Class in Soweto

Class in Soweto
Peter Alexander, Claire Ceruti, Keke Motseke,

Mosa Phadi and Kim Wale
iii
Contents
Photographs fall between pages 84 and 85

Abbreviations vii
Acknowledgements ix
1 Affordability and action: Introduction and overview 1

Peter Alexander
2 Historical introduction to class in Soweto 35

Kim Wale
3 Contemporary Soweto: Dimensions of stratification 55

Claire Ceruti
4 A proletarian township: Work, home and class 96

Claire Ceruti
5 Underemployment: Too poor to be unemployed 127

Peter Alexander and Kim Wale
6 Models, labels and affordability 142

Mosa Phadi and Claire Ceruti
7 Perceptions of class mobility 164

Kim Wale
8 The language of class: Confusion, complexity and difficult words 190

Mosa Phadi and Owen Manda
9 Class and religion: Denominations in Soweto 210

Keke Motseke and Sibongile Mazibuko
v
10 Conclusion 234
Peter Alexander
Appendix 1: Methodolgy 249

Peter Alexander and Claire Ceruti
Appendix 2: Sowetans in the Classifying Soweto survey 259

Claire Ceruti
Bibliography 274
Notes on Contributors 295
Index 297
vi
Affordability and action 1
Affordability and action

Introduction and overview
Peter Alexander
‘The discrepancies in livelihoods across the world are so large that they are without
historical precedent and without conceivable justification’ (Pieterse 2002: 1–2). This
gulf has fed commentary about ‘class’ both in the popular media and among political
analysts. In many countries, the expansion of a new middle class has attracted
attention, and this has sometimes been matched by concern about the plight of the
poor and the working class. Academic research on class has lagged behind, but even
here there has been renewed interest over recent years. However, this scholarship
emanates overwhelmingly from the global North, where class structures and social
dynamics are different from those in the majority of the world. This book enriches
the debate through a study conducted in Soweto, South Africa’s largest black
township, which provides a home for about one in 30 of the country’s 50.6 million
inhabitants (Stats SA 2011a).1
In South Africa, income inequality and unemployment have reached extreme
levels. In terms of Gini co-efficients for income, it is the most unequal country in
the world (with the exception of a few with small populations), and Johannesburg
– which is where Soweto is located – is the most unequal major city (with a Gini
co-efficient of 0.75) (UN HABITAT 2010: 73, 193). The county’s unemployment
levels are also among the worst in the world, with the official figure standing at
24.9 in early 2012 (Stats SA 2012b). However, a very high proportion of adults is
‘not in the labour force’, so excluded from this statistic, and others are engaged in
survivalist activities. All this may lead one to dismiss South Africa as so exceptional
that its class structures and identities can be written off as unique, and thus extraneous
to a wider analysis of class. Our view, however, is that the intensity of the contrast
1
2 Peter Alexander
between rich and poor in South Africa and the magnitude of joblessness in Soweto
and elsewhere, can assist us to comprehend the class nature of those who are excluded
by occupation-related analyses or marginalised by a focus on direct relationships
to production. In other words, it can contribute to the study of class, not only in
lower-income countries, but also in much of the global North, where unemployment
and underemployment are now significant issues, both socially and theoretically.
In Spain, the second quarter 2012 official rate of unemployment was 24.6 per cent,
only 0.3 per cent lower than South Africa’s for the same period (Washington Post
2012). The urgency of such analysis was underlined by, for instance, the 2011 ‘riots’
in Britain, in which the majority of the participants were outside the labour force
and viewed their actions in terms that were diametrically opposed to the majority
of the population (Roberts 2011: locations 545, 638).
Our research was motivated by two concerns. In addition to interest in the
impact of heightened inequality, there was a related political question. While this
had broader implications, we posed it locally. Earlier research had revealed that the
more radical of South Africa’s social movements were composed overwhelmingly
of people who were jobless (Alexander 2006a: 49). In contrast, trade union member-
ship invariably excludes those without jobs. Since 2004 this gap has been highlighted
by the development of, on the one hand, ‘a rebellion of the poor’, in which workers
are peripheral, and, on the other hand, a level of strike action that, measured in days
lost per capita, was among the highest in the world. Furthermore, there have been
political differences between leaders of the Congress of South African Trade Unions
(COSATU) and the social movements, and there has sometimes been friction between
employed workers and rebelling youth. In this context, we pondered the following
question: Is it possible that workers and the poor constitute distinct classes?
We have called this book Class in Soweto, a title intended to convey a sense of
how people see themselves in class terms, as well as the differences in employment,
housing, income and culture we discerned as researchers. The project from which
the book emanates was known as ‘Classifying Soweto’, with this name embodying
an understanding of ‘classifying’ in line with academic usage that signals the
importance of social class for analysis of social change. Recognising that apartheid
discourse still has purchase and that ‘classifying’ might be acssociated with racial
classification, we decided on a new title. The chapters in this volume were written
by different authors, all of whom were engaged in the project and participated in
discussions on how to interpret its findings. This opening chapter outlines the
project’s theoretical and methodological framework, comments on some relevant
literature, summarises the various chapters, and then draws on all of these to highlight
novel aspects of our account.
The book is innovative in a number of respects. With 69 per cent of adult

