Abstract
Although discourse analysis is a well-established qualitative research methodology, little attention has been paid to how discourse analysis may be enhanced through careful supplementation with the quantification allowed in content analysis. In this article, we report on a research study that involved the use of both Foucauldian discourse analysis (FDA) and directed content analysis based on social constructionist theory and our qualitative research findings. The research focused on the discourses deployed, and the ways in which women were discursively positioned, in relation to abortion in 300 newspaper articles, published in 25 national and regional South African newspapers over 28 years, from 1978 to 2005. While the FDA was able to illuminate the constitutive network of power relations constructing women as subjects of a particular kind, questions emerged that were beyond the scope of the FDA. These questions concerned understanding the relative weightings of various discourses and tracing historical changes in the deployment of these discourses. In this article, we show how the decision to combine FDA and content analysis affected our sampling methodology. Using specific examples, we illustrate the contribution of the FDA to the study. Then, we indicate how subject positioning formed the link between the FDA and the content analysis. Drawing on the same examples, we demonstrate how the content analysis supplemented the FDA through tracking changes over time and providing empirical evidence of the extent to which subject positionings were deployed.
Keywords
Discourse analysis is a well-established qualitative research methodology that is used in a range of disciplines. Although there are a diversity of approaches within discourse analysis (including linguistic, ethnomethodological, semiotic, Althusserian, Gramscian, social constructionist, psychoanalytic, and poststructuralist variations), the commonalities underpinning these various methods center on the significance of language in structuring and constraining meaning and their employment of interpretive, reflexive styles of analysis (Jørgensen & Phillips, 2002).
What has received little attention in the methodological discussions concerning, and applications of, discourse analysis is how it may be enhanced through careful supplementation with the quantification allowed by content analysis. Jørgensen and Phillips (2002), in their review of discourse analysis, indicate that “it is possible to create one’s own package by combining elements from different discourse analytical perspectives and, if appropriate, nondiscourse analytical perspectives” (p. 4). How this kind of multiperspectival approach may be achieved, in particular in mixing quantitative analysis with the qualitative elements of discourse analysis, has not received, to our knowledge, any systematic attention in recent qualitative or discourse analysis methodological books. For example, The Sage Handbook of Innovation in Social Research (Williams & Vogt, 2011), The Sage Handbook of Qualitative Research in Psychology (Willig & Stainton-Rogers, 2008), and Discourse Analysis: A resource book for students (Jones, 2012) do not address such application.
In this article, we report on a research study that involved the use of both Foucauldian discourse analysis (FDA) and a directed content analysis based on social constructionist theory and our qualitative research findings. We found that while the discourse analysis was able to answer the questions about the constitutive network of power relations constructing women as subjects of a particular kind, there were questions that emerged which were beyond the scope of a discourse analysis. These questions arose from our wanting, first, to understand the relative weightings of the deployment of particular subject positions in the data set and, second, to trace historical changes in the deployment of these subject positionings. The underlying logic for the mixing of the methods in our research study was, thus, to complement the qualitative analysis with quantitative analysis to yield a comprehensive understanding within and across the data set (as suggested by Creswell, Fetters, & Ivankova, 2004) and to allow for the multifaceted and historically contingent character of the phenomenon under study to be revealed (Greene, 2008).
Importantly though, a feminist, poststructuralist theoretical grounding underpinned both analyses. Following on from Mertens (2007), our intention was to utilize mixed methods research approaches originating from a transformative paradigm so as to allow a deeper understanding of the role of power differentials in the construction of women as subjects of articles written about abortion. The research focused on the ways in which abortion was constructed, and women were discursively named and positioned in relation to abortion in newspaper articles, published in 8 weekly national and 17 daily regional South African newspapers over 28 years, from 1978 to 2005.
The explication of the discursive power relations within these media representations achieved two aims. First, by examining newspaper articles published over nearly three decades, the sheer variety of discourses drawn upon to construct abortion and the subject positions made available by those ways of speaking about abortion were made apparent. Second, the fluid, multiple, nuanced, and contradictory identities as constructed through discourse became visible. Some of these data and the manner in which we combined discursive and content analyses can be viewed in Macleod and Feltham-King (2012) and Feltham-King and Macleod (2015). In this article, we draw on examples from Feltham-King (2010) to illustrate the methodological points that we make.
