‘ I just don ’ t think it ’ s that natural ’ : adolescent mothers ’ constructions of breastfeeding as deviant Kimberly

Breastfeeding is recognised globally as the optimal method of infant feeding. For Murphy (1999) its nutritional superiority positions breastfeeding as a moral imperative where mothers who formula-feed are open to charges of maternal deviance and must account for their behaviour. We suggest that this moral superiority of breastfeeding is tenuous for mothers from marginalised contexts and competes with discourses which locate breastfeeding, rather than formula feeding, as maternal deviance. We draw on focus group and interview data from 27 adolescent mothers from socio-economically deprived neighbourhoods in three areas of the UK, and ﬁ ve early years professionals working at a Children ’ s Centre in the Northeast of England. We argue that breastfeeding is constructed as deviance at three ‘ levels ’ as (i) a deviation from broad social norms about women ’ s bodies, (ii) a deviation from local mothering behaviours and (iii) a transgression within micro-level interpersonal and familial relationships. Given this positioning of breastfeeding as deviant, breastfeeding mothers feel obliged to account for their deviance. In making this argument, we extend and rework Murphy ’ s (1999) framework to encompass diverse experiences of infant feeding. We conclude with re ﬂ ections on future research directions and potential implications for practice.


Introduction
Breastfeeding is associated with a range of short-and long-term positive health outcomes for mothers and infants (Horta et al. 2007). Despite these outcomes, breastfeeding rates are strikingly low in developed countries (Perez-Escamilla 2019). In the UK, although 81% of mothers initiate breastfeeding (defined as putting a baby to the breast at least once), only 46% of babies [Correction added on 6 September 2020, after the first online publication: The abstract has been updated in this version] are exclusively breastfed at 1-week old, which decreases to 17% at 3 months and 1% at 6 months (McAndrew et al. 2012). Patterns of breastfeeding mirror wider health inequalities where mothers with low educational attainment, those living in socioeconomically deprived areas and young mothers are the least likely to breastfeed (McAndrew et al. 2012). This is likely to compound and exacerbate existing health inequalities.
This article focuses specifically on the infant feeding experiences of young mothers living in areas of high socioeconomic deprivation who, given the intersection of their age and socioeconomic background, are among the least likely groups to breastfeed. The UK's Infant Feeding Survey 2010 highlighted that mothers under the age of 20 are four times less likely to breastfeed than those aged over 35 (McAndrew et al. 2012). While there are parallels between the experiences of young and older mothers (for both, breastfeeding is an innately gendered act mediated by wider cultural expectations about women's bodies and behaviours), the particularities of adolescent motherhood mean the experiences of young mothers cannot easily be conflated with those of mothers more generally (Hunter and Magill-Cuerden 2014). Rather the distinctiveness of adolescent motherhood, being often unplanned and highly stigmatised (Yardley 2008), means feeding decisions are often conceptualised differently and different challenges occur.
To capture the distinctiveness of adolescent motherhood, we draw on, and expand, Murphy's (1999) analysis of infant feeding choices through employing a deviance framework to argue that negotiating complex discourses and expectations around infant feeding is particularly challenging for young mothers living in socioeconomically deprived contexts. In these circumstances, the nutritional and moral 'superiority' of breastfeeding is questioned and breastfeeding is often rendered as a deviant act. We explore the ways this deviance is constructed and accounted for. While Murphy (1999) shows that accounting for breastfeeding commonly focuses on its effects and consequences (i.e. how, where and for how long it is done), we show that for some mothers, it is the act of breastfeeding itself which requires justification. In making this argument, we rework Murphy's (1999) framework to show how deviance is a useful lens for understanding infant feeding but applies differently to different mothering contexts. We also contribute to sociological analyses of breastfeeding which, though wide ranging, have not focused extensively on young mothers' feeding decisions despite wider continued sociological concerns with adolescent parenthood (e.g. Brown 2016, Ellis-Sloan 2014, Mannay et al. 2018a, 2018b, Moore and Reynolds 2017 and marginalised mothers' infant feeding experiences more generally (Gillies 2007, Grant et al. 2018, Mannay et al. 2018a, 2018b. Experiences of breastfeeding as a key aspect and decision in young mother's reproductive lives are comparatively absent. Foregrounding young mothers' perspectives and experiences of infant feeding also has the potential to shape interventions seeking to increase breastfeeding rates in this population. Before turning to our theoretical perspective, we outline the complex moral landscape surrounding breastfeeding in industrialised nations followed by an overview of the data collection, analysis and findings around how breastfeeding deviance is constructed and accounted for by young mothers. We conclude with reflections on directions for future research and practice. Breastfeeding: a moral minefield Much like other aspects of women's reproductive lives, breastfeeding is 'open for comment' where women's decisions, bodies and experiences are subject to scrutiny, intervention and moralising (Stearns 1999, 308). For Brown, breastfeeding straddles the personal, public and political spheres and is mediated by family, partners, policymakers, the infant formula industry and wider culture (Brown 2016). Within this context of scrutiny, breastfeeding is positioned as a moral imperative intricately linked with 'good' mothering practices (Crossley 2009, Knaak 2010, Wall 2001 and the 'breast is best' message 'dominates the context' in which feeding decisions are made (Murphy 1999, 187). This 'breast is best' discourse goes beyond the nutritional value of breast milk to position breastfeeding as best for the environment, global economy (Schmeid and Lupton 2001), mother-child bonding (Knaak 2005) and womanhood as a whole (Malka 2007).
