“ YOU GET ANGRY INSIDE YOURSELF ” : LOW-INCOME ADOLESCENT SOUTH AFRICAN GIRLS ’ SUBJECTIVE EXPERIENCE OF DEPRESSION

Informed by the feminist social constructionist approach this study aimed at exploring the subjective experiences of depression of low-income South African adolescent girls. Participants in this study (girls between the ages of 12 and 14) live in a semi-rural low-income coloured community in the Western Cape. Participants were familiar with the concept of depression, but it seemed that for them the central emotion associated with depression was anger, which often manifested in destructive behaviours. Furthermore, participants seemed to construct depression as a relational problem, suggesting that psychotherapy may be indicated as an important intervention strategy.


INTRODUCTION
Various studies have noted a gender difference in depression, arguing that depression tends to be up to two times more common in women than in men (Accortt, Freeman & Allen, 2008;Fan & Ji, 2004;Noble, 2005). Noble (2005) claims that this ratio is found across various countries, cultures, and ethnicities. Depression has been found to be one of the most common mental health problems afflicting women (Sides-Moore & Tochkov, 2011).
The impact of depression on children and adolescents has traditionally been understated, due to the erroneous belief that children could not become depressed (Mash & Wolfe, 2013). However, it has been documented that up to 5% of all children, and between 10% and 20% of adolescents, are or have been suffering from serious depression (Avenevoli, Knight, Kessler & Merikangas, 2008). Depression in young people is thus an important concern for the mental health profession (Mash & Wolfe, 2013).
Longitudinal studies investigating depression throughout adolescence have consistently found linear increases in symptoms of depression with age, and a significant gender difference regarding these increases: "…samples of females demonstrated greater increases in symptoms than mixed-sex or all-male samples" (Olino, Stepp, Keenan, Loeber & Hipwell, 2013:1). Studies have also consistently found that depression in adolescent girls after puberty is already two times more common than that in adolescent boys after puberty (Thapar, Collishaw, Pine & Thapar, 2012).
Given the historical lack of emphasis in research on depressed youth, this feminist social constructionist study examines how young females themselves experience depression or their exposure to depression, specifically also investigating the underlying discourses.

METHODOLOGY
The current study forms part of a larger qualitative long-term research project aimed at exploring the emotional experiences of women living in a low-income semi-rural community within South Africa (Kruger, 2014). The larger project, as well as the current study, is informed by a feminist social constructionist perspective.
For the current study, following the principles of convenience sampling (Bless, Higson-Smith & Sithole, 2013;Riley, Wood, Clark, Wilkie & Szivas, 2000), coloured 1 girls (between the ages of 12 and 14) attending a local high school were invited to participate in a psychoeducational/psychodynamic group intervention (Jewkes, Gevers, Cupp, Mathews, Russell, Lefleur-Bellerose & Flisher, 2011;Smith, van Wyk & Alkana, 2013;Way, 1998;Yalom & Leszcz, 2005). The girls who expressed an interest and handed in their consent forms on time were divided into two small groups, with each group being facilitated by two postgraduate psychology students, and two of the Masters students in Clinical Psychology.
The participants took part in 14 group sessions, but for the current study the focus is mainly on data collected during a session where the topic of discussion specifically was "depression". During this particular session on depression, the questions about depression were kept open-ended and the discussion informal (Miles & Gilbert, 2005). Girls were encouraged to free associate, with the emphasis being on the participants' subjective experiences of depression. As such, the discussion simulated a focus group discussion (Babbie, 2013) in which participants were encouraged to free associate.
After transcription of the audio-taped group sessions, data was analysed using social constructionist grounded theory (Charmaz, 1995), a data analytic approach consistent with social constructionist feminism. In both these approaches the focus is on the subjective accounts, experiences, and perspectives of the participants (Charmaz, 1995). The four steps of data analysis included data collection (the creation of rich data through open-ended interviews); coding (which included line by line, open-ended searches for codes); the focused recoding of data according to categories which emerged from the first step of coding; an examination of the relationships between these categories; memorandum-writing (the personal reflection of researchers); and finally, theorybuilding (a conceptual analysis informed by the current analysis and by existing scholarly literature on depression in South African adolescent girls) (Charmaz, 2008(Charmaz, , 2014. In terms of the principle of triangulation (Yardley, 2008), the developing codes and the interpretation of the results were discussed with the whole research team and experts in the field. The analysis was also deepened by referring to other research studies, ethnographic observations and clinical work conducted in this community by the authors over a period of 14 years. To allow the reader to formulate his or her own opinions on the researchers' interpretation of the data (Potter, 1996), the researchers are providing quite lengthy verbatim quotations.
The study was approved by the Research Ethics Committee (Humanities) at the University of Stellenbosch.