In order to orient readers, we provide, initially, some background in terms of the context within which our research was conducted. Then, we discuss sampling questions. The mixing of content analysis with discourse analysis meant that we had to refine the manner in which we sampled the data. This is followed with a discussion, using examples from our data, of the usefulness of the FDA in answering particular questions. Next, we show how the content analytic method that we used complemented the discourse analysis by providing insights into the variability of use of particular subject positionings over time and the relative weightings of use of these subject positions.
Context
The transition from the oppressive, racially based Apartheid system to democracy started in South Africa in 1990, with the unbanning of previously banned political parties, the release of political prisoners (most famously Nelson Mandela), and the beginning of negotiations that led to the first democratic elections in 1994. A number of sociopolitical changes have been implemented at the demise of Apartheid, with the post-Apartheid government systematically setting about reversing the Apartheid era, racialized and gendered legislation and policies that ensured that all Black people, and in particular working-class and rural Black people and Black women, were discriminated against (Ngwena, 2004).
With respect to our research study, changes in abortion legislation and in the newsprint media are the most pertinent social contexts within which our data were generated. South Africa’s Abortion and Sterilisation Act (ASA; Act No. 2 of 1975) was introduced by the Apartheid regime as an extremely restrictive political tool that served to encourage unsafe abortion among the majority of women. The legislation was differentially applied. Black women, who comprised 87% of the population, had very limited access to state-funded medical and legal services and virtually no economic resources to access private health care. In practice, the ASA resulted in high mortality and morbidity rates for Black women. White women, who had the economic resources to fight the bureaucratic system, received the vast majority of legal abortions during the Apartheid years (Cope, 1993).
The radical transformation of the abortion legislation happened as part of the broader democratization process initiated in 1990. The Choice on Termination of Pregnancy Act (CTOP; Act No. 92 of 1996), one of the most liberal globally, allows for women (including minors) to request (without parental consent) an abortion in the first 12 weeks of gestation. Thereafter, a medical practitioner must recommend abortion under specified (relatively open) conditions. The implementation of the CTOP Act has provided evidence that the introduction of liberal abortion laws drastically curtails the number of deaths owing to unsafe illegal abortion (Klugman & Varkey, 2001). Despite problems around the delivery of abortion services (Guttmacher, Kapadia, Naude, & de Pinho, 1998) and three court challenges by antiabortion groups, this liberal abortion legislation is still in place.
Turning to the media, this was strictly controlled under the Apartheid government, which used states of emergency, warnings to newspapers, and the Bureau of Information to restrict press freedom (Tomaselli & Louw, 1989). The transition to democracy saw not only the lifting of press restrictions but also diversification of commercial publications. While ownership patterns have begun to change to reflect the demographics of South Africa, the initiation of a democratic media culture has not been without conflict (Botma, 2011; Mabote, 1998). Class, race, and gender inequalities continue to play themselves out in the media as the transformation process is ongoing (Berger, 2001).
Sampling
In order to ensure that we were able to perform the kind of integrative work that we envisaged, we had to begin the conceptualization of our research study with important sampling decisions. In these decisions, we needed to keep the requirements of both the discourse analysis and the content analysis in mind in order to ensure the integrity of the data analysis at a later stage.
Our approach to sampling eschewed the all-or-nothing, probability/nonprobability sampling binary often used simplistically to distinguish the differences between qualitative and quantitative sampling decisions. We used a multistage purposeful stratified random sampling strategy (Collins, Onwuegbuzie, & Jiao, 2007). Thus, we combined purposive sampling, often employed in qualitative research, with random and stratified sampling, most often used in quantitative research. This sampling method, although relatively complicated, allowed us to fulfill the premises of sampling for qualitative research—sampling for richness of data—and those of quantitative research—sampling to ensure a reasonable level of representativeness.
Our sampling frame was the South African media archives housed at the University of the Free State (http://www.samedia.uovs.ac.za), which is the biggest noncommercial press-cutting archive in South Africa consisting of more than 3 million print media articles that have been electronically indexed since 1978. From this frame, we first used purposeful sampling by selecting 25 (17 daily regional and 8 weekly national) newspapers that fulfilled the criterion of having published 30 or more articles about abortion over the 28 years. The purpose in using this criterion was to ensure that we included newspapers that had engaged substantially with, and made a reasonable contribution to, the debate around abortion. These 25 publications were all on the list of 37 major urban commercial newspapers in South Africa as certified by the Audit Bureau of Circulations (www.abc.org.za). This sample was further refined through multistage stratified and random sampling. The 28-year period was stratified into seven 4-year epochs. Using the individual newspapers and the epochs as axes on a grid (25 × 7), we then randomly sampled 10% of articles in each of the 175 cells (which were rounded up or down within a range of 10 units). This resulted in a sample of 300 articles. This kind of multistage purposeful stratified random sampling enabled us to feel confident about the richness of our data and to perform the quantitative comparisons across time referred to in the next section.