Sowetans either not in the labour force, unemployed or engaged in survivalist
activities, traditional occupation-based categorisation had limited value. Instead we
developed nine employment categories that reflected different relationships to
production (mostly indirect). Each individual could be allocated to an employment
category (EC), which was valuable, because it is individuals who engage in different
kinds of economic activity, participate in struggle and hold particular identities.
This takes us to the second innovation. Unlike previous studies of class identity,
and in line with the recognition that people have multiple identities, we permitted
respondents to describe themselves using more than one class label. That is, they
could say they were middle class or working class, but they could also describe
themselves as middle class and working class. It turned that out that more than 90
per cent of Sowetans could identify themselves using class terms, but only 38 per
cent used a single label.2 Most Sowetans had multiple class identities. We found that
there was a material basis for the different class labels. A close association existed
between regular employment and ‘working-class’ identity, and people described as
‘lower class’ or ‘poor’ (the two terms were virtually synonymous) tended to suffer
greater deprivation than other Sowetans. We also discovered that distinctions
between lower-, middle- and upper-class labels were based mainly on a notion of
affordability, a term derived from popular perceptions that linked identity with
EC. When it came to action, strikes mobilised people who generally regarded
themselves as working class and the rebellion involved mainly people who were
not workers, a high proportion of whom were regarded as lower class. However,
the distinction was about ‘relationships to means and ends of protest’, rather than
class in any fundamental sense, and we reached the conclusion that, in class terms,
Soweto was best understood as a ‘differentiated proletarian unity’.
Theoretical and methodological framework

Erik Olin Wright (2005: 180) once asked: ‘If “class” is the answer, what is the
question?’ His point was not only that definitions of class are embedded in distinct
theoretical approaches, but also that these approaches are anchored in different
kinds of question. Each approach has a primary anchor that establishes the main
criteria that ‘class’ as a concept is designed to fulfil, one or more secondary anchors
that help delineate the theoretical reach of the concept, and, in most cases, further
questions that are empirically relevant but do not impinge on the definition of
class. Wright (2005: 180–92) then sketches out six key questions that, depending
upon one’s approach, could be regarded as a primary anchor, a secondary anchor
4 Peter Alexander
or merely empirically relevant. These six questions are about distribution of

inequality, subjective location, life chances, antagonistic conflict, historical variation
and emancipation. He proposes that the last of these provides Marxism (and himself )
with its primary anchor, arguing that, within this tradition, ‘class contributes to
the critique of capitalist society rather than just to description and explanation’.
This book derives from a project animated by just this position – it stimulated the
founding questions and it grounded my conceptualisation. But there have also been
secondary anchors. Like Wright, we have been concerned with antagonistic conflict.
Unlike him, we also prioritised subjective location (and I will return to this later).3
Extending Wright’s metaphor just a little, an anchor prevents a boat from drifting
out to sea, but it allows that boat to rise with the tide and sway with the wind. Our
researchers did not have to swear an oath of allegiance to Karl Marx when they
were appointed, and discussions between colleagues over theory and over
interpretation of data have been part of the pleasure of this project (and, one hopes,
these have improved our analysis). We gained a great deal from the work of scholars
strongly influenced by Max Weber and we read Pierre Bourdieu’s Distinction as a
team. This affected the design of our questionnaire, and it has flavoured our
conclusions. However, the more I have reflected on our findings, the less explanatory
value I have found in the work of these two great writers. Of course this is partly,
or even largely, because the opening questions had Marxist anchors, but there is
more to it than that. Critically, the assumption of relatively stable boundaries
implicit in the key concepts of status group (Weber) and habitus (Bourdieu) make
better sense in the contexts in which these terms were developed (respectively,
early twentieth-century Germany and 1960s France), than in Soweto at the end of
the first decade of the twenty-first century.4 Had we looked at South Africa when
it was more rigidly divided along racial lines, or even, perhaps, if we had considered
the whole country today (rather than one large township), it might have been
necessary to nuance this point. But the Soweto we researched was the product of
rapid social change, and within it there was a good deal of social fluidity and blurring
of boundaries. I make use of the term ‘poor’, deploying it analytically in
contradistinction to ‘workers’, but am not suggesting that the poor constitute a
status group (even if we allow for the possibility that a status group can be negatively
privileged). A worker might be materially and culturally ‘better off’ than somebody
regarded as ‘poor’, but workers and the poor tend to share households and
neighbourhoods, a person can be a worker one day and ‘poor’ the next (and vice
versa), and disposable income is linked to household demographics. The idea of
habitus fits, at best, uncomfortably with people holding multiple class identifications,
which is what we found (see also Phadi and Ceruti 2011: 91).
There are three theoretical issues that require clarification before proceeding.
First, one of the strengths of Marx’s ‘class’ is that it was principally an abstraction
(part of his theorisation of the laws of motion of the capitalist mode of production
and a prediction about a socialist future). In this sense, there were ‘two great classes’,
capitalists and workers, each internally united by their antagonistic relationship to
exploitation. Because class was abstract and exploitation remains central to social
dynamics in capitalist society, the concept has continuing salience in the form in
which it was initially advanced. However, Marx never formally defined ‘class’, and
he (and Frederick Engels) also used the same word in ways that were historically
specific (that is more sociologically and/or politically concrete) (see also Wolpe
1986).5
The second issue is this. At the more concrete level, class is about production,
but it is also about aspects of social reproduction.6 The importance of reproduction
was acknowledged, especially by Engels (1884: preface), and Marx (1954: 603) writes,
for example, about the way the unemployed were ‘thrown’ onto the ‘shoulders’ of
the working class and lower middle class. Today, there is often a larger gap than in
Marx’s time between classes defined in terms of (1) production and (2) distinctions
within the space of reproduction (that is, consumption as well as the reproduction
of labour). This arises from a growth in the size of the middle classes, and, in cases
like South Africa, an increase in the proportion of people who are either unemployed
or underemployed. Consumption can encourage quiescence, but absence of things
such as housing of a certain quality can be an important source of struggle. Action
may occur in the sphere of production (for instance, a strike) but may also occur in
the sphere of reproduction (such as a service delivery protest). Understood in this
way, Marxism provides us with the greater flexibility we need to make sense of
contemporary society.
The third issue concerns the importance of subjectivity (which Wright does
not stress but we do). Here my concern has three components. (1) Changing the
world involves agency, and to address this we need to understand how people see
themselves consciously and behave unconsciously in class terms. The emphasis
on agency derives from the weight Marx and Engels placed on working-class
action. This comes, in particular, in the formulation: ‘the emancipation of the
working class must be the act of the working class itself’ (Engels 1969: preface to
the English edition of 1888, though similar formulations exist elsewhere). (2) As
Alex Callinicos (2006: 2–3) argues, Marx’s distinction between class-in-itself and
class-for-itself is closely related to one between class as something objective, what
Callinicos terms simply ‘class’, and subjectivity. In this formulation, subjectivity
could include subjectivities related to race, gender and so on, but we focused
6 Peter Alexander
narrowly on class subjectivities. (3) Formulations that clearly distinguish between