RESULTS AND DISCUSSION
In the process of social constructionist grounded theory, emphasis is placed on the reciprocal and mutual relationship between data analysis and the developing theory (De Vos, Strydom, Fouché & Delport, 2011). Results are therefore presented and discussed simultaneously.
Following line-by-line and focused coding, five main categories emerged:  Depression as a disturbing topic;  Depression as a complex, but known phenomenon manifesting in a variety of feelings, thoughts, and behaviours;  Depression as relational construct;  Depression as gendered;  The healing power of relationship.

Depression as disturbing topic
Firstly, it was found that participants were very uncomfortable with the topic of depression, with the discomfort manifesting both physically and emotionally. During the sessions in which depression was discussed, participants initially seemed to be unable to talk about the topic of depression, focusing rather on feeling hungry, feeling tired, and feeling cold. There was also a lot of silence, uncomfortable coughing, clearing of throats and gigglingmuch more than usual. For instance, one participant, rather suddenly, stated: Participant 5: "It would be very nice now to eat a muffin or McDonalds".
When directly asked about this apparent discomfort, participants responded to facilitators in monosyllables rather than in full sentences, using words such as "bad" and "uncomfortable".
Later on in the session participants were able to articulate their discomfort with the topic more explicitly: Participant 1: "Can't speak out about it…I think it's a private thing… (sigh) I don't know anymore".
Participant 6: "I would become shy to speak…I would be shy to speak".
Participant 7: "It doesn't want to come out". While some of the discomfort with the discussion is acted out during the session, participants are also able to give reasons for the discomfort: it is difficult to find the right words; it is an embarrassing topic; they are not sure whether they can trust the space; and, finally, they are also aware of the fact that such a discussion potentially can cause them to "think back to these happenings".
The literature also suggests that talking about depression is difficult, specifically for children with depressed mothers (Focht-Birkerts & Beardslee, 2000). Pennebaker, Zech and Rimé (2001) describe how, when an emotional thought is accessed, elements of the corresponding emotional reaction can also be activated. This includes physical, experiential, and sensory elements, as well as memories of emotions and emotional experiences (Pennebaker et al., 2001). Certain emotions have also been argued to trigger specific bodily reactions (Nummenmaa, Glerean, Hari & Hietanen, 2013).
Their discomfort about the topic of depression and their initial reluctance to speak about it did not, however, mean that the girls did not know depression. Subsequent to their initial avoidance of the topic, it was clear that the girls have been exposed to depression and were very able to describe the feelings, thoughts, and behaviours associated with it. Their understanding of this phenomenon, however, differed from traditional notions of depression.

