisted me in transcribing and translating the recorded discussions. The conversation starters consisted of vignettes, role-plays, and debates fleshing out the following questions:
A younger woman comes to tell you she’s had enough of being beaten. She wants it to stop. What would you say?
A man tells you he beat his wife yesterday. What might he tell you for you to chastise him about it? What might he tell you, for you to say he was right?
A man says he’ll never beat his wife. What do you think of this man?
Do you think women sometimes get beaten without deserving it?
What should one do if one hears one’s neighbor beating his wife?
In accordance with the procedures set out in the terms of my research permit from the Tanzanian Commission on Science and Technology, I asked research permission at regional, district, and village levels before recruiting the participants at the village level. In addition to ensuring the formal requirements of informed consent, I created opportunities for any participants who might have felt pressured to give me their consent to opt out. The fact that no researcher can guarantee that FGD participants will respect the confidentiality of what others say in the discussion (Barbour & Kitzinger, 1999) had been among the ethical concerns that made me fine-tune the FGD method. Some of my tweaking, then, was aimed at steering the discussion clear of any references to actual events or people the participants knew about. Thus, although I still cannot guarantee that participants did not cause harm after the discussions by passing on information from it, I reduced the likelihood that the FGDs would produce any such “juicy gossip” worth sharing.
Analysis: Constructing “The Good Beating” From the Data
Having generated data on hegemonic norms, I analyzed these data using a construct I derived from them. I assembled this by looking through the transcripts for the respondents’ own logic on wife-beating, identifying themes that recurred across religious, ethnic, socioeconomic, age, and sex differences. I then examined these themes iteratively with theory through analytic induction. I treated what people chose to say to one another as reflecting the speakers’ judgment of how various arguments would be received in the group (Jakobsen, 2012). In this sense, I biased my analysis toward what speakers reckoned to be hegemonic norms and discourses (Jakobsen, 2012). I treated the totality of these as making up a hegemonic ideal, an ideal I call “the good beating.” I used the “good beating” construct to group findings that spoke to the beating(s) that participants refer to across all discussions as being widely supported. To present an accurate picture of these findings, I will first describe how they came to be grouped as representing the good beating. In part, this involved constructing two counter-categories to the good beating: the objectionable beating and the indefensible beating.
Participants did not discuss all wife-beating as justifiable. Certain types of beating were found objectionable, in that they were objected to in all groups. For example, beating that resulted in injuries needing medical attention was objected to in all groups as “ukatili”—cruelty. Beating a wife in a fit of anger, or without checking whether she actually “deserved” it, was also objectionable. For example, in a women’s group, a discussion of how a certain beating was justified changed tack after this factor was introduced: Amina
3
: But what if it wasn’t your fault that the dinner was late? Let’s say, your neighbor had an accident, she called for help. You were late, yes. But late because you did the right thing.
That’s something else!
He needs to ask first, to check: why is it late: Is it your fault?
If he’s one of these men, who just comes home, in a bad mood, and BAM! Without any thinking. He doesn’t care what happened. He is just angry, and you are there.
Yes! Those guys-!
But then it’s your duty, isn’t it? The dinner.
As in this exchange, an objectionable beating was not one that everyone objected to, but one that it was possible to object to. Other forms of beating were not only objectionable but indefensible, in that nobody in any of the groups made any serious effort to defend them. For instance, no participant made any effort to defend a husband who beat his wife out of sheer drunkenness. When one participant found that his role as devil’s advocate committed him to defending such a case, he expected that no argument he could make would be taken seriously by his peers. For them to ask him to defend it, and for him to try to defend it, was a joke, as they all knew it was indefensible:
But in the case of a guy who just comes home his head full of pombe [home brew], ready to fight, that (beating her) is just silliness.
What do you say to that, Siliani? Can this too be ok?
(chuckling) Argh, you caught me! Hold on . . .
Drunk! Even the goat can smell it!
Yes! But then, you know, why did he drink? Maybe this wife was so cruel he had no choice (laughter)
She forced him to drink and forced him to beat her?
