Jezebel at the welfare office: How racialized stereotypes of poor women's reproductive decisions and relationships shape policy implementation

Abstract

Current welfare scholarship lacks an analysis of how caseworkers discuss sexuality-related issues with clients. Seventy-two of 232 transcribed welfare interviews in three states included discussion of reproductive decisions and relationships. Overall, caseworkers’ language reflected negative myths regarding African American women's sexuality and motherhood. By virtue of their status as welfare recipients, regardless of their individual races, clients were placed into racialized myths through workers’ talk. This analysis demonstrates that though not present in every welfare interview and often veiled in bureaucratic language, negative ideas about poor women's sexuality persist in welfare policy and are deeply embedded in its day-to-day implementation.

Welfare workers’ conversations with clients about fertility and family formation have the potential to illuminate the discourses on poor women's sexual behavior that are embedded in present-day welfare policy. In order to understand how poor women's sexuality and fertility are framed during welfare office interviews, this study examined interactions about reproductive decisions and relationships between caseworkers and clients, using transcripts from a multi-site study and focusing on caseworkers’ language.

A substantial amount of scholarship has grappled with Temporary Assistance to Needy Families (TANF), the program resulting from 1996's welfare reform. Much of this work critically analyzes the policy itself (e.g. Smith 2001; Soss & Schram 2007), assesses the policy's effects (e.g. Burger 2004; Corcoran, Danziger, Kalil & Seefeldt 2000; DeJong, Graefe & St. Pierre 2005; Loeb, Fuller, Kagan & Carrol 2003), or examines aspects of the policy's implementation (e.g. Hays 2003; Meyers 1998; Meyers & Lurie 2005; Ricucci 2005). Implementation research like ours focuses on interactions between clients and caseworkers in order to understand how policy mandates are translated within social welfare systems. Research on policy implementation has shown that street-level workers exercise great discretion in their application of formal policy mandates (Lipsky 1980; Maynard-Moody, Musheno & Palumbo 1990; Watkins-Hayes 2009). Frontline workers are both constrained by policy requirements and able to interpret them in such a way as to facilitate or hinder client access to services (Meyers 1998; Meyers & Lurie 2005).

An important element of current scholarship on welfare reform implementation studies these efforts at the ground level, using data from interviews or observation and recording in welfare offices. Hays’ 2003 book, Flat Broke with Children, uses such data to examine the cultural norms, beliefs, and values embedded in welfare reform. She points out that TANF, both as written and as implemented, enforces “family values” by discouraging women from raising children alone. Watkins-Hayes’ (2009) work focuses on how welfare workers’ professional identities and their race and gender locations interact to influence their implementation of policy as they work with clients. Both of these works use day-to-day talk in welfare offices to study implementation, but neither focuses on how workers and clients discuss issues of sexuality and reproduction in the context of the welfare interview.

Welfare office talk matters both as a measure of policy implementation and as the kind of language in action that can contribute to perpetuating stigmatizing discourses of poor women's sexuality. Reich's study (2005) of clients’ interactions with child protective services sheds light on how everyday practices in government offices both draw upon and contribute to racialized and gendered stereotypes of bad parenting; no equivalent work has been done using data from welfare offices. Women on welfare have long been the subjects of derogatory stereotypes, and their sexual behavior and reproductive decisions have been a major element of these negative portrayals (Cassiman 2006; Collins 1990; Collins 2004; Lens 2002; Roberts 1997; Roberts 2002). Such myths, as Dorothy Roberts puts it, are “more than made-up stories” (Roberts 1997, 8). Myths about people (e.g. poor women, welfare mothers, Black women) affect not only the treatment of specific individuals – unjust in and of itself – but also the treatment of everyone in those people's groups: All women, all mothers, all African Americans. As such, these demeaning myths are of concern in the realms of both scholarship and citizenship (Harris-Perry 2011).

Gilens (1999), in his study of national polling data about welfare policies, found that white Americans responded to racially encoded language: they disapproved of “welfare” (something received, as they saw it, by African Americans), but approved of “assistance for the poor.” Likewise, Hancock's (2004) study of references to welfare recipients in newspaper reports and the Congressional Record showed how the public identity of welfare recipients was cast as sexually permissive African American women, and Cassiman (2006, page 57) delineated the social construction of the “welfare queen” as “the embodiment of the deviant... black, poor, and dependent upon an unearned income.” Empirical falsification of such negative myths is often insufficient to diminish their power. For example, the US General Accounting Office reviewed research on the effect of TANF policies on out-of-wedlock births, and concluded that there was not evidence that these policies affected the childbearing choices of women on welfare (US General Accounting Office 2001). In spite of this authoritative statement, the dehumanizing myth of the welfare client having another baby to increase her check persists. Critical analytic descriptions of such myths in action, such as those produced by this study, can potentially contribute to lessening their influence among those who make or implement welfare policy, or who shape it with their public statements or their votes.