Breastfeeding, particularly in public, troubles normative constructions of breasts in terms of (hetero)sexuality (Rodriguez-Garcia and Frazier 1995), creating ambiguity of their purpose (Carter 1996) and positioning breastfeeding as a 'scandal' at the boundary 'where maternity meets sexuality' (Bartlett 2002, Boyer 2016, Carathers 2017, 71, Grant 2016. As Stearns (1999) points out, the good maternal body is imagined as asexual so lactating breasts represent a transgression of the cultural scripts available to, and about, women. Paradoxically, the right kind of breasts, suitably (un)dressed suggesting sexual availability, are ubiquitous in Western society (Acker 2009, McConville 1994. Murphy (1999, 205) argues infant feeding is a 'moral minefield' as it involves 'moral work' to externally account for feeding decisions and to (re)construct a sense of self as a breastfeeding mother (Ryan et al. 2010). For Lee (2007), much of this moral framing of infant feeding sits within a 'new health paradigm' where individuals are responsibilised for preventing ill health through 'good' behaviour such as breastfeeding. Lomax (2013, 97) demonstrates that this moralising of infant feeding through a 'health paradigm' underpins breastfeeding policy and practice whereby breastfeeding confers a positive identity and assumptions of responsible behaviour while formula feeding 'is associated with a deficit identity'. Such balancing presents a challenge for all mothers (Avishai 2007). Medical and nursing literature has demonstrated that some factors underpinning young mothers' low incidence of breastfeeding, such as anxiety about public breastfeeding, lack of interest and complications with work or education (e.g. Brownell et al. 2002), are common to all mothers. However, negotiating competing expectations and tensions around breastfeeding is even more challenging for women in marginalised contexts where the equation of breastfeeding with 'good' mothering may not have the same purchase for all groups of mothers (e.g. Grant et al. 2018). Researchers have highlighted the tensions and challenges particular to adolescents including limited prenatal engagement, lack of tailored support and feeling overwhelmed by breastfeeding without confidence to ask for help (Apostolakis-Kyrus et al. 2013, Hall Smith et al. 2012. Moreover, Condon et al. (2013, 156) suggest that, among adolescent mothers, breastfeeding sits outside 'normal' mothering behaviours 'in their social milieu'. Employing the notion of deviance, we take the suggestion of Condon et al. (2013) further through examining what contributes to the perception of breastfeeding as abnormal among adolescent mothers.
Breastfeeding and deviance Murphy (1999) argues that given the link between breastfeeding and maternal morality, formula feeding leaves women open to charges of deviance. However, she suggests that formula feeding is not ipso facto deviant but contingent upon the conditions within which the behaviour takes place (Hillman 2014, 487). For McHugh (1970, a behaviour must fulfil two conditions to be characterised as deviant. Firstly, it must have 'conventionality' which means being carried out despite behaviour 'in accord with the rules' being possible (Dingwall 1983, 131). Secondly, it must have 'theoreticity' which means the actor carrying out the behaviour knows what they are doing.
In the case of infant feeding, maternal deviance occurs when mothers knowingly break the rules by eschewing breastfeeding despite being aware of its status as 'best' for children (Murphy 1999). However, women who intend to formula-feed can challenge the understanding of their behaviour as deviant. Through 'techniques of neutralisation' (Copes 2003, Sykes andMatza 1957), formula-feeding mothers can account for their behaviour through excuses or justifications (Scott and Lyman 1963). Excusing a deviant act accepts its immorality but understands that the person committing the act is not fully responsible. Justifying an act refutes its immorality but accepts that the act was intentional or avoidable (see Scully and Marolla (1984) for an overview of this distinction applied to rape). Against this theoretical backdrop Murphy (1999) argues that women who formula-feed are required to account for this decision to try to neutralise its deviance and avoid social sanctions.
Within Murphy's framework, however, only formula feeding may be understood as deviant, requiring accounting for. Given the association of breastfeeding with good maternal behaviour, breastfeeding does not 'require the same kind of legitimation' (Murphy 1999, 201). That is not to say that breastfeeding goes without comment; Murphy's breastfeeding participants felt obliged to account for the potential impact of breastfeeding on their intimate relationships and to offer assurances that they would feed discreetly and for an appropriate length of time. Nonetheless, this accounting work is focused on the consequences and effects of breastfeeding, rather than the act of breastfeeding itself, whose moral superiority goes largely unquestioned.
While we take Murphy's framework as our starting point, we suggest that breastfeeding does not automatically occupy a universal morally superior position. Instead for some mothers, in our case adolescent mothers from socioeconomically deprived areas, it is breastfeeding, not formula feeding, which is understood as deviant and requires accounting for. Accordingly, we offer an extended, reformulated version of Murphy's (1999) model to reflect our participants' understandings of infant feeding deviance and demonstrate the complexities of competing deviance discourses. We have visualised this extension to Murphy's work in Figure 1 where Murphy's model is shown in dashed boxes (Positions A-D) with our extension in solid boxes (Positions E-I). For the remainder of this article, we outline why breastfeeding was considered

Rule-breaking act (formula-feeding)
Position A: Non-deviant if: -Non-conventional (no alternative, could not breastfeed) -Non-theoretic (mother did not know what she was doing)

Formula-feeding is excused
Position B: Deviant if: -Conventional (had an alternative, could have breastfed) -Theoretic (mother knew what she was doing) Formula-feeding is not excused (mother must account for deviance) Position C: Formula-feeding is acceptable if: -Conventional and theoretic but justified (immorality/negativity is refuted: formula-feeding presented as being as good as/better than, breastfeeding) Position D: Formula-feeding is not acceptable if: -Conventional and theoretic but not justified (immorality/negativity is not refuted) Given breastfeeding is deviant (not excused), mother should formula-feed Wider society's framing of infant feeding Local framing of infant feeding Figure 1 Deviance framing of infant feeding (adapted from Murphy 1999: 191) deviant by many of our participants and how young mothers who breastfed and those who held pro-breastfeeding perspectives accounted for their deviance. Throughout we refer to Figure 1 to demonstrate participants' navigation of this complex moral landscape.