(laughing) Ok, I didn’t say it’s ALWAYS ok to beat . . .
In contrast to these objectionable and even indefensible beatings, which were condemned as mere cruelty [ukatili], a certain type of beating was referred to across all discussions as a widely supported one. Participants who opposed this type of beating did not expect their objections to succeed. They did not support these objections with dominant discourse, nor did they expect a beaten wife could. As participants argued for and against beating, their discussion left an outline of what this certain type of beating, that they presumed most others to consider a “good” beating, was. I analyzed this outline of the ideal they presumed to prevail in the group as the conventional wisdom which Gramsci says constitutes cultural hegemony: “common sense” in that it is common to the social group (as described by Jones, 2007). As the outline was not identical in each FGD, I focus only on the parts of it that recurred across all FGDs. I treat these characteristics as forming a common core of a beating participants believe to be widely approved of as good and necessary. In other words, I treat them as forming the hegemonic ideal which I call the Good Beating.
Findings
A Good Beating is Reasoned and Reasonable
The good beating was different from how husbands beat their wives in the past. Especially, men distanced themselves and the good beating from their forefathers’ practices. You could not be as quick to beat a wife as husbands in “times of dictatorship” could:
It’s not like, you know, in centuries past. Our ancestors would just force things right away. Like if he says something, it already has to be as he says, she can’t answer anything back. Nowadays it’s been mixed up with modern practice, generations are different. You’ll find a husband and wife like managers together. When he tells her to do something, she can answer, they can discuss. A man may even tell his wife to do something, and she’ll say no.
It’s true. Our grandfathers-
The times of dictatorship are over. We still beat them, but—for faults.
And not straightaway like that. Like if the food isn’t ready. Like you get home and you’re expecting the food to be ladled onto plates. Say it should be ready at six. And you get there and she’s not even started. You look first: is there nothing to make it with? But there is! You see everything’s there. And then she does this not one time, but three or four times, she persists. Or let’s say she goes out in the middle of the night and you find her drunk. And you tell her the children haven’t eaten, and she answers back. She does this several times—of course then you need to-?
-Flog her. 4
This group contrasts the beating they can justify with their forefathers’ “dictatorial” ones by presenting theirs as reasoned and reasonable. A husband must reason with his wife (Mohamed), have reason to beat her (Karim), and be reasonable in ascertaining that the reason truly is a good one (Ricardo). This does not show how beating has actually changed over time, but rather on what basis beating is presented as good, and differentiated from “bad” beating. This in turn reveals much about the legitimated social role of violence in constituting power. Participants across discussions referred to reason when differentiating good beatings from bad. For instance, “the tendency to beat without reason”, according to many women, would reflect badly on a man. A good beating was to a bad one as rational government was to rash and arbitrary dictatorship. Legitimate violence was rational: not because violence should not be used to govern, but precisely for more rational government. Men should govern, and if they governed rationally, they could legitimately use violence to govern.
A Good Beating Steers a Wife’s Choices
Participants also talked about the legitimating goals of beating in terms of a wife’s governability. In all discussions, the purpose of a good beating was to effect behavioral change. A husband must be willing to beat his wife, in the words of one respondent, “because you need to stop her mistakes, change her behavior. That’s why.”
You beat her with sticks that you want her to be changed.
You could just tell her to change, of course—but would she listen?
In saying “beat her that,” Jonas is replacing “tell” in “tell her that” with “beat.” This semantically integrates the purpose of beating into the doing of it: to beat is to give the beating’s message. Marko’s response shows this was how he interpreted Jonas, that is, as saying the beating was a way to communicate that the wife’s behavior must change. Communicating this demand through violence makes it more difficult to ignore.
The behavior a good beating aimed for was for a wife to “listen”: to be attentive to what her husband wanted, and responsive to his direction. All groups discussed how a wife should be an “mtii”—an obedient tractable person. If she had “kiburi”—if she was proud and headstrong—she needed to be beaten to behave with “heshima”—respectful obedience. Men described the hardships a family suffered when the mother had “kiburi,” and how, if only a wife was “mtii,” a man could be the most quarrelsome drunkard, and still live peacefully with his family. Life was much easier with an “mtii” wife.