objective and subjective aspects of class discourage slippage into assumptions that
capitalism leads inevitably to socialism, and instead opens possibilities for researching
the nature of the relationship between the two dimensions.
We started from the principle that there is value in undertaking an empirical
study of class. In order for Marx to theorise the centrality of the working class, he
first had to experience it as a collective subject. For this there was probably a
seminal moment in August 1844, when he wrote glowingly about Parisian workers
and Silesian weavers, and Frederick Engels joined him fresh from Manchester (where
he was working on The Condition of the Working Class in England).7 Marx and
Engels reflected on what they witnessed, and interpreted this by drawing on their
existing theoretical positions. The aim in this project was to proceed in a similar
manner, albeit in a more consciously sociological way. In a parallel project,
summarised in Chapter 10, we look more directly at contemporary struggle, doing
so through the lens of the ‘rebellion of the poor’. In neither of these projects has
our interest extended to what I defined above as the abstract level of Marx’s theory,
which was certainly beyond our reach.
What we have done is investigate the interplay between an objective marker of
class related to production (what we termed ‘employment categories’), some aspects
of reproduction (particularly housing and household composition) and subjectivity.
The last of these might be further divided into questions of identity (both identity
labels and the way in which people describe themselves within broader systems of
meanings), culture (subjectivity practised in everyday life) and conflict (how people
organise in relation to class struggles). This gives us five main themes, which, in
short-hand, we can refer to as production, reproduction, identity, culture and
struggle. While chapters cut across themes, Chapters 4 and 5 are mainly about
production-related questions, Chapter 3 deals with issues related to reproduction,
Chapters 6 and 8 are about identity, and Chapters 7 and 9 look at culture (though
in very different ways). The fifth theme, conflict, forms the background to the
book and is discussed in Chapter 10. None of this sits perfectly or fully with
distinctions outlined above, which have emerged from an attempt to integrate
empirical evidence with my own pre-existing theoretical assumptions. The outcome
is complex and somewhat messy, but that is the nature of empirical research.
We operationalised our key concepts in various ways. For our ‘production’
theme, we introduced the notion of employment category (EC). Given that some
ECs are associated with lack of employment, the term might seem like a misnomer,
and it is not ideal, but we used it for want of something better. We consciously
avoided ‘occupation category’ and ‘class location’, and we did not regard ECs as
‘classes’ (in any sense). While there was a theoretical basis to the ECs, they were
modified to some extent by an iterative engagement with the data. One can see
them as defined in terms of relationships to production, whether direct or indirect,
and of relationships between the different categories. However, the categorisation
should be regarded as provisional, because an acceptable theorisation would require
a broader frame of reference, both geographically and in terms of the questions
being posed. A further explication of the distinctions and the procedure for allocating
individuals to the various categories – a complex and time-consuming process – is
provided by Claire Ceruti in Chapter 4.
We ended up with nine ECs. First, there are the capitalists, the exploiters.
Secondly, there are regularly employed workers who stand on the other side of the
main class divide within capitalist society. We lumped all workers together, whether
or not they directly produce surplus value. The third group, managers and
supervisors, were employees who exercised some measure of control and surveillance
over workers (albeit, in practice, with limited powers). Fourthly, there was the
petty bourgeoisie (which included self-employed workers), who owned their own
business but exploited themselves and employed no more than a handful of staff
(commonly family members). The fifth group, the unemployed, resemble Marx’s
‘reserve army of labour’ (see Chapters 4 and 10). The next two groups are akin to
that part of Marx’s ‘surplus population’ engaged in some kind of work (so not
actually part of the reserve army). They include the survivalist self-employed, who,
unlike the pretty bourgeoisie, would have preferred to be regularly employed. Then
there are partial workers, whose precise relationships to exploitation varied, but
whose conditions were always inferior to those of the regularly employed. In
practice, most were engaged on a short-term basis (for example, as day labourers),
but we also included people working just a few hours per week, who might, literally,
be regularly employed, but were actually unemployed for more than half of what
one can regard as a normal working week (that is, 40 hours). Finally, there were
two groups that were outside the labour force (in other words, neither employed
nor available for employment). These were, eighthly, students and, ninthly, pensioners
and others not in the labour force (for instance, disabled). In some respects these
two categories were similar, but they had contrasting expectations of joining the
labour force (that is, making themselves available for exploitation).
For our identity theme, we learned from initial qualitative research that people
could hold more than one class identity. While this was a surprise, it seemed plausible
theoretically.8 For the survey, the discovery enabled us to ask our key identity
question in a novel way – one that moved beyond expecting respondents to select
just one identity from a predetermined list. We asked ‘would you call yourself . . .
8 Peter Alexander
[followed, in turn, by one of eight labels, all of which were derived from the
qualitative interviews]’. Our other themes do not require further elucidation at
this stage. For ‘reproduction’ and ‘culture’, colleagues drew together responses to
several, sometimes many, questions in our survey. This was linked with data from
qualitative interviews.
A case study of the kind undertaken was suitable for developing and testing
concepts and relationships, which is what we needed in order to provide a response
to our opening questions. But, why select Soweto? Firstly, it was accessible (indeed,
during most of the research, our team was situated on the university’s Soweto
campus). Secondly, linked to this, we had some prior knowledge of the township’s
geography and political life. Thirdly, its size, location, leadership and past events
meant that it was – and still is – politically important. Fourthly, related to this,
there was literature that helped with historical context. Fifthly, its size also meant
that it provided a reasonable cross-section of class backgrounds (although this proved
to be more limited than anticipated). Sixthly, we wanted to undertake a study that
was comprehensive and representative, and Soweto is sufficiently compact and
geographically defined to make this possible. Lastly, it provided the possibility to,
albeit cautiously, draw out some implications for South Africa as a whole.
Appendix 1 provides a detailed account of our research design and its
implementation. The main component was a survey conducted in Soweto in June
and July 2006 (which covered adults of sixteen years and over). This included a
representative sample of 2 340 respondents and had a response rate which exceeded
90 per cent (that is, very high). Quantitative research was sandwiched between two
main rounds of qualitative fieldwork (with some chapters, especially the one on
churches, involving further research). Additional data was collected for our research
documentary, Phakathi. We regarded our approach as broadly Marxist, albeit
heterodox. It was intended to capture a particular situation in which those officially
regarded as employed were a minority of the adult population (less than one-third
if one excludes the underemployed). However, we planned the survey in a way
that left open a range of interpretations.
What were the limitations? Firstly, we did not survey Johannesburg’s ‘suburbs’.
This has three significant implications. (1) In the course of our research, it became
clear that there had been significant out-migration of better-paid Sowetans, and
these were not captured in the survey.9 (2) There were no white or Indian
respondents in our survey and only a few so-called ‘coloureds’, so the significance
of racial differences is largely absent. (3) The research misses out on the wealthy in
general and capitalists specifically (though this would probably have been true of
any survey-based project). Secondly, we administered our questionnaire and