Depression as multi-faceted phenomenon manifesting in feelings, thoughts, and behaviours i) Feelings of a depressed person
In a free association exercise, participants predominantly associated depression with emotions of anger and rage. The terms "anger" and "rage" were always what came up first (note the monosyllabic responses): Facilitator 1: "Ok, what do you think when I say depression?" Participant: "Angry".
Facilitator 2: "What do people say when someone is depressed?" Participant 2: "I think people would say: 'Why is she so angry all the time'".
As they themselves anticipated, talking about the anger seemed to trigger angry feelingsand the participants then proceeded to talk about their own anger, who they are angry with, what they are angry about, and how they express anger. Depression is typically associated with sadness (Fivush & Buckner, 2000), but the relationship between anger and depression is less clear. Studies suggest that up to 40% of patients diagnosed with Major Depressive Disorder experience intense anger (Abi-Habib & Luyten, 2013;Vliegen & Luyten, 2008), but in formal diagnostic systems anger is not regarded as a symptom of depression (American Psychiatric Association, 2013;Vliegen & Luyten, 2008).
Researchers have different ideas about the role of anger in depression. It has been suggested that women's persistent anger may play a dominant role in their higher rates of depression (Simon & Lively, 2010). Kruger, Van Straaten, Taylor, Lourens and Dukas (2014), doing research in the same community as the participants, found that depressed mothers often experience severe anger, often expressing their anger through aggressive behaviour towards their children. While the anger and aggression reportedly caused these mothers great distress , it is highly likely that some of the girls in our study have experienced the rage of a depressed mother.
While participants spoke quite openly about anger and rage, they did not talk much about sadness.

ii) Thoughts of a depressed person
Participants seemed to suggest that depression is associated with negative thoughts. The most prominent negative thought that was identified by participants was that depressed people are perceived negatively by others. Depression seemed like something one should feel ashamed about: Participant 9: "Its almost as if people say they're stupid, they're crazy…".
The emphasis on what other "people think of me" rather than on what the depressed person thinks of herself is interesting. This supports other research in this community that suggests that people from this community feel as if they are under constant surveillance, being watched and judged (Fleming & Kruger, 2013;Hamman & Kruger, in press;Kruger, in press). Fleming and Kruger (2013:114) state that "while shame leads to silence, there is another dangerif you do speak about your shame or about what you are ashamed about, you may be shamed again". This fear of being "shamed again" (Fleming & Kruger, 2013:114) by the community may add to a fear of judgment and exposure. The girls in our study seem to suggest that this feeling of being under surveillance, specifically when one is depressed, is very prominent in the experience of depression. It also suggests the importance of relationships (specifically the threat to relationship) in the experience of depression.
Other researchers have also found that the sense of stigma surrounding depression has led to feelings of shame, embarrassment, and fear of others' judgment in those suffering from depression (Gask, Aseem, Waquas & Waheed, 2011). Rose, Joe and Lindsey (2011) found that higher perceived stigma was found in teenage girls who were more depressed.