“But would she listen?” This pressing concern underpinned men’s discussions. What did it take to make a wife listen? “A wife can be so stubborn, you’d think she was deaf.” If it made her listen and do as she was told, then beating her was good:
When you beat her rightly, you don’t just order her and then when she doesn’t do it, you beat her straightaway. No. It’s when you order her today, tomorrow, and the day after.
If you leave her that way without beating her, she’ll continue in her faults. Like she won’t wash the children, or she doesn’t cook when she should, you know, so then you’d have to beat her.
You may not even want to beat her. But then she may have a mouth on her. Maybe you tell her to do something, you come back, and she hasn’t done it, and instead of apologizing she asks you where you’ve been. Then you just have to beat her. But if she just knows you may beat her, you can just tell her to work, and she will: she’ll keep quiet, and you won’t have to beat her.
A strategically administered beating could make a wife more governable. Samuel’s final sentence (“if she just knows you may beat her”) spells out how the intended effect of such a beating went beyond the actual incidence of beating, influencing a wife’s decisions also when she was not beaten, encouraging her to censor her behavior to avoid censure. In addition to targeting the specific mistake she was beaten for, the beating that was expected to be widely approved also made a wife responsive to future demands by the mere knowledge that she could be beaten. This was how it made an mtii of her.
A Good Beating Makes Wives “Choose” Obedience
Steering a wife’s choices, then, was a beating’s legitimizing goal. This good beating could only succeed, however, insofar as also women approved of its goal. The fact that this norm was shared by both men and women raises the issue of how women comply in their own subjugation. Women responded to the threat of the good beating by “choosing” to obey, and urging others to do so, because they accepted that beating a wife to make her behave was a good beating.
In discussions, as women approved of the good beating, they disapproved of the wives it targeted. They talked about how women who got beaten more often, had “kiburi,” and must learn to be “mtii.” They described how they could guide and help such a woman please her husband, not only for her own good, but also because it was right and proper.
Women also discussed how a woman should adjust her behavior in response to a beating. This can be seen in the following discussion of a vignette where a younger woman came for advice, saying her husband had beaten her four times in the past week, and she was tired of it. The idea that the woman should adjust her behavior to avoid beatings was a dominant one in this and similar discussions:
It sounds like she’s ready to break the marriage. Hold on, I’d say: are you listening to him? You might find she herself is the one who’s at fault in that house. She sounds like a disrespectful woman. I’d advise her like she were my own child, saying “don’t even think of divorce, you’d lose your standing in the community.”
You’d need to ask her though. Maybe she actually wants to ask you about sex, and is embarrassed to ask directly. But if it’s beating, then I’d sit with her, I’d say, “daughter, tell me how you behave in the home. Do you cook for the time when he wants to eat?”
And also what she cooks. Maybe she doesn’t cook what he likes. Like he asks her to cook with palm oil, and she cooks with vegetable oil. Or he says “don’t put chillis in the sauce,” and she decides “I’m the one cooking, I make the sauce I like.”
“My daughter, the problem is with you.”
But even maybe she is not doing it on purpose, she is young. I would tell her, “daughter, learn to observe what your husband likes. How to please him.” Or-?
I personally, I’d ask her, “Mightn’t it be he’s slapping you because you’ve made a mistake?” I’d make her sit down, and calm her so she tells me why she’s beaten. I can’t give her advice before asking her well. After she tells me, maybe I’ll find in some way, her husband has wronged her. But even if he wrongs you, try to change, try to be this way: maybe he’ll change too. I’ll tell her “you’ve gone off-course here, you’ve gone off-course there. In order to live well with your husband, you must be in this way.” I’d advise her, so she can continue to live well, to live in her marriage.
Teresa and her co-discussants agree that if this is a good beating they are dealing with, then being beaten is the wife’s cue to behave more the way her husband wants her to, and censor away the parts of her behavior he might not like.