Methods
The findings presented here are drawn from focus groups with 27 young mothers, follow-up interviews with three of these women and interviews with 5 early years professionals working at a Children's Centre in Northeast England. These data were obtained through two projects.

Project 1: a week in your life
The first project used photo elicitation and focus groups to examine adolescent mothers' health beliefs and behaviours (Hackshaw-McGeagh et al. 2018). Twenty-seven young mothers (women with at least one child by age 21) were recruited from three areas of the UK -Belfast, Bristol and Middlesbroughselected for convenience and high levels of deprivation. Participants were aged between 16 and 24 years old at the time of recruitment. Twenty-one participants had one child, four had two children and two had three children. Participants' children were between 5 years old and newborn, and three participants were pregnant at the time of the research. Participants were aged between 15 and 21 at the time of their first child's birth. Twenty-six of the participants were white British and one was mixed race. Participants were recruited through Children's Centres and organisations providing services for young parents. Participants were requested to take photographs depicting a typical week as young mothers which were intended as stimulus material for later focus group discussions and not analysed separately. Photo elicitation has been used in other marginalised populations, including young people, to reduce researcher/participant distance (Capello 2005, Leonard and McKnight 2015, Young and Barrett 2001. We used photo elicitation in much the same way as suggested by Brady and Brown (2013, 101) to engage young mothers in research through 'exciting, fun and inclusive' methods which would give our participants a degree of ownership over the project. Enabling participants to generate visual images which then formed the basis of later discussions had several advantages. Firstly, the use of photographs engendered informal discussions where asking participants to describe what was happening in images elicited a more relaxed and natural flow of conversation than without visual prompts. This informality also allowed researchers and other participants to find common ground through seemingly innocuous visual details such as (humorously) comparing the messiness of homes or complementing others' clothes. Secondly, asking participants to document their everyday lives in photographs positioned participants as the experts in the study materials and enabled them to direct the research and introduce pertinent topics outside of the original research agenda (Brady et al. 2011, Mannay et al. 2018a, 2018b. This reversal of power through using participant-generated prompts also enabled us to avoid beginning conversations with direct questioning which might have been perceived as confrontational or reminiscent of approaches used by authorities (e.g. social services) which many of the participants were wary of. Moreover, there was an epistemological synergy between our focus on everyday life as a young mother and photo elicitation methods; as Mannay et al. 2018aMannay et al. , 2018b argue, visual methods enable an 'in-depth understanding of the banal everyday elements of participants'. Following photo elicitation, participants were invited to focus group 1 where they were asked to discuss everyday life as a young mother using the photographs to guide discussion. To avoid any tension arising from sharing personal photographs in the focus group context, we gave participants several minutes at the beginning to sift out any images they did not want to share or discuss. While some participants discarded a small number of photographs where they perceived themselves looking unattractive, all participants willingly shared personal images of their homes, friends, families, children and bodies. In several instances, photographs censored at the start of focus groups were later shared once a group rapport was established.
Although a topic guide was used to facilitate discussions, conversation was mostly led by participants, focusing on topics and experiences salient to them. After focus group 1, fully anonymised transcripts were analysed using constant case comparison and deviant case analysis to identify commonalities between participants' stories and to interrogate instances where experiences differed from the norm. This initial analysis identified themes such as stigma, marginalisation, family/ network support, the influence of social media and coping mechanisms, which characterised participants' everyday lives. Following this stage of data analysis, participants were invited to a second focus group where we presented initial findings for discussion and feedback. This facilitated further data collection and refinement of the salient themes. Sixteen of the 27 original participants took part in focus group 2. Focus groups consisted of between two and six participants; took place within gatekeeper organisations; lasted between 90 and 150 minutes; and were audio recorded and transcribed by a subcontracted transcriber.
During analysis of the focus group data, breastfeeding emerged as a surprising finding and a salient experience shared by many of the women; even those who did not breastfeed discussed feeding decisions and practices with little prompting. Data pertaining specifically to breastfeeding, identified in 19 participants' accounts, were extracted for further analysis. Given that the research had not been specifically focused on breastfeeding, there were several occasions in these extracted data where follow-up questions to expand participants' reflections would have been useful but were not asked or where discussions of breastfeeding occurred as part of wider conversations about body image or food practices, rather than breastfeeding per se. Despite these limitations, given the informality of the focus groups, participants spoke openly and often at length about breastfeeding which generated significant amounts of data. As the project was not originally focused on breastfeeding, we were not guided by existing theoretical frameworks and, instead, inductively interrogated the data. We began by ascribing descriptive codes to initial cases to identify commonalities and tentatively develop explanations. This analysis of initial cases identified breastfeeding as a process of negotiating competing discourses and perspectives. What was particularly salient in these initial analyses was a widespread ambivalence towards breastfeeding. As we examined additional cases, these experiences and explanations were redefined and refined with negative cases being particularly valuable for testing emerging explanations (Katz 2001). As such, pro-breastfeeding participants' accounts were particularly useful for unpacking the discourses circulating around breastfeeding and the factors behind participants' widespread ambivalence. This refining process and negative case analysis was particularly important given that the original research was not specifically focused on breastfeeding. Through this inductive work and use of negative cases, we were able to identify deviance as a core factor shaping participants' experiences of infant feeding.