iii) Behaviours of a depressed person
It was clear that the participants associated depression with acting out, specifically destructive acting out such as attempting suicide or committing suicide (by hanging oneself or slitting one's wrists), self-destructive behaviours (such as alcohol abuse), destructive behaviours (such as breaking things) and defiant behaviours (such as refusing to clean one's house): Participant 4: (Giggles) "She just ran away and and uh her alcohol outside was a problem, drank her alcohol".
Participant 1: "She asked if there was someone who could help her".
Participant 2: "She didn't want to clean her house".
Participant 3: "And throw and so…". Anecdotal evidence suggests that suicide rates at the local high school are particularly high. In the foyer of the high school, in a display case celebrating sport and cultural achievers, there also is a macabre gallery containing photographs of scholars (all of them female) who have committed suicide. Not surprisingly, then, participants seemed to be very familiar with suicide, a topic that came up frequently during these sessions in various ways.
Participant 3: "She mustn't kill herself…I had a friend that said that. And then I told her that she mustn't listen to others who tell her something like that….She told herself that she wanted to kill herself".
Participant 7: "And there are girls at school who drink pills because of boyfriends…Even now, but all the time they just say pills pills pills…And then she slit her wrists, probably sad".
This discussion on depression and the destructive behaviours associated with it also led to other accounts of violence in the community: Participant 5: "Its awful…I don't know what the guy did for her, for them. It was up here in the mountain man. Then they argued about drugs, then the guy hung her up with barbed wire" (giggles). All of these acting out behaviors seem angry or aggressive, and tie in with participants' descriptions of a depressed person as feeling angry. However, very prominent in the above quotes is also a sense of the depressed person withdrawing, withdrawing from her household tasks, withdrawing by drinking -"she just ran away".
One participant, in the middle of the discussion on the angry acting out behaviours, very poignantly said: Participant 3: "Can I say? She just stayed quiet and didn't talk about anything".
Depression, as constructed by the girls, seem to range from emotions of anger to sadness, from destructive acting out to quiet withdrawal. Again these descriptions seem in line with studies conducted with adult women in this and other similar communities, where both the experiences of sadness and badness emerged as prominent properties in the experience of depression (Dukas, 2014;Kruger et al., 2014;Lourens & Kruger, 2013). Lourens and Kruger (2013) describe a range of feelings and behaviours reported by depressed women. These included sad feelings and behaviours, such as feeling down, isolated and withdrawn, and behaviours considered as social withdrawal or as self-injurious. These also included bad feelings and behaviours, such as feelings of rage, guilt and stress, and behaviours such as smoking, drinking, and externalizing anger (Lourens & Kruger, 2013). Dukas mentions feelings such as "hopelessness" and "mortification" (2014:118) in discussing the subjective experience of a depressed woman's shame; whereas Kruger et al. mention "anger and hostility" (2014:465) as well as "violent fantasies, violent behaviour" (2014:466) as being associated with women's experiences of being depressed.

Depression as a relational construct
When asked about the causes of depression, participants clearly constructed depression as a psychosocial, or more specifically, a relational phenomenon. Participants did not mention a single biological cause of depression. The girls seemed to think that childhood depression is caused by relationships or the loss of relationships at home or at school.

i) Relationships at home
The dominant cause of depression was described as problems at home, including domestic violence: Participant 2: "She has house problems… Your dad probably drinks and then he hits your mom".
Participants also specified mothers or their relationships with their mothers as causing depression. Often it was the anger and/or aggression of mothers that was seen to cause depression: Participant 6: "If I maybe break a plate or saucer…".
Participant 4: (Giggling) "She has problems at home…Because her mom hits her like this and she can't be late".
Participant 6: "A falling out with your mom".
It also became clear that participants viewed an inability to communicate important things with their mothers as a cause of sadness: Participant 8: "But she's a stepmom…all stepmoms are evil". Mirowsky and Ross (2003) state that women and children from lower-income homes are more likely to be depressed than those from higher-income homes. Beck and Alford (2009) argue that problems at home may result in chronic stress for the children and that this chronic stress, combined with poor parent-child relationships, increases the children's risk of developing depression. Furthermore, domestic violence has been found to have a negative impact on mental health, specifically on the development of depression and anxiety in mothers and children (Ulrich, Cain, Sugg, Rivara, Rubanowice & Thompson, 2003;Walby, 2004). Adverse childhood experiences, such as domestic violence, also increase the risk for an individual becoming depressed (American Psychiatric Association, 2013;Holt, Buckley & Whelan, 2008).
Studies have shown that homes with a depressed mother are more violent, hostile, and filled with conflict, than those without a depressed parent (Kohl, Kagotho & Dixon, 2011;Silk, Shaw, Prout, O'Rourke, Lane & Kovacs, 2011). Family situations of anger, conflict, and punishment of the children may increase the risk of a child becoming depressed (Flykt, Kanninen, Sinkkonen & Punnamäki, 2010;Rishel, 2012;Turney, 2012;Watson, Potts, Hardcastle, Forehand & Compas, 2012). Maternal depression diminishes a mother's ability to meet the basic needs of her child or form a healthy bond with her child (Coyne & Thompson, 2011;Lourens, 2014), and may lead to criticism and resentment in the mother towards her children (Goodman, 2007;Lourens, 2014).
It has been indicated that bereaved youth are more prone to developing major depression following the loss of a parent, especially if that loss was due to suicide (Brent, Melhem, Donohoe & Walker, 2009). Apart from problematic family or peer relationships, the loss of a loved one was also mentioned as a depressogenic factor in the current study: Participant 5: "She just slit her wrists… Oh, her dad died".
Towards the beginning of the discussion, a focus on depression already seemed to trigger thoughts of loss in the girls: While the girls are acutely aware of how relationships can cause emotional distress, the above paradoxically suggests that relationships, specifically mother-daughter relationships, are profoundly important to the girls.