conducted interviews in or outside people’s homes, and our findings about identity
may have been different had we undertaken research in workplaces. However, one
of our researchers conducted preliminary fieldwork in two factories, and his
interviewees tended to stress workers’ lower-class – as distinct from working-class
– identity (Manda 2009). Thirdly, we cannot say anything about the rural areas
(including former homelands, farms and mines), the population of which constitutes
nearly half the country’s total (Gaffney Group 2012: 130). The fourth limitation is
that our survey provided a snapshot of a particular moment in time. One significant
change was that in 2009, mainly as a consequence of the global financial meltdown,
nearly one million jobs were lost in South Africa.10 In addition, while the 2011
local government election showed that the African National Congress (ANC)
remained the dominant political force in Soweto, the proportion of the township’s
residents who were ‘unhappy’ with service delivery had increased significantly and
many participated in strikes and urban unrest.11 Despite these limitations, we feel
that the class structure and dynamics around class subjectivity that we have revealed
have broader relevance, especially for South Africa. Beyond the country’s borders,
our research will expand the geographical scope of scholarship on class, providing
pointers for ongoing theoretical debate.
Some relevant international literature

This section offers reflections on theoretical work that had direct implications for
concepts we have used or arguments we have advanced. It is also aimed at establishing
a foundation for demonstrating the originality of our findings, analysis and
theorisation, which is advanced later in this chapter. A fuller reflection on the
literature is available elsewhere (Alexander 2010b). Studies on class in South Africa
have more immediate significance and will be considered in the section that follows.
Much debate on class in the 1980s and 1990s pitted Wright, champion of the
neo-Marxists, against John Goldthorpe, standard-bearer for the neo-Weberians.
Wright’s (1978: 30–110) initial attempt to apply Marxism to the class relations of
advanced capitalism led him to distinguish six categories associated with control
over production.12 With a concern to place greater weight on disaggregating the
middle class, he subsequently developed a scheme that had twelve class locations,
with these defined by relations to authority and to scarce skills, as well as means of
production (Wright 2000: 15–26). In contrast, Goldthorpe’s approach was initially
founded on assessments of common life chances that were related to market and
work situations, but in a further iteration he stressed the importance of employment
relations (distinguishing, on the one hand, employees who are tightly supervised
10 Peter Alexander
and paid a wage, and, on the other hand, cases in which work requires greater
autonomy, and pay and conditions reflect the need for loyalty) (2000: 206–29).
Arguably, Wright moved closer to Weberian formulations and Goldthorpe shifted
nearer to Marxism (Roberts 2001: 27–31). While Wright’s revised approach has,
perhaps, made less impact than the earlier one (Roberts 2001: 37; Seekings 2005:
27–8), Goldthorpe’s new scheme has had significant influence in Britain, shaping
the socio-economic classes used by its Office of National Statistics (ONS). The
ONS, however, made one important change. It added an eighth category to
Goldthorpe’s seven (ONS 2008). This included the long-term unemployed and
students, a so-called ‘residual’, thus recognising the growing importance of these
segments of the population and the need to move beyond specific employment
relationships. The approaches of both Wright and Goldthorpe have merits. The
former contributes to our understanding of class conflict and social transformation,
while the latter reveals the extent to which educational achievement, social mobility
and voting behaviour are linked to ‘class’. However, both have the limitation of
seeing class as something rooted only in work relationships, what Rosemary
Crompton (2008) termed an ‘employment aggregate approach’.
In contrast, there is ‘culturalist class analysis’, influenced to a greater or lesser
degree by different readings (perhaps misreadings) of Bourdieu (who actually placed
a good deal of emphasis on production relations) (for example, 1984: 372). The
value of this literature is that it challenges one to think about the way in which
culture marks out class boundaries, albeit in ways that are often ‘ambiguous’ and
‘fuzzy’ (Skeggs 2004: 5). While methodologies have generally been ethnographic –
limited in scope, if insightful – Tony Bennett et al. (2009) drew on some of Bourdieu’s
quantitative work to produce a statistically sophisticated study, conducted in 2003/
04, which covered the whole of Britain. They found that cultural variables – covering
participation, knowledge and taste – were associated with respondents’ ‘class’, gender
and age. While there were no homologies (that is, clearly defined groupings of a
Bordieuvian kind), the categorical midpoints were distinct in each case.13 The class
model that best fitted the data – that is, that best described how people in Britain
practice class – was one that distinguished the middle class, an intermediate class
and the working class. The research found that the working class ‘neither likes nor
is interested in legitimate [high] culture’ (Bennett et al. 2009: 252) and that the
middle class, especially its higher levels, simply engages in more cultural activity
(its members are ‘omnivores’). In reading this account, one might conclude that
the lack of homologies and the need for an intermediate class reflect the importance