Project 2: breastfeeding Given the salience of breastfeeding as a shared experience in the focus group data, we undertook a second project to explore the particularities of young mothers' breastfeeding experiences and the tailored support available to them in the Northeast of England. We focused on the Northeast of England for convenience. We conducted in-depth interviews with three of our original participants (Jade, Georgia, Emily) with diverse breastfeeding experiences. Additionally, we invited five early years professional staff from one of the gatekeeper Children's Centres to be interviewed about the support offered to young mothers and about their perspectives on young mothers' feeding practices and experiences.
The aim of this second project was twofold. Firstly, we used follow-up interviews to test and refine our tentative theorisation around breastfeeding deviance. To do this, we returned to participants' original data and used the follow-up interview to expand their insights. We analysed the interview data thematically to explore and confirm the potential for deviance as a conceptual framing. Secondly, given that breastfeeding was identified as a key priority area for most of the gatekeeper organisations, data were also collected on the specific support offered by local organisations, which fed into site-specific recommendations. For this project, interviews took place in participants' homes (for young mothers) or at the gatekeeper organisation (for early years professionals), lasted between 60 and 90 minutes and were audio recorded and transcribed. While, for both studies, funding conditions did not permit remuneration of participants' time, participants were reimbursed their travel expenses, childcare was paid for and catering was provided for each research encounter (i.e. each focus group, each interview).

Ethics and positionality
Ethical approval was obtained from Durham University before each project commencedin December 2014 for the first study and July 2016 for the second. Given the methodology, there was an ethical tension in ensuring participants' anonymity and confidentiality while also ensuring that participants were seen and heard through the research (Wiles et al. 2012). To address this tension, at the beginning of the first project, we organised introductory briefing sessions at each of the gatekeeper organisations to discuss the project in detail with a particular emphasis on what exactly participants were agreeing to and the range of potential spaces where their visual data might be shared (see Mannay et al. 2018aMannay et al. , 2018b. In addition, we took an 'ongoing consent' approach meaning that, although written consent was obtained at the start of the research, we revisited this consent regularly by checking with participants that they were willing to continue their involvement. This 'series of permissions' enabled us to have continual conversations with participants about research progress, findings and dissemination which also engendered deeper engagement with the project (Renold et al. 2008, 429). Throughout this paper, we use pseudonyms to report participants' responses to ensure anonymity.
In addition to these ethical tensions, the focus on the lives of young mothers from areas of socioeconomic deprivation also brought dilemmas around positionality and insider/outsider status. All of the researchers on the project could be classed as 'outsiders' given that all were child-free at the time of the research and had no personal connections to the localities in which the research took place (Wigginton and Setchell 2016). Moreover, researchers being highly educated and in relatively well-paid, prestigious academic jobs contributed to this 'outsider' status and power imbalance. We were, of course, keen to negate these status and power issues for both practical (to ensure adequate recruitment and the collection of quality data) and ethical (to avoid further marginalising our participants) reasons. To do so, we adopted many of the same approaches of Wigginton and Setchell (2016) and engaged in an ongoing process of self-monitoring and reflexivity to ensure that we were framing and approaching the topic itself and participants respectfully and flexibly. We ensured, for example, that we were clear with participants that we were focused on their experiences and narratives rather than normative issues about young motherhood such as 'reasons' for conception and measuring long-term outcomes. We also paid close attention to our physical, verbal and written self-presentation to reduce the distance between researchers and participants. For example, although we initially informed participants which universities we were from, we did not mention our institutional affiliations beyond this given universities' inherent elitism and power, and their negative imagine within low participation neighbourhoods.
Despite our symbolic distance from participants, our 'outsider' status proved to be beneficial as it allowed researchers to ask na€ ıve questions and deconstruct taken-for-granted experiences which might have been missed by researchers coming from an 'insider' position (see Hayfield and Huxley 2015). Moreover, our outsider naivety also cemented participants' status as experts in the research topic which negated potential power imbalances and enabled participants to direct the research conversations to topics pertinent to their experiences. Moreover, our 'outsider' status was not all-encompassing and there were several points of commonality such as relatively close age (all researchers were in their late 20s or early 30s), sex and researchers' future plans for becoming mothers. This enabled a flexible conversation, often tangential to the core research focus, which built rapport with participants and facilitated data collection.

Findings
Breastfeeding as deviant Our participants found themselves in a quandary around infant feeding. On the one hand, participants were well aware of the nutritional benefits of breastfeeding. Given this awareness, some participants who formula-fed engaged in similar justificatory work to Murphy's (1999) participants. For example, they described being physically unable to breastfeed, challenged the 'breast is best' assumption, located breastfeeding as incompatible with work or education and expressed concerns about the quality of breast milk. These discourses can be read as attempts to challenge the conventionality and immorality of formula feeding and neutralise potential deviance (i.e. to occupy Positions A or C, Figure 1 where formula feeding is acceptable).
On the other hand, for many participants, the moral framing of breastfeeding as 'best' did not have the strong purchase that Murphy suggests. While the 'breast is best' rhetoric was certainly ubiquitous, for many of our participants, it was competing equally with, and commonly losing out to, discourses and perspectives framing breastfeeding as a problematic transgression of bodily behaviours. As such our participants were caught between two deviancesformula feeding framed as deviant by wider society and medical literature and breastfeeding framed as deviant within their local areas and social networks. For most participants, this latter deviance attached to breastfeeding within everyday spaces and interactions took primacy in shaping feeding decisions. In other words, it was breastfeeding which was understood as deviant and required accounting for. In contrast to Murphy's participants' justification of breastfeeding (Murphy 1999), which centred on accounting for the potential consequences of breastfeeding, for our participants, it was breastfeeding itself which required justification.