A Good Beating Deters Deviance
Groups discussed good wifely behavior not as something particular husbands demanded, but rather as what society expected. They talked about the good beating as a way to make married women behave normally by penalizing actual and potential deviance from the norm. Participants explicitly described dominant community norms—or rather, what they reckoned they could refer to as community norms when speaking to other members of the community.
Let’s say you’re just plain lazy. Your husband comes home, there are no clothes for him to change into. Everything is dirty. If you’re beaten then, it’s for a real mistake. You’d keep quiet about that, wouldn’t you?
Your mistake, your shame. You wouldn’t even want your friend to know.
Laziness to the point of beating! Shame.
Aza expected all others in the community (even a woman’s friend) to consider the woman’s act a deviation from the norm (“mistake,” ”shame”), to support the violated norm, and thus support the beating that punished the deviance. In all discussions, when a woman was beaten for deviating from a norm, the community was presumed to support the norm and to condemn her for deviating from it.
Imagery of deviation from the standard permeated discussions of beatable behavior. For example, Emma summarized her group’s take on a wife-beating vignette in terms of veering from the correct path, the one that everyone else knew to take: “You just went your own way, for who knows what reason. So of course you must be punished.” One phrase used exclusively in reference to wives going their own way, instead of the way all wives should go, was “hapaswi kabisa,” meaning it was completely inappropriate, out of order, uncalled for, and unseemly. Respondents contrasted such behavior with that which was seemly and fitting, or “inafaa.” A good beating steered a deviant wife back to the norm, and was often referred to as “adhibu”: chastising, punishing, or judging. Its goal was to achieve “heshima”—respect and obedience—and “mtii”—a docile and biddable manner—in a wife. Thus, if she showed signs of “kiburi,” that is, if she seemed “headstrong or recalcitrant, you must beat her so that she continues in the way that is her duty, instead of developing bad habits to the left and right.” A good beating deterred deviance.
A Good Beating is One Means of Control
Above all, participants saw the good beating as a means of control. This became clear from the way they placed it in the context of many other control methods. They discussed beating as one of an array of “punishments” a husband could use to make a wife behave. Thus, one woman’s statement that beating was the only way husbands coerced their wives drew spontaneous protests from others, who could list a seemingly endless array of other means of control:
Our punishment as grown women is just with sticks-
Caning-
We have no other punishment. Sticks, well another is slaps, punching with the fist—just beating.
Wait-
Let me join, another is if a man is abruptly angry, he could send you home, or deprive you of clothes or food-
(He’ll say) “No food for you, you’ll know where to go eat,” “no new clothes for you”—like that.
He couldn’t do that: I’m the one cooking! I’d just eat! But clothes, yes.
Being sent home, it happens.
So many ways. He says you’re disrespectful: “Let me bring another (wife) who’ll respect me and listen to what I tell her!”
Ah, that one!
Then-! He’s caught you.
Pili’s co-discussants interpret her claim as her just getting ahead of herself. Once they made her consider some examples, she recognized them (“but clothes, yes”) as other ways that a husband could enforce his will (“he’s caught you”).
Men’s discussions often involved the suggestion that instead of beating a misbehaving wife, a husband could find other ways to make her do as he wished, for example, “Why not call two or three people to talk with her, and see if that changes her behavior, before beating her?” That the goal of a beating was to make a wife behave, as explained earlier, was a given. The question was whether the beating was the best means to the goal:
I would not beat her, no matter how severe her mistake. I would just take her to court. And I would admonish her: “wife, why are you ruining everything?” That’s why it’s not right to hit her.
But don’t you know that if you don’t punish her for her mistake, it can become her habit? Your life could be destroyed by the stupid things she does.
Yes but beating her doesn’t make her quit her bad behavior. Not like when you put her in front of a judge. And if you hit her in anger, who knows, you may be unlucky and kill her, and then they’ll take you too.
Ok, but if you take her to the elders or the church leaders or sheikhs like you say, don’t you see that’s punishment too?
Yes, that’s why I’m saying I won’t beat her, but I won’t leave her the way she is, I will look for another way to correct that bad behavior that she has.