The sexual behavior of women and the structure of their families have long been a concern of social policy. The federal Aid to Dependent Children (ADC) program was established in 1935. As the Social Security system was modified in subsequent years, more widowed white women were transferred from ADC to Social Security benefits, and unmarried mothers, particularly African American mothers who were excluded by design from other social assistance programs (Jones 1985), comprised an increasing percentage of ADC caseloads. ADC became a lightning rod for criticism as this transition occurred, particularly among conservative leaders who believed that providing cash assistance to mother-only families enabled women to avoid the traditional bonds of marriage (Gordon 1994; Murray 1984). Poor women were stigmatized by their use of welfare; this stigma was particularly harsh for women of color, as racist attitudes linked the African American women heading these families with irresponsible sexual behavior (Berkowitz 1991; Collins 1990; Jones 1985; Roberts 1997).

As discriminatory state welfare practices were struck down by the Supreme Court, women previously denied benefits began to receive cash assistance for the care of their children, and welfare rolls grew dramatically (Quadagno 1994). Cyclically, whenever a system of overtly biased practices was legally overturned, another set of welfare implementation practices seemed to emerge to exclude Black women. Thus as welfare officially became more accessible, state policies also increased scrutiny of poor women's sexual behavior. For example, in 1960, Louisiana discontinued benefits to 95% of its African American clients on the basis of a “suitable home” provision that prohibited welfare benefits for families in which any child had been born out-of-wedlock (Lindhorst & Leighninger 2003). Welfare caseworkers were also required to conduct raids to find men who were not married to the welfare recipient but were living in the home (Abramovitz 1996), until legal and administrative challenges by the National Welfare Rights Organization and other groups ended these practices (Neubeck & Cazenave 2001).

Explanations for welfare use focused increasingly on a “culture of poverty” argument. The culture of poverty thesis, resurrected by Charles Murray (1984), suggests that poverty is caused by the “deviant” values of poor people, particularly poor Black people. One relevant example is his conclusion that poor women, especially African American women, have babies in an entrepreneurial fashion to increase their welfare checks rather than engaging in paid work. Instead of an overtly racist focus on the creation of laws that limit African American women's access to financial assistance, culture of poverty arguments focus on women's transgressive moral and sexual behavior and advocate for conventional gender structures, themes that persist into the current round of welfare reform. While important refutations of the culture of poverty argument exist (for example, Katz 1989; Piven & Cloward 1971; Quadagno 1994), it nonetheless continues to exert a powerful influence over US welfare policy.

With welfare reform under the Personal Responsibility and Work Opportunity Reconciliation Act (PRWORA) in 1996, the federal government once again asserted its right to set normative standards for women's sexual behavior, including the prevention of out-of-wedlock pregnancies and the encouragement of marriage. This assertion reinforced welfare's historical devaluing of poor women and women of color as mothers (McCormack 2005; Roberts 1997, 2002). One example of PRWORA's regulation of reproduction is “family cap” policies that allow states to prohibit increasing a family's financial assistance if the mother has another child while receiving welfare. Twenty-two states have currently implemented family cap provisions (Smith, 2006). This policy and others sought to eliminate any perceived financial incentive for women receiving welfare to have more children, but empirical evidence of their effectiveness has been mixed (Horvath-Rose, Peters & Sabia 2008; Jagannathan & Camasso 2003).

Welfare rolls dropped dramatically after the passage of PRWORA. Debate continues as to whether these declines were due to discontinuation of recipients who would have been eligible under previous welfare rules, improvements in national economic conditions immediately after PRWORA passed, or increases in the stigma associated with welfare receipt (Danziger 1999). Under the Bush administration, further efforts were made to encourage marriage through the reallocation of TANF monies to marriage and fatherhood promotion efforts which focused on patriarchal, heterosexual marriage as the primary solution to poverty among women and children (Edin & Kefalas 2005; Hays 2003; Weigt 2010).

The Current Study

Based upon this historical context, we would certainly expect the regulation of women's reproductive decisions and relationships to be a substantial element of frontline workers’ implementation of TANF. Our objective with this study was to explore the relationship between welfare workers’ conversations with clients and dominant discourses regarding sexual behavior and family structures that are embedded in welfare policy, while also attending to the ways in which gendered and racialized assumptions intertwine in workers’ implementation of these policies. Our focus was not on connections between worker rhetoric and welfare case outcomes (e.g., whether a client was granted benefits), but on caseworker language itself. We addressed our research questions by examining conversations related to fertility and family formation in worker-client interactions in welfare offices, using transcripts of these conversations from a multi-site study. To date, scholarship on welfare reform has not focused directly on interactions between welfare workers and clients in relation to issues of sexuality and fertility.