Our data showed that breastfeeding was considered deviant on three 'levels'; it was considered (i) a deviation from broad social norms about women's bodies, (ii) a deviation from local normative mothering behaviours and (iii) a transgression within micro-level interpersonal and familial relationships. Given that the 'rules' governing expectations of women's bodies at these three 'levels' were well known by participants, the act of breastfeeding can be understood to be theoretic; participants who breastfed did so despite knowing it was 'questionable' behaviour (McHugh 1970). Additionally, given that infant formula was available to participants as a more acceptable alternative (albeit one which might also require some justification), breastfeeding also fulfilled the condition of conventionality. As such, using McHugh's (1970) and Murphy's (1999) frameworks, we can understand breastfeeding, for our participants, as a form of deviance occupying position E in Figure 1.
For most participants from whom breastfeeding data were collected (n = 19 of 27 participants), this multifaceted, multilayered deviance of breastfeeding contributed to ambivalence towards, or a decision to forgo, breastfeeding. The positioning of breastfeeding as deviant required participants who breastfed (n = 3) to account for their behaviour. Also notable was that participants who were pro-breastfeeding but unable to do so for their own children (n = 4) also appeared to feel obliged to account for their pro-breastfeeding perspective. Below we elaborate on participants' construction of breastfeeding as deviant at the three 'levels' mentioned before describing the accounting work undertaken by breastfeeding participants to justify their perspective and behaviours (to occupy position H in Figure 1).
Deviance from societal norms about women's bodies: breastfeeding as unnatural Firstly, a number of participants located breastfeeding as an act which deviates from societal norms and expectations of women's bodies. Within this framing, participants' constructed women's breasts in terms of femininity and sexuality, with their primary purpose understood as related to men's pleasure and heterosexual desire. Emma, who breastfed, was aware of this dominant sexual framing and made her decision to breastfeed within this context: Emma: I think quite a lot of people see boobs as more of a sexual thing instead of a natural thing This sexual framing created ambiguity over 'what breasts are really for' (Carter 1996). For some, breastfeeding was understood as 'weird' (Emily) because breastfed babies were assumed to enter the physical and symbolic space of heterosexual intimacy. This sexualisation created a dilemma for participants around infant feeding where the predominant norms governing appropriate (sexual) functions of breasts had to be weighed against the health benefits of breastfeeding. For most participants, their internalised sexual norms surrounding women's bodies took primacy in feeding decisions and they avoided the rule-breaking behaviour of breastfeeding altogether. Rebecca explained: Rebecca: I just don't think it's that natural, I know it is natural so that sounds silly, but to me. . .I think the public eye nowadays has made boobs a sexual object, well men have haven't they in magazines and page three and everything else? And I see them, I know they're not there for men, but I think that is how today's society, that boobs are there for men to look at whereas that isn't actually what they're there for but that is what I saw it as. I thought I don't want my child sucking on my boob, that sounds really silly, but that was my feeling Within this comment, Rebecca displayed knowledge of the 'rules' around breastfeedingshe knew what the rules were (breasts are sexual), where they came from (the media) and for/by whom they were (re)produced (society, specifically men). Such knowledge of the rules leaves little room for manoeuvre and limited capacity for claiming non-theoreticity. In other words, given that the rules were clear and all-encompassing, it would be unfeasible for Rebecca to claim she did not know what she was doing and occupy Position F in Figure 1. Rebecca's conceptualisation of breastfeeding as 'unnatural' is powerful here and locates breastfeeding as implicitly deviant and against some kind of 'natural order'. Such discourses around the immorality of 'unnatural' acts serve as powerful devices to condemn particular beliefs and behaviours with limited room for resistance (Takala 2004). Rebecca, however, seems to be using 'natural' primarily to imply ordinariness and normality rather than biological naturalness. Indeed, in the above comment, after condemning breastfeeding as socially unnatural, she immediately went on to acknowledge its biological naturalness and the flawed logic of her own perspective. This flexibility of 'naturalness' as a discourse in infant feeding decisions is echoed in Murphy's work and highlights well the quandary at its centre (Murphy 1999, 195).
Also present within this dilemma were concerns about the necessity of public breastfeeding, which was assumed by many participants to be conspicuous and involve bodily exposure, as Faith and Hailey described: Hailey: I didn't want to like walk around and just get my boob out [laughter] Faith: I didn't think I could do it because I didn't want to go out and just breastfeed in front of everybody, I'm not comfortable with that Within this concern about public breastfeeding is some anxiety over deviation from normative societal expectations of women's bodily behaviours, both in terms of breastfeeding itself as an unnatural and sexually ambiguous activity and public breastfeeding as a transgression of women's expected modesty. Previous research has repeatedly shown that sexualised norms surrounding breasts complicate breastfeeding and necessitate a degree of negotiation and management for women from across a diversity of groups (Avishai 2007, Grant 2016). Indeed, for Murphy's participants, assurances of discreet feeding were a key aspect of their accounting for breastfeeding. Others have, more specifically, argued that young mothers are particularly likely to internalise these norms and forgo breastfeeding (Johnston-Robledo and Fred 2008). This was evident in many of our participants' stories, such as Faith and Hailey, where they described eschewing breastfeeding in order to avoid deviance and adhere to dominant norms about women's bodies.