Many exchanges resembled that between Abass and Isac. Respondents like Abass deemed beating necessary as opposed to letting the wife go unpunished. However, once a respondent like Isac explained that there were other ways to control her, they reconsidered.
A Good Beating is Asymmetrical
Groups discussed beating at great length as a way of simply ensuring that vital chores were done, for the survival of the family. However, in some discussions, participants were called out by their fellow discussants for presenting this as mutual, lest they gloss over the fact that it was husbands beating wives, and not vice versa. For example, after Pedro and Pablo elaborated on how the good beating controlled the amount of damage one person could cause society, Marko began to challenge them by pointing out that this control only went one way:
So your wife, you find her in the wrong, you give her a warning, for example you flog her three strokes. Now, if she finds you doing wrong, if she tells you, for this mistake today you’re required to get four strokes, will you lie down?
Now you go to rights! Whose rights?
It’s not for any mistake you flog her—it depends on how serious, if she needs warning-
Yes, but you, what kind of mistake can you do that your wife is required to give you punishment, what could be a good reason?
“Husband, I must beat you.” Will you lie down?
Maybe if she knocked him to the ground-
Not “would she overpower you”: would you lie down?
(Laughter) He would not! How could he lie down?
Who’s ever heard that anywhere there’s a woman who told her husband, “lie down, today you have done this, I must flog you”?
Whenever the image of a “good beating” from wife to husband was raised in a discussion, it was as a role reversal so absurd that it was either hilarious or perverse. This does not mean that wives never beat husbands; FGDs are unreliable for telling us which events actually occur. It does mean that women cannot beat men with the approval with which men can beat women. The good beating ideal, as Marko pointed out, was wife-beating, not husband-beating. It was husbands, not wives, who should control their spouse by violence if necessary, and it was wives, not husbands, who should submit to this coercive control.
Discussion
This study has illuminated several sociopolitical aspects of wife-beating in Tanzania. Rather than isolate these as unique to Tanzania, I will consider them in light of existing scholarship on the sociopolitical dimension of VAW.
The discussions revolve around the singling out of particular types of beating as “bad” without condemning beating per se. In other points in time and place, such selective condemnation of VAW has served to normalize the more common VAW. For instance, Clark (2002) has shown how, historically, the same discourse legitimated beating as a tool to uphold hierarchy in England. Authorities there approved of beating a wife to enforce her duty, “though not outrageously” or in a “cruel manner” (Clark, 2002, p. 191). Hegemonic norms in Tanzania approved of beating a wife to enforce her duty, but not unreasonably or with cruelty (ukatili). Clark found that sanctions that limit the brutality allowed in wife-beating actually safeguarded the legitimacy of the social order the beating maintained: My data suggest the same can be said of the condemnation of “bad beatings” in Tanzania.
Feminist research on how women seeking to avoid violence are encouraged to police their performance of “good wife/daughter/mother” roles can also aid in understanding the findings (Stanko, 1996). In discussions of the good beating, women responded to the violence by imposing behavioral restrictions on wives. The good beating was a punishment for inappropriate behavior, and her cue to correct it. This self-censoring to avoid violence is how violence against women serves as social control elsewhere (Gagné, 1992; Stanko, 1997). It is also how censure works. Sumner (2012) describes censure as a political economy aimed at self-disciplining and self-motivation, in which the self-imposition of behavioral restrictions—which I call “self-censoring”—is key. In other words, censure works when people self-censor to avoid being censured. 5
Self-censoring is a defining characteristic of what both men and women discussed as a good beating. The reason unreasonable and irrational beatings were “bad” was that they did not efficiently change the wife’s course of conduct. By contrast, a good beating, when administered strategically, changed her conduct by its mere potentiality. As Samuel says in one of the cited discussions, if a wife just knows she may be beaten, then that knowledge can make her obey her husband without him having to beat her. That the latent possibility of violence influences the choices of all women regardless of whether and to whom the violence actually materializes was a central tenet in the initial activism against VAW in the 1970s, especially in DV (Riger & Gordon, 1981). The findings serve as a timely reminder that women’s compliance may be violently enforced compliance even when the physical act of violence does not materialize. As Samuel points out, the mere threat of violence may suffice, if it is credible. This resonates with Gramsci’s point that hegemony is dependent on the individual who can govern herself without the need for violent coercion (Gramsci, Nowell-Smith, & Hoare, 1971, p. 268). When the target of a threat avoids the harm that is threatened by complying, that compliance signifies the success of the threat, not resilience or resistance to it (Stanko, 1997).