The Usefulness of an FDA Approach in Relation to this Study
Social constructionist approaches to discourse analysis highlight the role of language in constructing reality and the manner in which discourses provide space for particular subject positions. Despite these commonalities, there is significant debate concerning underpinning theoretical resources and analytical focus (e.g., the everyday discursive practices focused on by discursive psychology vs. a more overarching abstract mapping of discourses operating in society performed by Laclau and Mouffe; Jørgensen & Phillips, 2002).
The feminist poststructural theoretical framework adopted in this research lent itself to a FDA that explicates how culturally located discourses, positioning strategies, and practices are intricately interweaved with, and serve to reproduce or to undermine, particular power relations (Arribas-Ayllon & Walkerdine, 2008; Parker, 1992, 2005; Willig, 2008). Drawing on Foucauldian understandings of discourse (Foucault, 1972), we conceptualized discourse as a system of statements and practices that are constitutive of the objects and subjects of which they speak. In conducting the discourse analysis of the newspaper text, we utilized the criteria for distinguishing discourses suggested by Parker (1992, 2005), vis-a-vis that discourses are realized in text, are about objects, contain subjects, are coherent systems of meaning, refer to other discourses, reflect on their own way of speaking, and are historically located.
One of our aims in relation to our research study was to identify the discourses deployed in relation to women who present for abortion. Several contradictory discourses were evident across the data set. In the following, by way of example, we highlight two: a discourse of autonomy and a discourse of victimhood, indicating how these discourses were put to work in the politics of abortion. We take both of these up later in discussing how the content analytic aspect of the project enriched the discursive analysis.
A discourse of autonomy, in which people are seen as either potentially capable (potentially autonomous) of or actually enacting self-determination and independent action (autonomous), has underpinned many of the demands made by social movements in a range of spaces (Böhm, Dinerstein, & Spicer, 2010). This discourse was taken up in the data set in combination with a discourse of reproductive rights and to emphasize actions that lead to empowerment, as illustrated in the following extracts: [1] Women want the right to take responsibility for their lives. They should be given the right to choose. (McGibbon, 1992, p. 17) [2] Women’s right to choose has to be seen in the context of the interim Constitution, the Bill of Rights, and the Government of National Unity’s commitment to a nonsexist society and the empowerment of women. (“Probe team in favour of easier abortion laws,” 1995, p. 3) [3] It was virtually impossible to make a professional decision to the severity and enduring impact [of abortion] on a young girl or adult woman. Women in the context of counselling should be enabled to make the decision for themselves. (Peacock, 1994, p. 9) [4] The Reproductive Rights Campaign has been launched by a group of concerned women to fight for women to gain control over their own bodies and over reproduction. (“Reproductive rights campaign,” 1995, p. 30)
In contrast to a discourse of autonomy, a discourse of “victimhood” positions people as disadvantaged, traumatized, and/or maltreated through a set of circumstances beyond their control (Hoijer, 2004). As indicated by Jeffrey and Candea (2006), a discourse of victimhood appeals to something “nonagentive” such as poverty and “poses itself as the neutral or indisputable starting point from which discussion, debates, and action … can and must proceed” (p. 289).