Method

Data for this study consisted of transcribed interviews between welfare caseworkers and clients applying for TANF benefits at public welfare agencies in three states. Data were originally collected as part of the Frontline Management and Practice Study conducted by the Rockefeller Institute of Government for the U. S. Department of Health and Human Services (see Lindhorst, Meyers & Casey 2008, and Meyers, Riccucci & Lurie 2001, for further description of the study). The goal of the parent study was to investigate the implementation of welfare reform policies in states with diverse administrative and political structures. The present research is a qualitative analysis of a subset of data from this larger study.

The research design for the parent study maximized organizational variation by drawing samples in three stages. The first stage used purposive sampling to select states, based on variation in location, political culture, and TANF agency structures. The states chosen were Georgia, Michigan and Texas. The second stage used purposive sampling to select 9 sites (2-3 per state) which represented urban and either rural or suburban locations. Table 1 describes differences in the three states in terms of their administrative structures, economic indicators, policies, and benefit amounts. The states differed from one other on these dimensions, with Georgia and Texas showing the most variation between them, while Michigan was similar to Georgia in some regards, but more like Texas in others. Each of these states was below the national median in child well-being indicators (Annie E. Casey Foundation, 2003). The lower child well-being ranking of these states indicates that while they differed from one another somewhat in terms of policy and organizational structures, they were similar in terms of their levels of poverty and of underlying need for TANF assistance.

Table 1.

Areas of Difference between Sampled States.

Areas of Difference between States (Number of Interviews; Proportion of Study Sample)	Georgia (90; 40%)	Michigan (85; 37%)	Texas (57; 24%)
Administrative Structures
TANF and employment services integrated	Yes	No	No
Decentralized vs. centralized (State) administration	Decentralized	Centralized	Centralized
Economic Indicators 1
% of children in poverty (2000)	18%	14%	22%
% of families headed by single parent (2000)	30%	28%	27%
Policies 2
Lifetime limit on TANF benefits	48 months	48 months	60 months
Initial limit on benefit receipt	None	None	12-36 months
Sanction policy	Gradual, Full Family	Immediate, Full Family	Immediate, Full Family
Work exemptions for illness	No	Yes	Yes
Family cap	Yes	No	No
Maximum TANF benefit for family of 3 with no income (2001)2	$280	$459	$201
Maximum income eligibility (2001)2	$514	$774	$401
Proportion (and number) of TANF interviews in state including discussion of sexuality3	44% (40)	26% (22)	17% (10)
Proportion (and number) of sexuality conversations in state including discretionary moralizing3	30% (12)	36% (8)	40% (4)

¹Annie E. Casey Foundation. 2003. Kids Count Databook. Baltimore, MD: Author.

²Rowe, Gretchen and Mary Murphy. 2009. Welfare Rules Databook: State TANF Policies as of July 2008. Washington, D.C.: The Urban Institute.

³Current study.

Data Collection

Data collection took place in 1999-2000. Frontline welfare workers’ encounters with clients were captured either through tape recording and transcription or through the taking of detailed notes by a trained observer. Informed consent for observation and recording was obtained from the worker in advance and from the client at the start of the encounter. Research assistants tape recorded and transcribed data from in Texas and Michigan. In Georgia, we were unable to obtain permission to tape record encounters, so research assistants were instructed to record verbatim conversation on 25 topics and activities that were of particular interest, including discussions of reproductive decisions and relationships. Comparison of the Georgia transcripts to the recordings in Texas and Michigan showed no meaningful differences in the length of transcripts or the topics discussed.

Sample

Our aim was to characterize the discourses of fertility and family formation reflected in the routine social and linguistic practices of the welfare encounter. Thus, we focused our preliminary analyses on interviews in which these topics were likely to be discussed. In most welfare systems, clients must participate in an “application interview” which determines their eligibility for benefits; after this, they usually have minimal follow-up contacts with caseworkers. It is during these initial applications (or reapplication interviews for clients renewing their benefits) that conversations about sexuality, fertility and family formation are most likely to occur.

This study's sample has limitations. Data were collected over a decade ago and thus may not reflect current welfare office operations, which may be different in ways to which this study cannot speak. Since locations for data collection were purposefully selected in the context of the parent study (Meyers 1998), they may not be representative of all U.S. welfare offices. Because we worked with written text, nonverbal worker-client communication that may have provided further insight into these transactions was lost, and the presence of research assistants in offices during welfare interviews may have altered worker behavior in unknowable ways. The data from Georgia may differ from the recordings obtained in Texas and Michigan, potentially affecting its trustworthiness. In spite of these limitations, the available transcripts represent a unique opportunity to examine the spoken discourses with which the welfare system contributes to the construction of poor women's gendered positions through talk about reproductive decisions and relationships.