of ‘ambivalence’ (the same might be said of ‘fuzziness’).
In response to culturalist approaches, Goldthorpe and Tak Wing Chan (2007;

Chan and Goldthorpe 2010: 12) argue that we should distinguish, along neo-
Weberian lines, between class positions (particularly employment relations) and
status order (expressed in a hierarchy of lifestyles associated in particular with ‘who
eats with whom and who sleeps with whom’).14 Goldthorpe (2008: 351) concludes:
. . . when we look at things like risk of unemployment or the long-term

lifetime development of earnings, class not status is important. If we shift,
however, to another field, that of cultural consumption – the extent to which
people participate in various forms of music, theater, dance, cinema, and
the visual arts – we get the reverse result.
Given that Bennett et al. (2009) use Goldthorpian economic categories (albeit
modified), it is perhaps hardly surprising that their conclusions had much in
common with those advanced by Goldthorpe. Moreover, both are consonant with
a model that recognises that, while people in Britain frequently have difficulties in
consciously positioning themselves (in other words, they are often ambivalent),
their cultural practices still recognise a distinction between working class and middle
class.
In terms of class identity labels, the existing literature makes interesting reading.
From 1983 until 1991 the British Social Attitudes Survey (BSAS) included a question
about the class people ‘belonged’ to (Jowell et al. 1993). In 1991 the figures were
29 per cent for the middle classes, 64 per cent for the working classes, and 4 per cent
for ‘poor’. Regrettably, the series was discontinued, but more recent studies still
show that a majority of Britons regard themselves as working class.15 The US
data is better. For instance, American National Election Studies (2010) has asked
about class identification since 1956. It requests respondents to select one of four
classes, with the followings result in the most recent survey: average working class –
41 per cent; upper working class – 10 per cent; average middle class – 39 per cent;
upper middle class – 10 per cent. While the numbers contradict much public
discourse about the archaic nature of working-class identification, they go some
way to explaining why politicians, especially in the US, but now also in Britain,
are concerned about the working-class vote.16
However, simple choices of the kind offered in these studies miss the problem
of ‘ambivalence’, discussion on which was initiated by Mike Savage (2000). His
interpretation of 200 interviews conducted in Manchester was that, although about
two-thirds of respondents could define themselves in class terms, ‘identification is
12 Peter Alexander
usually ambivalent, defensive and hesitant’ (see also Savage, Bagnall and Longhurst
2001). He argued: ‘class does not seem to be a deeply held personal identity, nor
does “class belonging” appear to invoke strong senses of group or collective
allegiance. In so far as class is significant, it is largely with respect to politics.’
However, Geoff Payne and Clare Grew (2005) re-evaluated the Manchester data
and combined this with their own fieldwork, which used similar questions but
was conducted in a rural area. They concluded that hesitations and qualifications
do not reflect unease about class as such, but rather the genuine difficulties that
exist in reaching conclusions about a complex phenomenon. For their respondents,
they argued, class was not a simple matter of occupations, but involved assessments
about income, attitudes, housing, education and much more.
Discussion about the nature of such ambivalence was taken up in a US study
by Michael Hout (2008), which draws on data from the General Social Survey.17 In
this he distinguishes between people who were unambiguously working or middle
class and those whose class position was ambiguous. The latter included individuals
who were ‘living on the border’, that is, lower professionals, routine white-collar
workers and self-employed manual workers, and/or those subject to ‘status
inconsistency’, such as people with a low-status job and a high income. Among
employed people who were unambiguously working class, 71 per cent identified
themselves as working class, with a further 8 per cent saying they were lower class;
and among those in unambiguously middle-class positions, 77 per cent said they
were middle class, with a further 9 per cent self-identifying as upper class. For
those in ambiguous positions, the split was 47 per cent middle class, 53 per cent
working class and less than one per cent for each of the other two options. Thus,
Hout’s study suggests that class ambiguity is an important source of ambivalent
class identity.
A further twist is provided by Paula Surridge (2007), who probed responses to
a question on self-identity asked in the 2003 BSAS. This offered respondents a list
of sixteen possible identities from which they were asked to choose three, placing
them in order of importance. While those who were most obviously disadvantaged
were strongly attracted to a working-class identity, the ‘salariat’ had a preference
for non-class identities. This should be linked with evidence from Bennett et al.
(2009: 178), who observed that, while the higher echelon of the middle class was
culturally dominant, it rejected any clear class identity, as doing so might invite
‘contestation’.18 So, one might hypothesise that in Britain there is a tendency for
different classes to be constituted differently. From a functionalist perspective, this
would make sense. For workers, an emphasis on economic boundaries adds
legitimacy to arguments on the injustice of domination by a rich minority and, by