A discourse of victimhood was deployed in our data set to highlight how women were victimized through abortion legislation and unfair circumstances surrounding abortion such as poverty, stigma, and lack of access to health care facilities. This way of talking about women often utilized words that were emotive and evocative, intended to evoke sympathy and compassion for the woman in that situation. [5] Mothers of unwanted children–and the children themselves–suffer and women deserve the right to terminate a pregnancy safely. (“Call for abortion reform,” 1978, p. 9) [6] When a pregnant woman believes her only option is to have an abortion she will stop at nothing to do so, even if it kills her. They will risk their lives and die, often leaving young children behind. I feel compassionate towards these women but also angry that they are forced to mutilate themselves when if abortion was legal they could undergo a simple procedure with no fear. (Krost, 1995, p. 9) [7] Young women were marginalized by age, gender and race. Their lack of status in the family, in schools and training institutions and in society at large ensures that they bear the brunt of social ills that face our communities. (Seale, 1995, p. 5) [8] Women in rural areas will be worst hit by the lack of resources, with many having to travel 250 km or more to the closest facility offering terminations. (Rickard, 1997, p. 5) [9] She [Dr Naude] said a large body of evidence showed that abortion was detrimental to the health of mothers. (“When nurses have to do their jobs”, 2002, p. 11) [10] Women, who have abortions, quickly learn that it is not as safe and easy as proabortionists would have us believe. Instead, abortion is dangerous to both the physical and mental health of women even if done under clinical conditions. Finally, instead of being a huge step forward for women’s rights, abortion on demand is the most destructive manifestation of discrimination against women. Proabortionists are trying to sell abortion to women under false premises and are taking advantage of the ignorance and vulnerability of women. (“Women will suffer,” 1996, p. 26) [11] Women who have had abortions up to 30 years ago and have married and had children in the interim, start thinking that they are mad until they realise that there is a name for what they have been struggling with for years. Kotze said that some of the women have no respect for their bodies after an abortion and will sleep around, get pregnant and simply undergo more abortions. (Fourie, 2004, p. 11) [12] We should remain aware of social realities and have compassion for the many women who are victims of tragic circumstances. (Auerbach, 1996, p. 11)
What is clear from this range of extracts is that a discourse of victimhood can be deployed in very different ways. In Extracts 5 and 6, women are seen as victims of unfair legislation that denies them the autonomy referred to earlier. In Extracts 7 and 8, we see how gradations of victimhood are constructed: Young women and women who live in rural areas are singled out for victim status (the latter in the context of liberal abortion legislation, but where services are less than optimal). While the discourse of victimhood was deployed in much of the public rhetoric that supported the liberalization of the legislation prior to the CTOP Act, and thereafter to highlight the poor roll-out of services (as in Extracts 5, 6, and 8), we see in Extracts 9, 10, and 11 how a discourse of victimhood can equally well be deployed to underpin antiabortion arguments: Women who undergo abortions will be the victims of poor mental and physical health; they are victims of proabortionists who provide them with services but misuse their ignorant and vulnerable state.
Parker’s (1992) additional criteria in identifying discourses (that discourses support institutions, reproduce power relations, and have ideological effects) speak to the deconstructive aspect of the FDA that we used. This aspect of the analysis involved, inter alia, analyzing the assumptions on which the deployment of a particular discourse draws, what gains are made in such deployment, and the implications thereof (Macleod, 2002). Thus, for example, a discourse of autonomy draws off the fundamental liberal–political notion of individual rights and an understanding that women are competent and able to make independent decisions concerning their reproductive health. It is this discourse, complemented by the dual notions of “choice” and “rights,” that has been the cornerstone of mainstream Western feminist advocacy around access to abortion (Ferree, 2003). It and a plea to reproductive public health issues formed the main arguments for the liberalization of abortion legislation during the democratization process in South Africa (Klugman & Varkey 2001).
Although a discourse of autonomy is firmly entrenched in, at least liberal, feminist advocacy for abortion, it is not without its difficulties. The assumption of active agency on the part of women seeking abortions belies the power relations within which choices are made. This might lead to a lack of examination of “the social context and conditions needed in order for someone to have and exercise rights” (Fried, 2006, p. 240) and a failure to address power relations within which responsibility for pregnancy and children are assigned.
The discourse of victimhood, on the other hand, foregrounds social context, highlighting how women suffer as a result of circumstances. It is complemented by and draws off a discourse of protectionism, in which there is a social contract to identify, and to help, those most in need of help. Thus, a discourse of victimhood is, as indicated earlier, inevitably paired with a call for some appropriate response. As seen in the extracts presented earlier, victimhood requires a basic level of compassion (Extracts 6 and 12). But it is also suggestive of fundamental action, like liberalizing abortion law (Extracts 5 and 6), improving services (Extract 8), and restricting abortion (Extracts 10 and 11). These calls to action are buttressed by the shock and horror that we ought to experience at women’s victim status.
As the inverse of discourse of autonomy, the discourse of victimhood likewise has a mixed reception among feminists. On the one hand, the manner in which victimhood deprives women of agency and renders them into the grateful recipients of benevolent (often patriarchal) assistance has been problematized (McKenzie-Mohr & Lafrance, 2011). On the other hand, a denial of victimhood is seen as a failure to acknowledge gender inequities and the experience of disadvantage and discrimination (Baker, 2010). Feminists have grappled for some time now with the bifurcation that autonomy and victimhood present, attempting to find ways to nuance and trouble these homogenizing ways of viewing women (Schneider, 1993).