Our initial data set contained 232 observations of TANF interviews between welfare workers and female clients. The sample of interviews involved 60 percent female workers and 17 percent male workers; the remaining workers’ genders were not recorded. Forty-four percent of workers were African American, 25 percent European American, 4 percent Latino/a, 1 percent Asian American, and 25 percent of unknown race-ethnicity (percentages do not sum to 100 due to rounding). Seventy-eight percent of clients were African American, 21 percent European American, and 1 percent Latina; all were women.

To select a subset of interviews and data for in-depth analysis, the first author read and re-read all 232 interview transcripts and coded conversations on topics of pregnancy, birth control, family planning, conjugal relationships, and sexual behavior. This topic coding resulted in a subset of 72 interviews that included talk of reproductive decisions and relationships. The second author independently coded a subset of transcripts to validate the topic coding; convergence was excellent overall, and disagreements about topic coding were resolved through discussion. The inclusion criterion for child support conversations deserves special mention. Child support enforcement generally involves workers gathering information about the father's description, location, or social security number for use in finding him and compelling him to pay for the care of his child(ren), and these topics were a part of nearly every welfare interview. We differentiated these routine child support conversations from those that included a pronounced focus on the sexual aspects of the relationship, for example, establishment of children's paternity through questions about the client's sexual relationship with the father. We included child support conversations in the in-depth analysis subset of data only when they had this sexual content in addition to routine, non-sexual discussion of enforcement issues.

Analytical Approach

We drew on a discourse analysis framework to investigate how workers’ talk with welfare office clients perpetuated or resisted the negative stereotypes of poor women's reproductive decisions and relationships embedded in welfare policy. Discourse analysis operates from a theoretical perspective that treats social realities such as gender and sexuality as being socially constructed through language and social practices (Cameron & Kulick 2003). It focuses on what people do with their talk: they “produce versions of events, objects, and people” (Horton-Salway 2001: 153). Focusing on the 72 interviews in which workers discussed reproductive decisions and relationships with clients, we conducted across-case analyses and used techniques of data summarization and display (Miles & Huberman 1994), categorization (Saldaña 2009), and comparative analysis (Strauss & Corbin 1998). We characterized coded statements with brief descriptive labels and grouped related statements together into categories. We refined, split, or merged categories when associations, divisions, or overlap appeared in the data (Hall & Stevens, 1991). The first two authors worked together iteratively to produce the findings and verify their trustworthiness.

Findings

Only about one third of the 232 interviews between caseworkers and clients included conversations that were directly related to clients’ sexual behavior. We identified 72 interviews (31 percent) in which reproductive decisions and relationships were discussed; these conversations are characterized below. The remaining 160 interviews (69 percent) involved discussion of other welfare issues such as food stamp eligibility, access to child care, work requirement fulfillment, and routine child support enforcement with no discussion of sex or the conjugal relationship. We began these analyses with a focus on how race and gender intertwined in welfare office talk about sex, childbearing, and family organization. Through close reading of the interviews, we identified moments of informal, discretionary talk that focused on moralizing about client behaviors. The more predominant form of talk, however, was a highly bureaucratized exchange of questions and answers. We start with findings related to issues of race and racialized talk within these exchanges, followed by a description of the moralizing and bureaucratic styles of talk.

Race or Racialized Talk?

In the first phase of analysis, we looked at whether race/ gender concordance between welfare workers and clients was associated with moralizing or bureaucratic styles of talk. All of the clients and over half of the caseworkers in our sample were women. The majority of the clients were African American, as were nearly half of the caseworkers. The distribution of each of these groups in our sample, as well as our sample size, may have obscured patterns of interaction specific to particular pairings, for example, between a white worker and a Black client, or between two African American women. We did not identify distinct discourses on reproductive decisions and relationships associated with workers’ race or gender, clients’ race, or the racial or gender makeup of worker-client dyads. Workers were no more likely, for example, to moralize with an African American client, or with a client of a different race or gender than their own. However, the discourse of poor women's reproductive decisions and relationships that emerged from these data on welfare workers’ everyday talk with clients was overwhelmingly disrespectful. Regardless of their individual races, clients were subject to the negative myths that arose out of gendered racism against African American women so powerfully characterized by Roberts (1997) and Collins (1990, 2004).