Deviance from local behavioural norms: breastfeeding as abnormal Breastfeeding was also located as deviating from normative mothering behaviours inscribed and entrenched within local neighbourhoods (Condon et al. 2013). For many participants, this deviation from normative mothering behaviour was understood to be part of wider contemporary parenting practices in which breastfeeding is increasingly uncommon: Izzi: It's not normal now. . . I was at a pub somewhere and we were having lunch and this little old lady in a wheelchair, someone brought her over to see, and she actually said to me how nice it is to see a young person breastfeeding. . .Back when she was probably having kids that was the normal For a number of participants, the rarity of breastfeeding was particularly notable in their local areas, which Jade linked to the high rate of teenage pregnancy and other deprivations ('everything like that'): The early years professionals we interviewed echoed this localised framing of mothering behaviour, positioning low rates of breastfeeding in the Northeast of England as a symptom, and outcome, of multiple socioeconomic marginalities. Poorly paid employment, high rates of debt and poor education around key aspects of 'survival' (Denise) like nutrition, cooking and financial management were perceived as particularly problematic across the local area. While Children's Centre staff referenced wider socioeconomic inequality, behaviours associated with these marginalities, such as not breastfeeding, were also located as a failure of individual responsibility. In the case of infant feeding, the early years professionals perceived young mothers' apathy towards breastfeeding and related support services as more pertinent than, though influenced by, wider socioeconomic inequalities: Samantha: I don't think there's the money in local services to give the breastfeeding support and education they need and really they don't make any effort to find out for themselves or educate themselves about it This quote demonstrates the interplay between structural inequality and personal responsibility which fed into the understanding of breastfeeding as an abnormal deviation from local norms. For organisation staff, these local norms around breastfeeding were produced in three ways; (i) through the reproduction and reinforcement of bottle feeding as the default feeding approach within families and friendship networks, (ii) as a result of inadequate breastfeeding education locally and (iii) the invisibility of breastfeeding in everyday, non-medical local spaces such as parks or shops. This final point is particularly noteworthy for two reasons. Firstly, human geographers have drawn attention to the importance of visible breastfeeding in public, non-medical spaces as a way to 'break the taboo' surrounding it (Boyer 2011). This relies, however, on the availability of appropriate spaces and facilities for public breastfeeding which is much less likely in areas of socioeconomic deprivation (McFadden and Toole 2006).
Secondly, the local visibility of breastfeeding only in medical spaces, and within them most commonly through formal medical literature, was problematic for many participants who almost ubiquitously described fraught relationships with healthcare practitioners. As such, local medical spaces were often experienced as contentious and ambiguous for young mothers arising from perceptions of being judged and patronised. Daisy highlighted this medical paternalism well: In other words, health professionals reproducing the 'breast is best' rhetoric and handing out literature might be read, in Foucauldian terms, as another way that young mothers are positioned as docile bodies requiring submission to a clinical gaze. Emily's comment claiming equivalency between breast milk and formula echoes formula-feeding mothers in Murphy's work who justified their deviance by challenging the nutritional superiority of breast milk through 'denials of injury' (Scott and Lyman 1963). A similar, albeit reversed, process might be seen here in Emily's comment which introduced conventionality, and thus deviance, to the decision to breastfeed. By suggesting that formula and breast milk are equally beneficial she left little space for breastfeeding women to justify their deviance through the 'breast is best' rhetoric. Moreover, her comment not only challenged the validity of the 'breast is best' notion in and of itself but also located it as a form of problematic medical paternalism, rendering breastfeeding questionable in two different but related ways.
Deviance from interpersonal interaction expectations: breastfeeding as invisible Given its abnormality in the participants' local areas, breastfeeding was largely invisible within their everyday lives and interpersonal interactions. As such, breastfeeding was described by participants as a 'daring' deviation from expected behaviours within families and friendship groups: Emma: My friend came into my house, sat on my sofa and she was talking to me but looking at the ceiling. . .and I said, "what's the issue?" And she goes, "mate, I'm going to have to look at your boobs because I can't cope", and she had to have a look to be able to get over it Jade: My dad was a bit awkward about it, he'd go off for a bit. . . But I was like "it's going to happen. . .this is the new normal" This invisibility of breastfeeding within interpersonal relationships meant that many participants lacked everyday practical and emotional breastfeeding support. In some instances, such as those described by Jade above, in which friends were 'quite arsy', this lack of support might be understood as a deliberate challenge to participants' deviance where actively being unsupportive of breastfeeding becomes a way to problematise deviance and speed up cessation. In other instances, however, this lack of support might have stemmed from an absence of breastfeeding knowledge and experience within social groups. Several participants, such as Emily, described problems establishing breastfeeding and not having friends or family members to ask for help: Emily: You're not going to ask a man [partner] to help you with breastfeeding. And my mum never breastfed any of her five children so she wouldn't have known This lack of kinship support is in contrast to the experiences of middle class mothers of nonadolescent age who receive breastfeeding support from family and friends and, in some cases, new friendships are developed through breastfeeding support (e.g. Nolan 1997). For many of our participants, however, the unavailability of practical support from friends and family meant that difficulties establishing breastfeeding early on led to breastfeeding cessation. In this sense, eschewing breastfeeding or early cessation emerged from a lack of support, which was itself rooted in understandings of breastfeeding as deviant. In other words, many participants lacked breastfeeding support because their friends and family had largely opted to formula-feed as a result of the dominant norms which locate(d) breastfeeding as deviant.