Supplementing Discourse Analysis With Content Analysis: Positioning as the Link
The discourses that we analyzed in the data set were characterized by variability, contradiction, and tension. The flexibility with which they could be used in different contexts and at different times is the very feature that interested us in our analysis. This gave rise to the questions concerning the extent to which various discourses were deployed across the data set and changes in usage over time. It was to answer these questions that we supplemented discourse analysis with content analysis. We did this by homing in on a specific criterion in Parker’s (1992, 2005) list of criteria for identifying discourses: that discourses contain subjects. As Parker (1992) points out, a discourse allows space for a certain type of self—“it addresses us in a particular way” (p. 9). This has been referred to as subject positioning that “constitutes ways-of-being through placing a subject within a network of meanings and social relations which facilitate as well as constrain what can be thought, said and done by someone so positioned” (Willig, 2000, p. 557).
Davies and Harré (1990) use the terms interactive and reflexive positioning to indicate processes, whereby subjects are positioned by others and by themselves, respectively. Each description of women presenting (or potentially presenting) for abortion constructs a particular subject position for these women, simultaneously allowing and constraining particular ways-of-being within a particular system of power/knowledge (Foucault, 1980). As such, women were interactively positioned in the newspaper articles within various discourses relating to abortion.
We used this aspect of discourse analysis, vis-à-vis subject positioning, as the bridge to conducting the content analysis. Subject positioning has been proposed as providing a bridge between discourse analysis and conversation analysis (Wetherell, 1998), but we argue that it serves as well in integrating FDA and social constructionist content analysis.
Usefulness of Content Analysis
The basic assumption of the content analysis that we used in our research is very different compared to those underpinning a traditional content analysis. Rather than using quantification in an attempt to show up similarities among predetermined categories (conceptualized as fixed, stable, and objectively verifiable), the quantification in this directed content analysis was utilized to track the multiplicity, variety, instability, and historical contingency of the discursive constructions over almost three decades. Following Hsieh and Shannon (2005) who indicate that directed content analysis proceeds with relevant research findings as guidance for initial codes, we used our FDA as the basis for the conceptualization of the codes.
By utilizing a content analysis to quantify the ways in which discursive constructions positioned women, the focus shifted from understanding these positions qualitatively to revealing the shifts, changes, and the pervasiveness of particular positions. This mapping could only commence once the qualitative coding was complete. Thus, for example, having identified a discourse of autonomy and a discourse of victimhood, we started the process of coding each article for the presence or absence of the victim subject position and the autonomous or potentially autonomous subject positions.
A crucial requirement for content analysis is that the categories are sufficiently precise, and mutually exclusive, to enable different coders to arrive at the same results when the same body of material is examined. Thus, we created clear descriptions of what these subject positions entailed and included brief examples from the extracts: In the victim position, girls and women are positioned as in need of protection from circumstances, predators and abusers. Those talking about women in this way are thus implying that women do not have power or resources to protect themselves. They are described as vulnerable and at risk or as individuals who have suffered unjustly owing to circumstances beyond their control. Often a sensationalist style of writing is used with a liberal use of emotive words to evoke sympathy in readers. Examples of such words are: “desperate”, “degrading”, “humiliating” or “painful”. Positioning women in this way suggests a need for sympathy, the need to empower women or the need for powerful protectors. This can be seen in the extract [5]: [5] Mothers of unwanted children–and the children themselves–suffer and women deserve the right to terminate a pregnancy safely. (“Call for abortion reform,” 1978, p. 9) The autonomous position constructs women as full citizens of South Africa. She is not under the authority of any person or institution and is capable of mature decision-making about her reproductive choices. This positioning refers to talk which is based on an ideal and not actual lived experience. As such, the autonomous woman experiences no class, educational or gender disparities and is protected by law against abuse and considered economically productive and valuable. Such a woman has a right to safe, reproductive healthcare. Such a positioning brings the woman as a new rights-bearing South African citizen into being, as shown in extract [2]. [2] Women’s right to choose has to be seen in the context of the interim Constitution, the Bill of Rights and the Government of National Unity’s commitment to a non-sexist society and the empowerment of women. (“Probe team in favour of easier abortion laws”, 1995, p. 3) The potentially autonomous position is similar to the autonomous position. The difference is that those utilising the potentially autonomous position do so in an attempt to resist current circumstances that render women potentially powerless. They question the assumption that women should accept their powerlessness. The intention of positioning women in this way is to propose an imagined ideal in which women may experience autonomy. For this reason this way of talking about women utilises the future tense, as shown in the duplication of extract [1]. [1] Women want the right to take responsibility for their lives. They should be given the right to choose. (McGibbon, 1992, p. 17)
As indicated earlier, one of the limitations of discourse analysis is its inability to ascertain the extent to which a particular discourse is being deployed, beyond a general sense obtained by the researcher in reading the data. Following the discursive analysis, we coded each newspaper article for the presence or absence of subject positions suggested by the discourses identified. Thus, in line with the examples presented earlier, we found that women were positioned as potentially autonomous or autonomous in 12.6% and 17.3%, respectively, of all articles in the data set. In contrast, the victim position was deployed in 41.6% of the articles.