While we observed an occasional exception – discussed below – in general, workers’ language functioned to create negative subject positions for clients. These positions were remarkably congruent with the stereotypes of African American women's sexuality and reproduction discussed in Collins (1990) and further elaborated by Roberts (1997). Some welfare clients were positioned as the immoral “Jezebel,” focused on sexual fulfillment without responsibility; others as the devious “welfare queen” who tricked the state into supporting her and her children; some as “matriarchs” who abandoned marriage in favor of asserting dominance within the family; others as “Mammies” who bore children only to neglect them. In the next section, we present examples of exchanges between workers and clients, highlighting moments when these racialized myths are present in the underlying assumptions expressed in the interview. In quotations from transcripts, we have replaced names with initials, and specific dates with spaces, in order to protect study participants’ confidentiality. When portions of the transcript not relevant to the analysis have been left out, we use this symbol: [...].

Styles of Talk

Our analyses identified two overlapping rhetorical styles that supported the stigmatized positioning of welfare clients, with the occasional exception. We labeled the first of these patterns “discretionary moralizing.” This style extended the moralistic “suitable home” talk documented among welfare caseworkers of the 1960s (Lindhorst & Leighninger 2003) into the present. It comprised workers’ personal and optional talk regarding the social riskiness of having babies as a poor woman, and the moral deficiencies they ascribed to women who had children and sought state assistance for their support. Discretionary moralizing was layered over the second style, bureaucratic talk, in roughly one-third of the interviews analyzed. The bureaucratic style involved statements through which sex-related policy decisions such as the family cap were implemented in the worker-client transaction, reducing reproductive decision-making to issues of verification and rule imposition. Unsurprisingly, given that they were official transactions in government offices, all of the 72 welfare interviews involving fertility and family formation included this type of bureaucratic talk.

The bureaucratic examples we discuss below may seem to make less of a contribution to the perpetuation of negative myths regarding welfare clients’ sexual and relational behavior than do the examples involving discretionary moralizing. The distinction between these two styles is important to note, however, so that the more subtle action of the bureaucratic imperative is not overlooked. Both strategies – old-fashioned moralizing and neutral-appearing bureaucracy – functioned toward the same end. Through them, welfare clients were positioned as sexually irresponsible supplicants for aid whose fertility and family structures were in need of governmental regulation, rather than as citizens for whom assistance was a right and whose reproductive decisions and relationships were their own.

Discretionary moralizing

In these cases, in addition to the policy-mandated bureaucratic language present in all the interviews we studied, workers did what we have called “discretionary moralizing;” they spoke in ways that reflected a personal sense of entitlement to judge and regulate clients’ reproductive decisions and relationships. These discussions included talk of clients as negligent of their maternal responsibilities, critiques regarding children's paternity and clients’ conjugal choices, and talk in which workers took clients to task for their sexual behavior. The moralistic strategy overtly draws on the dominant discourse of irresponsible reproduction and inappropriate family structure in which PRWORA is embedded.

A prominent feature of workers’ moralistic talk is statements that position clients as negligent “Mammies” (Collins 1990; Roberts 1997), as in this example from Georgia.

Worker:	How many children are in the home?
Client:	Five and one on the way.
Worker:	Why would you go ahead and do that? We can talk to y'all until we're blue in the face. You know you won't get any extra money.

Here we see discretionary moralizing layered on top of the bureaucratic imperative as the worker conveys the substance of the family cap regulation in less neutral, more personally judgmental language.

Another instance of morality talk appears in discussions about the fathers of clients’ children during workers’ attempts to get information for use in child support enforcement. These examples demonstrate workers’ personal normative judgments of clients’ reproductive relationships.

Worker:	Do you have your current ID? Are you getting child support?
Client:	Um, well right now, he's just helping me pay the mortgage so, I guess, yeah.
Worker:	Is he the father of all your children?
Client:	The other three, no, he's in prison, the father of the other three.
Worker:	You don't pick them good.

Here the client is talked about as to blame for her own difficulties due to her deficiencies at partner selection and placed in the “Jezebel” role for having children by two different men. The Michigan worker conveys her verdict on the client's partner with the statement “You don't pick them good.” The client is asked to justify the fact that she is not married to the man who fathered her new baby: PRWORA aims to encourage marriage as part of its perspective on appropriate types of family structure, but the worker's discretionary moralizing takes this conversation to a more personally intrusive level.

Discretionary moralizing can also have a sarcastic quality, as when a worker in Georgia said: “you don't have ANY information on their father?” When the client stated that she and her children's father “didn't talk,” the worker responded “You had to talk sometime, you had three babies.” Another worker in Texas expressed disbelief regarding the client's knowledge of her former partner:

Worker:	What's the father's name?
Client:	I can only think of his first name, that's _.
Worker:	How long did you go out with him?
Client:	A while. But he's in denial. Only thing I can think of is his first name.
Worker:	If you went out with him a while you should know his last name. [...] Oh, I'm getting a message [from the computer]. I never got that before. [...] I guess the system couldn't believe someone would only know the first name.

In both examples, the workers take a tone of disbelief that the clients would not have certain information about their childrens’ fathers. This language contributes to the workers’ positioning of the clients as “Jezebels” who have sex (and children) with men they do not know.