ii) Relationships at school
Similarly, when discussing factors at school causing depression, relationships were prominent. Participants viewed ruptures in relationships, and conflict with their friends, as an important reason for becoming depressed. Prominent words that came up included "jealousy" and "gossip". Participants also discussed more specific reasons for becoming depressed: Participant 7: "A fight….with your friend…with your best friend". (2014) argue that peer relationships during childhood and adolescence consist of multiple constructs, such as "a broad range of peer behaviours…relationships…statuses and reputations…and developmental processes" (2014:185). They further state that later psychopathology or "emotional difficulties" can often be traced to previous troubling experiences with peers (in childhood or adolescence) (Choukas-Bradley & Prinstein, 2014:185). Research has also shown the strong influence adolescent girls' peer relationships have on their behaviour and emotional well-being. Recent literature has shown a correlation between increased adolescent girls' problem behaviours and deviant or problem behaviours in their close friends ( Negative self-evaluation was also mentioned as a depressogenic factor, but again, there was a relational element, in that the negative self-evaluation involves comparison with the other girls:

Choukas-Bradley and Prinstein
Participant 1: "Everyone looks pretty here but I am not pretty….and thin, thin".
Empirical confirmation has been found for the presence of negative cognitions about the self in depressed individuals (Beck & Alford, 2009;Brown, 2014) and value judgments attached to the self-concept are especially pertinent to the development of depression (Beck & Alford, 2009). Furthermore, studies have indicated the negative effects of girls comparing themselves to other girls (either socially or through the media). This negative effect was especially significant in terms of the girls' body image or body dissatisfaction (Carey, Donaghue & Broderick, 2013;Rodgers, McLean & Paxton, 2015). Depressive symptoms in adolescent girls were also correlated with technologybased "social comparison and interpersonal feedback-seeking" (Nesi & Prinstein, 2015:1). It is of interest to note the findings from a study by Rayner et al.: "…girls tended to select friends who were similar to themselves in terms of body dissatisfaction and bulimic behaviours…" (2014:93). It seems then that adolescent girls' social and body comparisons could lead to psychopathology, be indicative of depressive symptoms, and is regarded as an important behavioural trait in selecting friends.

Depression as gendered construct
In the current study, depression was clearly described as a gendered construct. Women, especially mothers, were thought of as more likely to become depressed: Participant 7: "And she was clever and then she hanged herself and then she was dead".
Participant 5: "And she was the the the…".
Participant 6: "They did, just Saturday, a girl just hung herself… And she has four children. She has four children".
As discussed above, the literature also suggests that depression is more prevalent in women than in men (Fan & Ji, 2004;Noble, 2005). There is a large prevalence of depression in women of child-bearing and child-rearing ages (between the ages of 16 and 53) (Burke, 2003;Rishel, 2012). Women in low-income communities are at an even greater risk of developing depression (Burke, 2003), making it possible that some mothers in the participants' community are depressed. This depression may influence the mothers' interpersonal relationships, especially with their children and partners (Rishel, 2012;Turney, 2012).
The fact that the participants, teenage girls living in a low-income semi-rural community, identified mothers first as those to become depressed, and told stories of depressed mothers, indicates that the presence of maternal depression in this community must be evident and visible. These participants might not necessarily have mothers who are depressed, but are considered to have witnessed depressed mothers in the community.