Accounting for deviance Given that breastfeeding participants, and those who held pro-breastfeeding perspectives, could not legitimately claim non-theoreticity and non-conventionality, breastfeeding was predominantly positioned as deviant occupying position E in Figure 1. As a 'daring' transgression of multiple norms and expectations, young mothers had to account for their breastfeeding behaviours and perspectives. To do so, pro-breastfeeding participants, including those who held pro-breastfeeding perspectives but were unable themselves to breastfeed, developed a number of strategies to move breastfeeding to position H in Figure 1. Firstly, breastfeeding participants repeatedly mobilised an adapted 'breast is best' discourse. This adapted discourse foregrounded breastfeeding as the 'best' option in the context of everyday family life:

Jade:
It was right for me and my baby. . .I found breastfeeding was easier for Sophie but I couldn't breastfeed with Harry. He was a bottle baby. Now I look back I think how did I do bottle-feeding? Because it was so draining. I think I'm more, with my sleep now, I'm just like go on the boob. Now she's nearly one I still do exactly the same. Neil [partner] said "I thought we were getting her off the boob". . .I'm like "it's three o'clock in the morning darling, shut up. She is screaming. Let's go to sleep" Giving primacy to these practical benefits of breastfeeding within everyday family life could be read as an attempt to temper its deviance. In positioning breastfeeding as practical, even selfish, participants were able to leave some ambiguity around their true commitment to breastfeeding and, as such, the degree to which they were truly deviant. This framing enabled participants some flexibility over how far they presented themselves as deviating from expected behaviours. This meant that breastfeeding participants were able to present a full commitment to breastfeeding where this was necessary or expected (e.g. in medical consultations) but also distance themselves from breastfeeding commitment in situations where breastfeeding was particularly deviant (e.g. with friends). In doing so, participants were not challenging their deviance by claiming non-conventionality or non-theoreticity or by refuting the dubiousness of their behaviour (i.e. occupying Position F in Figure 1); indeed, the theoreticity of their behaviour is front-and-centre. Rather, through this kind of justification, breastfeeding participants could minimise their infant feeding decisions whereby the breast or formula dilemma was framed as less about body or gender politics and instead a question of what is easiest. This is not, of course, without risk. By portraying the decision to breastfeed as primarily based on what is easiest for her, a mother opens herself up to the charge of being selfish and, by implication, a bad mother. This charge, however, may be offset by the purported health benefits of breastfeeding and so position I in Figure 1, where deviance is sanctioned and formula feeding is the logical alternative, can be avoided. In addition to breastfeeding participants, those who were pro-breastfeeding but unable to breastfeed also undertook a degree of accounting work to justify their perspective. Their justificatory work centred on dismissal and minimisation of anti-breastfeeding discourses, particularly those related to public feeding, and positioning breastfeeding as common sense. In a focus group interaction, Chloe (who was unable to initiate breastfeeding) and Daisy (who could not breastfeed for as long as she hoped) articulated this: While Chloe and Daisy demonstrated a degree of solidarity with those undertaking the deviant act of breastfeeding, it was not clear from the data how far they would, or did, take this probreastfeeding solidarity. Being pro-breastfeeding but not actually engaged in the act of breastfeeding allowed these participants a degree of flexibility in their support for breastfeeding. It is not clear, for example, whether Daisy commented 'are you mad or what?' publicly on Facebook thus outing herself as a supporter of breastfeeding deviance, or kept this thought to herself.
Moreover, the context of both of the above conversations (in Jade's case a focus group involving two women who did not breastfeed, and in Chloe and Daisy's case a focus group involving Emma who was adamantly pro-breastfeeding) should also be borne in mind as a factor potentially influencing their reflections. Jade's use of an adapted 'breast is best' discourse which centres the selfishness of breastfeeding may have been an attempt to distance herself from any implications of her being a 'better' mother for following medical advice. In other words, by reporting doing the 'right thing' by breastfeeding (according to medical advice) but for the 'wrong reasons' of ease and selfishness, Jade may have been attempting to validate the experiences of her focus group counterparts who did not breastfeed and so were at risk of charges of poor mothering (occupying Position B in Figure 1). In the latter case, Daisy and Chloe's comments about breastfeeding solidarity also cannot be easily separated from the focus group context in which Emma repeatedly expressed her pro-breastfeeding stance. In both cases, the importance of flexibility in a pro-breastfeeding stance becomes apparent as participants were able to move in and out of deviance according to the particular situation. As such, the accounting work of participants who were pro-breastfeeding but unable to do so is contingent and highly context dependent.
Participants' accounting work sometimes extended deeper than justifying their personal breastfeeding beliefs and behaviours to addressing the underlying norms which position breastfeeding as deviant in the first place (questioning the basis of Position E in Figure 1). In doing this, the theoreticity and conventionality of breastfeeding are not addressed and become largely irrelevant as they stem from the rules themselves which are being questioned. For example, Emma and Jade in particular described questioning the dominant sexualised framings of Emma: A load of my partners friends used to come round and obviously be males, I used to feed in front of them and I changed their view on things. It changed them from thinking that they were a sexual object. . .. When one's just gone on to have their first child and he actually was trying to get her [mother of the child] to breastfeed and was like, "no you need to do it and it is good for the baby". . . He went from thinking, oh "yes, sex", to "actually, no they're for babies" As well as challenging these norms informally, Jade, at the time of one-to-one interview, was in the process of formalising this by becoming a breastfeeding peer supporter and had also been accepted on to a midwifery course at university.