The moralistic rhetorical strategy also occurs in a way that exhorts the client to do better with the kind of tough talk that a coach or parent might use. In Georgia, after discussing tubal ligation with a pregnant client, a worker pointed out that the client's mother worked, saying “That's probably why she can take care of her children. Like you need to.” This quote from a Michigan worker is another minor exception to the overall negativity of caseworkers’ positioning of welfare recipients with their moralizing language.

Worker:	You're going to go on and get your GED [Graduate Equivalency Diploma]?
Client:	Yes.
Worker:	Because you need that. And you don't need any more children to hinder you. [...] You could go to a trade school and learn how to be a nurse. Don't say “don't know” – you just need to do it. And stop having babies. You're too young.

In both of these examples, the worker exhorts the client to pull herself up by her bootstraps. This coaching language is somewhat supportive of the client, positioning her as a potentially strong person who can succeed at school or a career and take care of herself and her children without assistance. It shows some worker resistance to the overwhelmingly negative perspective on poor women's fertility demonstrated in most other examples, and is also congruent with caretaking aspects of TANF policy. However, this paternalistic tone displays no less of an entitlement by the worker to decide what is best for the client regarding reproduction; it simply combines this entitlement with a more parental interpersonal style.

The bureaucratic imperative

The dominant discourse of poor women as sexually irresponsible affected talk in the welfare interview through workers’ personal and optional talk, but it did so more frequently through its influence on the policy and regulations workers were mandated to implement. The bureaucratic strategy did not involve overtly judgmental language regarding clients’ sexual behavior, merely a rote statement of the rules or a seemingly neutral inquiry for information. Much of the content of welfare office encounters was driven by the worker's task of gathering information to complete welfare forms or cover policy-mandated material. Topics of reproductive decisions and relationships (e.g., family cap, child support enforcement) came up in these contexts. Conversations involving the bureaucratic imperative reinforced negative myths regarding poor women's sexual behavior in a low-intensity way without using language that overtly denigrated clients’ reproductive decisions or relationships.

We often saw the bureaucratic imperative in encounters where the worker's goal was to get information about the father of the client's children for use in child support enforcement. In this Texas interview the worker's talk had the tone of an interrogation in a police drama on television.

Worker:	All right, this part. Do you know who M's father is?
Client:	Nuh-uh.
Worker:	Do you know anything about him? You never see his face or know where he is? OK. You never did find out who he is? (Client shakes her head.) [...]
Worker:	Do Z and P have the same father? (Client nods her head.) What's his name?
Client:	PM.
Worker:	OK. Um. When was the last time you had contact with him?
Client:	About a month ago. [...]
Worker:	Did he have any plans to marry you? (Client shakes her head.)
Worker:	OK, do you know his social security number? Date of birth?
Client:	September__, 197_.
Worker:	197_? Is he white? How tall is he?
Client:	I don't know.

Here the worker asserts her entitlement as a representative of the welfare system to get information from the client about the father of her children. At the same time, her language positions the client as an immoral “Jezebel” who has been sexually involved with men she cannot remember and as an unwed “matriarch” who is building a family outside marriage (Collins 1990; Roberts 1997). The worker demonstrates her skepticism about the client's truthfulness (i.e., “You never see his face?”), but the emphasis of the conversation is discovering information to track down the father rather than moral judgment of the woman.

In addition to across-state policies like child support enforcement, some states had their own reproduction related policies. One state, Georgia, had a specialized fertility policy, the “family cap,” that prevented women who were on welfare from receiving additional financial assistance if they had another child. In the encounter below, the bureaucratic imperative led to a conversation bordering on the absurd.

Worker:	Are you pregnant now?
Client:	No.
Worker:	You're not having any more?
Client:	No.
Worker:	Are you using any birth control?
Client:	I've had my tubes tied.
Worker:	I need you to sign this saying you understand that if you have any more, the baby will not be added to your check.

The dominant discourse of welfare that characterizes clients as at best careless, or at worst, entrepreneurial, about their fertility speaks through the worker so powerfully that even after the client states that she has been sterilized and cannot have more children, the worker continues to talk as if the client is imminently likely to conceive a baby and ask the state to support it. The worker's language enacts within the interview the negative view of welfare clients’ reproductive decision-making that permeates welfare policy.

The bureaucratic imperative is also visible in the requirement that pregnancy be medically verified. This documentation requirement in and of itself suggests that welfare clients are likely to fake pregnancy to obtain benefits, even in the absence of a negative tone from the worker:

Worker:	How many months pregnant are you?
Client:	7 months.
Worker:	Have you been to the doctor yet?
Client:	Yes.
Worker:	Did they give you a due date or a letter with the due date?
Client:	Yes, here it is.