contrast, the middle class benefits from the outward fuzziness, inner cohesion and
confidence associated with cultural capital. However, British reluctance to claim a
middle-class identity can be contrasted with the position in the US, where, according
to Fiona Devine (2005: 161), people from all economic backgrounds spoke freely
about their class identity (with ‘middle class’ seen as an inclusive category, rather
than an exclusive one, as in Britain).
This rich body of international literature has undoubted strengths. In particular,
while the importance of culture has been highlighted, the continuing salience of
employment relations in shaping critical aspects of social life is also apparent.
Further, while research on identity is underdeveloped, what does exist has tended
to show that in Britain and the US, class identification retains significance in people’s
lives, even if this is complicated by ambivalence about what ‘class’ means. The way
class is practised and the extent to which it is verbalised are different in the two
countries, and in Britain there is variation between classes. Thus, we are alerted to
the need for textured and context-specific analysis if we are to make sense of class,
and class identity in particular. However, there are gaps. Most importantly, leading
sociological journals have carried few articles on the character of class in the global
South and prominent scholars on class appear to be innocent of the theoretical
implications of the kinds of structures and identifications that exist outside the
North (see Alexander 2010b).19 In this book, we offer a contribution that addresses
this parochialism.
Some views of class in South Africa

The purpose of this review of local literature is broader than our brief reflection
on international texts above. In addition to providing a basis for showing how our
work builds upon and moves beyond existing scholarship, the aim is to expose
South African debates to a wider audience and sketch out some pertinent historical
context. There are, of course, ties to the international literature. Marx and Weber
have both been influential, and Wright, and, to some extent, Goldthorpe have also
made an impact. There have also been moments when South African scholarship
has had a resonance internationally, especially through discussion of race and class.
Within the local literature, four main problematics can be discerned and are
considered below. These are, firstly, the general structure of class relations, including
what might be termed the ‘white worker debate’; secondly, the contours of class
relations in townships; thirdly, the African petty bourgeoisie; and, lastly, class
analysis and the poor.
14 Peter Alexander
Structure of class relations and the white working class

Initially, discussion about apartheid, which was imposed from 1948 onwards, hinged
on the idea of ‘caste’, seen by Weber (2010) as an extreme form of Stand (usually
translated as status group). In an early contribution, Leo Kuper (1949: 152–3) wrote
about the ‘ “caste-like” characteristics of the social structure’, concluding: ‘Race
factors restrict the rational development of capitalism, by imposing the non-rational
use of subordinate labour.’ He added:
. . . the criterion of ‘propertylessness’ is insufficient to establish the class

situation of workers: race is an integral factor, securing and reinforcing
non-ownership of productive property as a determinant of the class position
of the Native worker, but tending to identify the white non-owner with
the owner group.
Pierre van den Bergh (1965: 52–9) reasoned along similar lines, arguing that ‘ “race”
is by far the most important criterion of status’ and that four ‘colour castes’ (whites,
coloureds, Indians and Africans) were each sub-divided by class. Asserting that
there was a lack of class consciousness among white workers, he dismisses – in one
sentence – the applicability of a Marxist theory of class to South Africa, adding
that the South African labour movement ‘has always defined its function as that of
protecting the White manual worker against non-White competition’.20 While ‘race’
posed a challenge for Marxist analysis, Van den Bergh misread the labour history,
which shows that white trade unionists, many of them Marxists, frequently acted
in sympathy with black workers (see Alexander 2000: 63, 120–3).
As Jeremy Seekings (2009: 877) observes, in the early 1970s ‘Weberian social
science was eclipsed, totally and quickly’. It collapsed under pressure from a Marxist
critique, which demonstrated that, rather than apartheid being ‘non-rational’, it
was functional to the development of capitalism. The migrant labour system had
delivered ‘cheap labour’ to South African employers and institutionalised racial
divisions among workers, thereby boosting profits (Wolpe 1972; Johnstone 1976).
The problem for the Weberians was largely political: following Weber, race and
class inhabited separate domains, and in the South African case the former was said
to be undermining economic development associated with the latter. As a
consequence, the Weberians appeared to be shielding capital at a moment when
class struggle was growing in importance and when solidarity movements were
challenging international businesses that had South African interests.21
But how would Marxists handle historical divisions between black and white
workers? One approach treated white miners as foremen and then assumed them
to be stereotypical of white workers (Simson 1973); another argued that a petty