Discussion and conclusions
This article has examined the infant feeding decisions of young mothers from socioeconomically deprived areas. While infant feeding decisions are a 'moral minefield' for all women (Murphy 1999), it has been shown that infant feeding is a particularly challenging terrain for mothers from marginalised contexts to traverse. For our participants, specific issues of age, class and locality intersected with broader discourses of bodily performance, sexuality and femininity producing competing framings of breastfeeding. In this context, the moral superiority of breastfeeding is not as abiding as Murphy (1999) and others (e.g. Wall 2001) have suggested. Rather, while adolescent mothers are aware of and mostly buy into the 'breast is best' discourse, localised perspectives positioning breastfeeding, rather than formula feeding, as abnormal are more salient in young mothers' feeding decisions (Condon et al. 2013). We have demonstrated that this deviance of breastfeeding is constructed at three intersecting 'levels' of societal norms, locality and neighbourhood and interpersonal interactions. At each, breastfeeding is understood as a transgression of women's expected bodily behaviours and young mothers who breastfeed are, thus, required to account for their deviant behaviour. We have demonstrated that our participants did this through mobilising an adapted 'breast is best' rhetoric where 'best' was extended beyond the nutritional superiority of breastfeeding, through minimising anti-breastfeeding discourses and through challenging the societal norms about women's bodies upon which breastfeeding deviance rests. In making this argument, the article contributes to the extant social science literature on women's feeding decisions. Most pertinently, we have offered an extension to Murphy's (1999) framework of infant feeding deviance, representing the complexity of this landscape for marginalised women. While our extension of Murphy's framework occupies positions E-I in Figure 1 (following Murphy's model in positions A-D), this does not mean breastfeeding deviance was a secondary phenomenon for our participants. Indeed, our participants' navigation of the infant feeding landscape most commonly started at Position E where 'breastfeeding as deviant' was the default within local and interpersonal cultures and navigation of broader social discourses around formula feeding (as per Murphy) was secondary. Our extension to Murphy's model also demonstrates a degree of circularity in infant feeding decisions where women are always rendered accountable given the liminal and ambiguous position of motherhood more generally.
The article also contributes more widely to discussions of infant feeding. While social scientists, particularly those approaching breastfeeding from a feminist perspective, have centralised the voices of mothers themselves, the breastfeeding experiences of women from marginalised contexts have not been as visible. Instead, breastfeeding experiences which transcend social, economic and cultural categories (e.g. public breastfeeding, extended breastfeeding) have been more readily explored. Within this literature, structural factors mediating breastfeeding experiences have been somewhat subsumed within an approach understanding breastfeeding as primarily gendered, rather than equally classed, racialised or framed by other marginalities like age. In centralising the experiences of adolescent mothers from areas of socioeconomic deprivation, we have demonstrated the ways in which multiple factors and structural inequalities coalesce to create ambiguity around infant feeding beyond those shared with less marginalised women. We have focused here on the intersections between, and impacts of, adolescence and socioeconomic marginality. However, all but one of our participants were white and all were able-bodied, meaning the ways ethnicity and disability may also frame breastfeeding could not be examined. Future research would benefit from focusing on the experiences of adolescent mothers with multiple marginalities, exploring the ways that diverse inequalities structure the infant feeding landscape and may add additional or amended factors to the model presented here. Moreover, it should be borne in mind that the first project from which these data were drawn did not specifically focus on breastfeeding but young mothers' health behaviours and experiences more generally. This meant that discussions of breastfeeding often occurred within wider conversations about issues such as body image or food behaviours which may have limited the depth of participants' accounts of breastfeeding specifically and opportunities for follow-up questions being missed. Conversely, the holistic focus on health in 'everyday life' created an open space for sharing otherwise unfavourable perspectives about a range of topics (including breastfeeding) because the research was not perceived as having a particular agenda. In other words, had the first study been focused on breastfeeding, participants may have been reticent about sharing links with deviance given the societal traction of the 'bottle feeding as deviance' rhetoric.
The article also contributes to work highlighting the importance of locality and place in reproducing norms around breastfeeding behaviours (e.g. Boyer 2011). In one way, we have shown that local area is important as a 'milieu' within which norms and expectations circulate through organisations and interpersonal networks. This focus on locality builds on the work of Condon et al. (2013) who suggest that young mothers' geographically bound social worlds create a culture where breastfeeding is questionable. Additionally, we have touched upon the idea that the physical space of a socioeconomically deprived area also reproduces infant feeding practices, drawing particular attention to the invisibility of breastfeeding outside of medical spaces, which are themselves sites of tension. This area of research warrants further work to explore the ways that local physical spaces are perceived, produced and take on meaning in adolescent mothers' everyday lives generally and breastfeeding practices specifically. Research of this kind lends itself to innovative, non-static research methodologies, such as walking interviews, which actively engage researchers and participants in the physical world under discussion (Jones et al. 2008).
We have relied in this article on verbal, mostly retrospective, accounts of experiences of infant feeding decisions and breastfeeding, which carry risks of participants misremembering or reconstructing narratives. Future research would benefit from working with adolescent and marginalised mothers throughout pregnancy and beyond to capture how breastfeeding is iteratively and continually constructed. This work would further elaborate and temporally locate the circularity of infant feeding decisions, perspectives and behaviours. This would, again, lend itself to innovative methodological approaches such as diary keeping or blogging, allowing researchers to track shifts in thinking about breastfeeding over time. Through the use of such methodological approaches, researchers may better capture the embodied and corporeal experience of breastfeeding, which was not possible in this current research. This approach would also embed research on breastfeeding within a wider holistic focus on women's embodied reproductive experiences rather than marking breastfeeding out as an extraordinary phenomenon for examination, which can itself oblige some women to undertake justificatory work (Amir and Cwikel 2005).
This article also offers some useful starting points for considering better breastfeeding support for adolescent mothers. The systemic roots of breastfeeding's location as deviance means arriving at a set of recommendations for practitioners is not easy. The issue does not lie with individual women but with communities and patriarchal society more widely. As such, interventions need to address the underlying assumptions about women's bodies and sexuality which framed our participants' approaches to breastfeeding. While continuous education about women's bodies may go some way to desexualising them, doing such work in a patriarchal culture characterised by ubiquity of sexualised images of women and easy access to pornography presents significant challenges. Moreover, unrelenting austerity in the UK makes funding for such projects and interventions precarious.

Data availability
The data that support the findings of this study are available on request from the corresponding author. The data are not publicly available due to privacy or ethical restrictions.