While this Georgia worker uses neutral language, a negative perspective on poor women's fertility-related behavior nonetheless enters this interview through the documentation requirement. The myth of the devious “welfare queen,” who uses her fertility to trick the state into supporting her, informs the policy that workers implement in conversation with clients (Collins 1990; Roberts 1997).

There were several instances in which workers discussed reproductive issues with clients using bureaucratic language, but resisted participating in the demeaning stereotyping of poor women that imbues welfare policy. These instances arose when workers emphasized assisting clients to comply with TANF's requirements while at the same time taking into account women's needs during pregnancy or early motherhood, as in this example from Georgia.

Worker:	What kind of work – what do you want to do? Do you have a GED?
Client:	No. [...]
Worker:	You may want to work towards the GED while waiting for the baby. [...] We've got a program called New Connections to Work that may be good. Do you know what kind of work you may want?
Client:	Nursing. [...]
Worker:	We should do the program. Here's some information. [...] After you have the child, you go to work – you'll be set. It's a very good program.

Discussion

This research indicates that though sexuality-related topics are not discussed in the majority of welfare interviews, the private sexual decisions and relationships of poor women continue to be a focus of intervention in the day-to-day interactions welfare caseworkers have with their clients. Negative myths regarding poor women's reproductive decisions and relationships (Collins 1990, 2004; Roberts 1997) are expressed in welfare offices both through their influence on welfare policy and regulations and through the effect these myths have on worker behavior. How workers talk with (and implicitly “frame” clients) is prescribed by bureaucratic mandates whose underlying assumptions reflect racialized myths antagonistic to poor women (Collins 1990; Hays 2003; Roberts 1997). Workers’ own informal, discretionary conversations with clients reflected moralizing messages that were also shaped by negative perceptions of poor women's sexuality. While workers’ talk expressing personal judgments of clients may seem more offensive than their bureaucratized language, both styles of speech contribute to the dehumanization of welfare clients. The more subtle action of the bureaucratic imperative should not be disregarded; both of these strategies reflect the widely accepted false beliefs that poor women on welfare make careless or financially calculating reproductive decisions and that woman-headed families are not legitimate.

Just as the style of mainstream U.S. talk on race has gradually become less overtly and personally racist, more subtle and bureaucratized in its action (Bonilla-Silva 2006), so analogously has welfare office talk regarding sexuality. By embedding conversations about sexual matters in bureaucratic talk about child support enforcement and the family cap, these conversations are transformed from overtly intrusive discussions of private issues into normalized outcomes of the need to ask for financial help. Women of other class backgrounds would be unlikely to submit to these interrogations, but the United States’ current approach to ensuring a social safety net allows this continued focus for poor women (Hays 2003). This analysis demonstrates that negative ideas about poor women and their fertility and family formation decisions persist in welfare policy and in its day-to-day implementation by caseworkers, but that they are often veiled in bureaucratic language.

Our findings regarding the apparent absence of differences in welfare office talk related to the race or gender of workers or clients are congruent with recent work by welfare implementation scholar Watkins-Hayes (2009), who observes that caseworkers’ attitudes and behaviors are not determined by their social group membership alone. Rather, we observed workers’ fairly consistent application of an ideology of irresponsible fertility to clients regardless of clients’ races or their own. We ascribe this pattern to the racialization of welfare through which the state sanctions discourse (bureaucratic scripts and discretionary utterances) which reinforce underlying negative racial stereotypes, and normalizes the application of these myths to all poor women, regardless of race. While Black and white women comprise roughly equal proportions of total TANF recipients (35.7% and 33.4%, respectively), African American women are disproportionately represented among welfare clients (Hays 2003; U.S. Department of Human Services 2006). Welfare has been associated with Blackness in the collective mind of the U.S. public ever since African American women gained access to these programs in the 1960s (Gilens 1999; Quadagno 1994; Roberts 1997); in these findings, we see negative myths regarding Black women's sexuality and childbearing extended to include all women on welfare, regardless of their individual races.

Our findings are based on data that are over 10 years old, and we cannot be certain that the same racialized discourses we identified are employed in welfare offices today. However, scholars of race and policy such as Michelle Alexander (2010) have illuminated a pattern in which new systems of “racial control” in the US, which tend to be more overtly “colorblind,” emerge to replace those that are overturned. In this analysis, which seems likely to extend to welfare office practices, it is probable that these deep negative myths persist and continue to shape caseworker language, though they may look slightly less racially influenced on the surface.