The self-censoring effect that defines the good beating is precisely how violence is a form of power. Walby (2009, p. 198) theorizes violence as a form of power on the grounds that it is used to change the course of conduct of the person it targets. As this is the very purpose of the good beating, it is as a form of power that DV is legitimated. In other words, the wife-beating that is supported is wife-beating that asserts and enacts a man’s power over his wife. A censure perspective yields the same analysis: censure that makes others self-censor, is power, and it is beating that affects self-censoring that is supported. Sumner’s seemingly abstract claim that “who censures, rules” (Sumner, 1997), then, was common sense to participants. They took it for granted that to beat was to censure through violence, and that who censured whom was a question of who ruled whom.
The good beating, then, is violence that orders society: supported by dominant discourse insofar as it enforced a specific power order between husband and wife. A wife who did not submit to her husband as expected was a disapproved-of rule-breaker: in short, a deviant (Marshall et al., 2007). Punishing her affects not only her: It maintains norms in society and deters deviance at large. It ensures that she and others comply with the norm and follow the rules.
At the same time, the good beating is violence that controls women. The controlling dimension of the good beating ideal is underlined by the way respondents discussed it as one among several means of controlling a wife. This resonates with Stark’s (2009) call for understanding wife-beating as part of a larger pattern of “coercive control,” within which violence is preceded, accompanied, or displaced by other control tactics, and perpetrators simply adjust their strategies so as to maintain their control within the domestic sphere. The good beating builds on the understanding that such control is legitimate and necessary.
Thus, the data are consistent with how Stark argues we should conceptualize wife-beating as one of many tactics by which men elicit obedience and submission by violence or the threat of it, so as to preserve an unequal gender order (Stark, 2009). The reason it is a “fallacy of misplaced concreteness” to measure violence by counting the number of bodily acts of violence, according to Stark, is that this fails to capture the very meaning of the violence according to perpetrators and victims (Stark, 2007, p. 198). Respondents’ conception of the good beating as only one of many means of controlling a wife coincides with how victims and perpetrators Stark worked with understood partner violence: as one option within a general pattern of coercive control. The good beating is violence that serves coercive control.
However, if the good beating constituted a social order that disadvantaged women vis-à-vis men, the fact that this ideal emerged from women’s discussions as well as men’s demands attention. Did women choose to be controlled? If so, how could this be coercion?
This is where the fit between Gramsci’s hegemony concept and the appearance of coercion and consent in the data is important. The data, it will be remembered, represent not what individuals believe, but what they say in groups. I constructed the good beating from the shared commonality of what respondents across all discussions referred to as norms they expected to have to reckon with as dominant in society. As such, it is made up of hegemonic norms: regardless of whether an individual agrees with them, she expects them to be dominant in society, and factors that into her decisions (Ridgeway & Correll, 2004). The hegemonic nature of the good beating is key to women’s compliance with it: It is because they expect society will blame a wife’s beating on her, that they encourage self-censoring.
Thus, the findings show how the perceived social legitimacy of violence is what makes it powerful. In this they illustrate the importance Gramsci (1971) ascribes to ideas in, social structure (Lears, 1985). In the data, hegemonic norms link coercion with compliance. What made women complicit? It was by their approval of beatings aimed at wifely submission that the violence secured behavioral change. Or, in Gramscian terms, it is by sharing the norms that legitimate their domination that dominated groups become complicit in their own domination. For its intended effect on women’s choices, the data show, the good beating relies on its supporting norms being hegemonic norms. In this way, complicity and force are not mutually exclusive, but support one another in hegemony.