The healing power of relationship
Participants suggested that when depressed it helps to talk, particularly to other women. Mothers and grandmothers were specified as important figures to talk to about depression: While depression was constructed as something that was shameful and thus to be kept a secret, participants suggested that it is good to share this secret with someone who can maintain confidentiality and who is strong enough to help, like a mother or a grandmother: Participants thus expressed the view that they can talk to their own and their friends' mothers about depression, and that female figures such as friends, sisters, and mothers could be trusted with confidential information regarding depression. Goodwin (2008) states that using social relationships in order to seek help with personal, intimate, and emotional problems is an active form of help-seeking. She then states that young girls are more likely to engage in this form of help-seeking than young boys (Goodwin, 2008). Studies also indicate that talking to a trusted individual may help someone who is depressed. Dukas found that depressed women viewed the ability to "share one's feelings in the context of a supportive relationship" (2014:181) to be valuable in alleviating the symptoms of depression. More specifically, research has found that, even while the mother-daughter relationship often is a fraught one, mothers are often regarded as the ideal go-to person by a depressed adolescent and are important in the maintenance of the adolescent's depressive symptomatology (Herres & Kobak, 2015;Waller, Silk, Stone & Dahl, 2014). Whether they in fact are that helpful when approached is of course not clear.
It seems, then, that seeking help in the relational domain was viewed as important by the participants. Furthermore, women or, more specifically, mothers were viewed as those who can be trusted with the confidential, intimate information regarding depression. This suggests that relational treatment strategies may prove successful in treating depression, especially in adolescent girls. Also, quite notably, participants did not mention biomedical interventions at all, confirming that depression was not constructed as a biomedical illness.

CONCLUSION
The current study aimed at adding to the literature concerned with South African girls' subjective experience of depression. Social constructionist grounded theory was used to analyse group interview data. Five main categories could be discerned: (a) depression as a disturbing topic; (b) depression as multi-faceted phenomenon manifesting in feelings, thoughts, and behaviours; (c) depression as relational construct; (d) depression as gendered; and (e) the healing power of relationship.
Depression was an uncomfortable topic, but participants certainly were familiar with the notion, suggesting that they may have been exposed to depressed people (such as depressed mothers), or to being depressed themselves. While they constructed depression as a complex phenomenon manifesting in a variety of emotions, thoughts, and behaviours, anger and aggressive acting out behaviours were particularly prominent in the discussions. This on the one hand may indicate that in this context, anger and aggressive acting out should also be considered as possible symptoms of depression Lourens & Kruger, 2013;Lourens, 2014). On the other hand, it may suggest that the construct of depression does not adequately capture the emotional distress of girls and women in these contexts. Participants' anger and rage were much more palpable than any of the other symptoms typically associated with depression. We have argued elsewhere that "the diagnosis of depression obscures the nuances of women's psychological distress. More specifically, the diagnosis may serve to obscure women's anger…" (Kruger et al., 2014:13). Working with girls rather than adult women, we again contend that anger which is not spoken about will either be repressed, which may result in impulsive violent behaviour, or be expressed habitually in the only language available, that of violence (Blum, 2007).
It was clear that participants did not think of depression as a biomedical disorder, but rather as something caused by relationship problems or relationship loss at home or at school. Conversely they also seemed to suggest that relationships can cure depression. It can thus be concluded that the participants situated depression within the relational domain. A recent study by Lourens and Kruger (2013) conducted in the same community with adult women also found depressed participants to situate their causes and experiences of depression within the relational domain, suggesting the appropriateness of a relational model of depression (Kruse, 2012;Kruse, Williams & Seng, 2014;Lund, Chan & Liang, 2014). Situating depression within the relational domain may hold implications for therapy: situating the cause and existence of depression within the relational domain means that treatment should address this domain. Therapy, a form of relational treatment, may thus be a successful treatment option for depression. In terms of interventions for young adolescent girls from lowincome communities, the findings could suggest that more energy and resources be put into providing therapeutic intervention measures to these girls.