bourgeois component of the ‘white wage-earning classes’ received some of the surplus
extracted from black labour, with these holding sway over the ‘white working
class proper’ through ‘ideological class practices’ (Davies 1973, 1979: 25); and a
third distinguished free white labour and unfree black labour (Johnstone 1976).22
In contrast, Harold Wolpe (1976) offered a more orthodox Marxist position that
emphasised class differences among white ‘employees’ and criticised writers who
exaggerated the political and ideological determinants of class. By 1980 the debate
had become scholastic, and a vibrant new labour history was highlighting
considerable complexity in the matters at hand (for example, Webster 1978; Marks
and Rathbone 1981).23 Moreover, while white workers were declining – both
numerically and in political influence – black workers were growing in number
and developing new unions that played an increasingly important role in the struggle
against apartheid.
By the mid-1980s, Wolpe (1986) had developed a critique of ‘economistic’
theories of class. He argued that, in constructing ‘class’ in purely economic terms,
these approaches – whether Weberian or Marxist in origin – assumed either that
class relationships were irrelevant to an explanation of apartheid or that an
explanation of apartheid could be reduced to class interests (including the interests
of different class fractions). Actually, Wolpe’s own earlier account of apartheid –
emphasising the benefits of apartheid for capitalism – was itself economistic (and,
as a consequence, obscured the potential for the system to unravel). His new position,
with which I concur, took a different tack:
. . . while at one level classes must be conceived of as unitary entities,

concretely, to the contrary, their internal unity is always problematic. In
the sphere of production and exchange, classes exist in forms which are
fragmented and fractured by politics, culture, ideology and, indeed, the
concrete organisation of production and distribution itself . . . One might
say that class unity, when it occurs, is a conjunctural phenomenon (Wolpe
1986: 121).
An abstract analysis of class is required, because it can reveal processes that are
otherwise hidden, but it is not sufficient. The significance of Wolpe’s approach is
not only historical (associated with the ‘white working class’ and so on), but also
relates directly to our own study. That is, one should not expect to find a direct
and simple relationship between class – as something objective – and its subjective
manifestations (whether self-identity, culture or struggle).24
16 Peter Alexander
Class in the townships