At the same time, women’s support for the norms legitimating the violence is itself supported by the violence. Women’s self-censorship is for self-protection: When women encourage conformity to the norm enforced by the violence (as Teresa, Nassra and Mwanapili do), they do so as a way to avoid violence. Women themselves enforce the norm so that men will not use violence to enforce it. This shows that it is not with full and free consent that women support the norms.
Thus, in a cyclically reinforcing dynamic, norms give power, which is power to enforce norms. In illustrating this, the data reflect the duality of social structure: structures of inequality both shape human agency and are themselves shaped by human agency (Giddens, 1986). The findings help us see how wife-beating works to control women. In these dynamics, women are both subject to constraining norms, and agents of them. Rather than frame women as fully victim or fully agent, then, the good beating ideal shows us how the violence, the norms it supports, and the norms it is supported by, together diminish women’s room for maneuver. It uncovers the “choice” of whether to conform to or deviate from dominant norms of how a wife should behave is one that women do not make freely, but rather at the pain of violent censure. The good beating ideal makes any woman’s choice to comply with patriarchal ideals unfree: in this it makes sense to think of the violence as “a liberty crime” (Stark, 2009, p. 1513).
Conclusion
This investigation of the meanings of support for wife-beating in Tanzania has uncovered ordering dimensions of the violence that we obscure when we frame it in seemingly apolitical “humanitarian” terms of health and suffering. The very basis on which beating is supported falls within this ordering dimension that we obscure, in that the sociopolitical role of the violence was central to the socially embedded meanings of support for the violence that emerged from the investigation.
In analyzing the dynamics of how the good beating constitutes a particular social order, this article illustrates the critical social theories it draws on. The good beating as one of many means of controlling a wife illustrates Stark’s warning against the fallacy of misplaced concreteness. Reconceptualizing social control and deviance as censure shows wife-beating as being about who has the ideological power to apply rules and thereby control the normative order. This illustrates how violence, insofar as it is regarded as legitimate censure, is both constitutive of and constituted by power. Finally, the analysis of how, when shared norms secure the controlling effect of the violence, coercion and complicity coexist, illustrates how hegemony plays out in gender relations.
The sociopolitical dimension of VAW described here has implications for policy and practice. Although addressing VAW as a health problem is vitally urgent, addressing it as if it did not have a sociopolitical dimension, and without a willingness to challenge hegemonic norms, can amount to sanitizing unjust power relations. What is needed, then, is policy that acknowledges that this violence affects the sociopolitical order in society as well as individual women. 6 Regarding practice, to intervene only on the VAW that the community already condemns can amount to legitimating the very sociopolitical order VAW is used to support. Practitioners should beware of the danger that they may be actively sanitizing the coercion and control of women if they are reducing injuries by leaving coercion and control intact.
The study also has implications regarding practices such as VAW, which feminists claim repress women, yet which some women support. The empirical account presented here of how coercion and consent combine in the hegemonic ideal of the good beating suggests new solutions for researchers, practitioners, and policymakers confronted with such practices. It shows that we do not need to choose between either taking whatever women say at face value, as their “free choice,” or ignoring their views. Instead, we can take a closer look at women’s support for the practice, as this study does, and see how the question of coercion and consent is not necessarily one of either–or. The account also warns against taking women’s compliance with community norms as simply a sign of women’s agency, and against taking the argument that “women themselves say so” as evidence that those norms are neither coerced nor coercive. When breaking a rule has negative repercussions like violence or ostracism, we cannot assume that following it is free. Can a person “freely comply” with rules that are enforced with violent censure? The censuring and censoring by which women comply warn against such uncritical proffering of what women say as evidence that they freely and happily oblige the patriarchy. When a woman knows that some women’s non-compliance is censured with the support of the society she lives in, then her choice to comply is not a free one.
Footnotes
Acknowledgements
I would like to thank Gro Th. Lie and Thera Mjaaland for supervision, Neil McLaughlin for guidance, Helen Baker for thoughtful comments, and especially Rosemary Barberet for encouragement and advice.
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) received no financial support for the research, authorship, and/or publication of this article.