We also noted differences among the three states studied in both the amount and the style of conversations on reproductive decisions and relationships (see Table 1). Caseworker-client interactions in Georgia were more likely than those in Texas or Michigan to include sexuality-related topics, however, these conversations were less likely to involve discretionary moralizing by workers than those in other sampled states. In Texas, while interactions were less likely than in Georgia or Michigan to include discussion of clients’ sexuality, when such discussions did occur, they were quite likely to involve moralizing discourses in addition to bureaucratic language; such moralizing was also more likely in Michigan than it was in Georgia. One possible explanation for these findings is that negative myths regarding poor women's fertility have a tendency to enter welfare office conversations, and if they do not do so in bureaucratically mandated ways, they may be more likely to do so through caseworkers’ personal moralizing. In states with explicit reproduction-related policies, like Georgia with its “family cap,” workers were mandated to discuss sexuality-related topics, and thus did so more frequently. Absent such policies in Michigan and Texas, conversations regarding sex and reproduction occurred less often, but because these conversations arose more at workers’ discretion, they often included optional, personal moralizing.

In light of these results, it is important to note that individual welfare workers are not solely responsible for talking in ways that position clients negatively (Lindhorst, Meyers & Casey 2008). The actions of these workers reflect a structural reality: Across states and settings, caseworkers are required by law to discuss aspects of the private sexual lives of poor women. Current welfare policy, informed by an ideology which positions poor women as irresponsible in their fertility and illegitimate in their family formation, speaks through these workers.

This study has implications both for the theoretical framing of future research on welfare, and for the implementation and creation of welfare policy. Our findings indicate the importance of careful theoretical attention to race and processes of racialization in any study examining welfare. Roberts’ statement that race “fuels the welfare debate even when it is not mentioned” (1997: 215) is exemplified by our finding that racialized myths regarding poor women's sexuality enter the welfare interview across multiple dimensions of potential difference. This dehumanizing framing of poor women was expressed both through workers’ optional moralizing talk about sexuality-related issues and in their bureaucratized language. Regardless of caseworker race, client race, or racial pairing of the worker-client dyad, these stereotypes affected workers’ language in welfare interviews. Viewed one way, this finding might be considered null. However, a theoretical attunement to processes of racialization supports a more convincing interpretation: that the dominant discourse of African American women's irresponsible reproduction and inappropriate family structure in which welfare interviews are embedded acts to “color” all welfare recipients and the language workers use with them.

Regarding policy implications, our analysis suggests that demeaning racialized stereotypes may influence welfare policy's creation and implementation in both obvious and subtle ways. Different training and supervision might decrease the amount of discretionary moralizing done by caseworkers. However, such a solution would only address part of the problem. Negative myths regarding poor women's sexuality not only shape workers’ optional talk, but may also play a role in welfare policy itself. Thus, it is likely that these myths will continue to affect day-to-day talk in welfare offices even if they emerge through the less obviously offensive bureaucratic language mandated by policy. This problem is sustained by tenacious stereotypes that act at many levels.

The pervasiveness of these stereotypes means that attempts to challenge them must focus on depths, not surfaces. To change the discretionary messages welfare workers give their clients, both their underlying perceptions of poor women and women of color, and more critically, the perceptions of policy makers, must be changed. This type of deep intervention will not proceed quickly or easily. It will require, at minimum, the kind of open discussion of race and racism that is rare in the U.S. (Alexander 2010; Bonilla-Silva 2006). However, such work has the potential to decrease the influence of negative myths upon those who implement welfare policy as caseworkers as well as reframing these issues for those who make it as elected representatives and those who shape it with participation as citizens.

Current welfare policy emphasizes marriage to “breadwinners” as a way of addressing poor women's need for financial resources for themselves and their children. This policy functions to place both women and men into a family structure in which men of severely limited economic means are made financially responsible for women and children, regardless of structural economic factors (e.g., low wages, unemployment) that make meeting this responsibility difficult or impossible (McCall 2000). It ignores the inconvenient fact that the majority of men who are the would-be husbands of women on welfare, and the fathers of their children, live in poverty as well (McCall 2000; Roberts 1997).

Regulating poor women's reproductive decisions and relationships has negative effects on women's freedom and well-being and does not adequately address child poverty (Smith 2006; Weigt 2010). Both through regulating sexuality and encouraging marriage, and through other elements of TANF (e.g. work requirements), US welfare policy treats financial provision for children as the almost exclusive responsibility of individual families (Cassiman 2006). Issues of structural poverty are not considered. Women are made ultimately responsible for ensuring resources for their children by finding husbands who can provide for them financially (Swift 1995). An alternative policy approach would be for the state to contribute to the support of children regardless of the sexual behavior or gendered positions (i.e. “breadwinner” and “homemaker”) of their parents. To take this approach would be to recognize that both earners and caregivers are required in families, regardless of whether men or women occupy these roles (Gornick & Meyers 2009). This perspective could decrease both child poverty and the stigmatization of poor women, which in the US often involves racialized stereotypes that exacerbate existing injustice.