Additionally, we were interested in changes over time. Some of the analyses traced changes across the seven 4-year epochs referred to earlier (see e.g., in Feltham-King, 2010), whereas others took a broader sweep, analyzing the deployment of subject positions in two significant historical periods, vis-à-vis the Apartheid period prior to 1990 and the transition to, and actualization of, democracy post-1990. Table 1 demonstrates the prior-/post-1990 analysis for the potentially autonomous, autonomous, and victim subject positions.
Percentage of Newspaper Articles in Which the Potentially Autonomous, Autonomous, and Victim Subject Positions Were Present Before and After 1990.
Note. 33% of these cells had an expected count of less than five; therefore, the chi-square statistic to establish exact significance was not meaningful.
This table illustrates how consistency or change in the deployment of a subject position may be tracked. We note the relative consistency in the use of the potentially autonomous subject position and the stark change in the deployment of the autonomous subject position over the two historically significant periods. The victim position is deployed in more than one third of the articles prior to 1990, with an increase being noted post-1990.
These illustrations demonstrate how our supplementation of FDA with social constructionist content analysis allowed us to make analytical points that otherwise might have been missed. We were able to show how the victim subject position outweighs the potentially autonomous and autonomous subject position. Given the different implications of the discourses of autonomy and victimhood as highlighted earlier, this relative weighting of the positioning of women in the data set with respect to these opposing discourses allows us to understand the politics, and pitfalls, of public representations of abortion in South Africa. The absence of the autonomous position during the conservative, pronatalist, and nationalist politics of the Apartheid regime speaks to the historical locatedness of discursive resources. While the ASA (1975) was enforced, autonomy for women could only be expressed as a possibility (as seen in 13.8% of articles published then using the potentially autonomous position). Even after the democratization process was underway and the issues of racial and gender transformation and equity were foregrounded, the fact that the victim position is used almost twice as often as is the autonomous position has implications for the politics of abortion. Although, as indicated earlier, neither of these positionings is unproblematic from a poststructuralist feminist perspective, the emphasis on victimhood in both conservative arguments to restrict abortion legislation and in progressive arguments for better roll-out of services is in need of careful inspection, debate, and nuancing.
Conclusion
As indicated at the beginning of this article, a feminist poststructuralist approach underpinned all our analytical work in this research study. The combination of FDA and content analysis allowed us to address different issues in the overall analysis of the data and, thus, deepened our analysis of the material, enabling both a critical engagement not only with the discourses being deployed but also with their variability and contingency over time.
The theoretically driven but pragmatic approach that we took in our research speaks to the compatibility thesis, which argues against the notion that paradigms of qualitative and quantitative research are inherently incompatible. Compatibility, argue Karasz and Singelis (2009), should be judged by whether the utilization of a mixed method is appropriate for the research question being posed. Furthermore, the underlying theoretical framework should be consistent.
Had we conducted only discourse analysis, our research questions would have been limited to discourses and subject positions appearing in the data over the 28-year period. Supplementing this analysis with a content analysis of the extent to which the subject positions were deployed in the data set and how these changed over time added depth to the analysis and allowed us to speak to the variability of discursive constructions over a period of time and within specific sociohistorical contexts.
Our method must be distinguished from Foucault’s genealogical method (Foucault, 1977). As a genealogy is a history of the present, tracing back particular discourses and practices to the conditions of possibility of their emergence, it envisages a less linear historical approach than our mixed method. We did not identify particular issues in the present on which we wished to perform a genealogical study. Instead, we left the field open, tracing in a linear fashion the range of discourses emerging across the data set and the changes in discourses over time.
Footnotes
Acknowledgments
This work is based on research supported by the National Research Foundation (NRF) South African Research Chairs initiative of the Department of Science and Technology and NRF. We thank Jabulile Mavuso for her assistance with coding the data.
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This research was funded by the National Research Foundation of South Africa (2054126).