Alongside this theoretical debate, a series of rich ethnographic monographs was
providing a picture of stratification in South African townships in the 1950s, 1960s
and 1970s. As Seekings (2009: 871) observes: ‘[T]hese studies . . . were important in
breaking with anthropological scholarship that prioritised either ethnicity or rural-
urban differences.’ They included books by Monica Wilson and Archie Mafeje
(1963), whose fieldwork in Langa was undertaken between 1955 and 1962; Leo
Kuper (1965), who did research in Durban between 1958 and 1961; Mia Brandel-
Syrier (1971), who worked on an East Rand township between 1958 and 1967;
Thomas Nyquist (1983), who conducted fieldwork in Grahamstown in 1966/67
and 1975; and Philip Mayer (1977), whose research in Soweto was undertaken in
1965 and again in 1975/76.
The subject matter of these accounts was similar, although the terminology
varied. Wilson and Mafeje write about ‘class’; Kuper was particularly interested in
the way that status (meaning ‘prestige’) was associated with occupation, but he is
not consistent and also talks about ‘class’; Brandel-Syrier’s account is about strata,
but she generally refers to ‘class’ or ‘status’ (which she used interchangeably);
Nyquist’s study ranks people according to occupation; and Mayer (1977: 16) analyses
‘class’ perceptions, which he said were ‘basically perceptions of status groups’.
Wilson and Mafeje (1963: 13–46) show people distinguishing between three
groups: migrant labourers, semi-urbanised folks and ‘townsmen’. The latter group
included ‘decent people’ (who ranged from university students, through secretaries,
nurses, ministers of religion and teachers, to lawyers and doctors), other respectable
people (mainly small traders), and a younger generation of ‘location boys’ and
‘girls’ that included factory workers. Kuper (1965: 127–9) includes a survey of 99
teachers, 77 of whom thought that ‘class differences’ existed. Of these, 34 said there
were three classes, sixteen said there were two and fourteen said four. His sample
was small and selective, but the proportions are remarkably similar to those we
found in Soweto. The way people justified these numbers varied greatly, but most
included an educated grouping – similar to the ‘decent people’ – as the top class.
Brandel-Syrier’s (1971: xxvi–vii) informants generally offered a three-class model
as well, but, she notes, the classificatory indicators varied, with locals offering finer
distinctions than Africans with a ‘national point of view’. In Nyquist’s (1983: 21–
5) study, 86 per cent of 301 respondents said that there were three ‘levels’ among
Africans in Grahamstown (the high, the middle and the low). Those regarded as
having a ‘high’ ranking were, once more, similar to Wilson and Mafeje’s ‘decent
people’ (with doctors at the top), but they also included carpenters; the ‘middle’
class included other artisans, police constables, shop assistants, waiters, petrol
attendants and, interestingly, domestic servants; and the ‘lower’ class comprised
herbalists, labourers, shebeen queens and, at the very bottom, latrine workers.
Mayer’s analysis, which is particularly pertinent (and is considered again in
Chapters 2 and 8), adds a new dimension to Brandel-Syrier’s notion of ‘point of
view’. He shows that, whereas the ‘class model’ – which identified an exploitative
relationship between white capitalists and black workers – was articulated by
‘ordinary working people’, educated respondents tended to express themselves in
terms of racial domination (Mayer 1977: 23–32).25 Within Soweto, working people
saw themselves as near the ‘middle’, but the educated and better off also placed
themselves at the centre, pushing ‘ordinary people’ towards the bottom (Mayer
1977: 123–4). The ‘educated rich’ were sometimes distinguished from the ‘rich
rich’, referred to as ‘tycoons’, who, Mayer (1977: 106) writes, were often no more
than shopkeepers. Lower down the hierarchy than factory workers came labourers
and domestic servants, who considered themselves, and were considered by others,
to be ‘poor people’; and beneath them there were the ‘very poor’, the destitute,
‘people who live through suffering’ (Mayer 1977: 111).26
We can add a few further points that provide valuable context for understanding
Soweto today – sometimes suggesting a counterpoint, more often a continuity.
Firstly, in all these cases, prestige was linked more to education than to money, but
in practice the better off were nearly always the better educated (although this was
already changing in Soweto before 1976). Moreover, education and income were
related to occupation, so one can have sympathy for writers who, in neo-Weberian
terms, muddled class and status.27 Secondly, the top stratum was marked off by
two kinds of cultural difference: (1) refinement, in particular the use of English
(even for conversations between friends), and (2) conspicuous consumption in the
form of housing, cars, furniture, fashionable clothes and appearance (for instance,
hair straightening) (for examples, see Wilson and Mafeje 1963: 137–8; Kuper 1965:
112, 138; Brandel-Syrier 1971: 87, 94; Mayer 1977: 68). Traders might have the
income required for consumption, but they rarely achieved refinement. Thirdly,
this top stratum was small. In Brandel-Syrier’s (1971: 94) case, the educated elite
was as small as half a per cent of the township’s total population and the business
elite was smaller still. Numerically, teachers were the largest educated group, but
there were social distinctions among them and not all were considered part of the
elite (Brandel-Syrier 1971: 64–5; Nyquist 1983: 44). Fourthly, while small, the elite
– at least at its higher level – was relatively cohesive, engaging in common social
activities and often intermarrying (Wilson and Mafeje 1963: 138–40; Nyquist 1983:
96–110). The elite generally preferred the old mission churches (Methodist, Anglican,
Catholic, etc.), which had higher status than the Zionist and Ethiopian churches
18 Peter Alexander
(Kuper 1965: 99, 137; Nyquist 1983: 69). Fifthly, the unemployed are barely
mentioned in these studies – which is not surprising, given that unemployment
had not yet become a visible problem – but Mayer (1977: 102) notes that ‘the poor
suffer from spells of unemployment’. Lastly, with the exception of a few activists,
class divisions among Africans were regarded as separate from those among whites
and only a few Africans were better off than the poorest whites.
The reader will find commonalities between these accounts and our own,
including the prevalence of a three-class model, the association between church
type and status, the use of the word ‘tycoon’ and the prevalence of the phrase ‘live
through suffering’. These studies also show that there is nothing new about
consumerism, even if its relative importance has increased as a marker of status;
this at the expense of education, which, though, is still important. The combination
of exploitation and hierarchical models of class remains widespread in Soweto.
African petty bourgeoisie

As contestation over the ‘white working class’ waned, interest in the African petty
bourgeoisie expanded. Again, the debate had political significance; this time because
of implications for how one characterised the leadership of the liberation struggle
and for the kind of class alliances that should be supported. By ‘petty bourgeoisie’
writers mean something analogous to those defined above as ‘decent’, or ‘elite,’ or
‘top stratum’, or ‘better off’ (and sometimes ‘bourgeoisie’ or ‘upper class’). While
‘petty bourgeoisie’ is subjective, it is more precise than ‘middle class’, a vague identity
term claimed by a much larger number of Africans (and, of course, others generally).
From historical and ethnographic accounts it is clear that, until the 1970s, the
petty bourgeoisie was numerically small yet consistently provided leadership for
urban Africans in general and the ANC in particular. As part of a social history of
the Transvaal Native Congress in the years 1917–20, Philip Bonner proposed that
the petty bourgeoisie – whether professionals or owners of small businesses – could
be pulled either in the direction of capital or that of labour, and the way this was
resolved related to place and time. In his case, the ‘place’ was a colonial setting
where the petty bourgeoisie was ‘stunted and repressed’ and the boundary between
it and the working class was blurred; and the ‘time’ was one of an ‘immensely
powerful upsurge of working-class agitation’ (Bonner 1982: 305). The outcome,
Bonner (1982: 305) shows, was a leadership that had its own ‘sectional interests’
and placed its own ‘stamp on events’, but ‘were not mere manipulators’. They
were ‘radicalised’ though also ‘fragmented’ by events going on around them.
Bonner’s was an insightful application of a Marxist conception of class to an
important phenomenon at a particular conjuncture. In other moments, the petty
bourgeoisie might be differently formed, come under different influences and play
a different role in history.28
His account was followed by a more theoretical dispute about the growth of
the African petty bourgeoisie from the 1960s onwards. In attempting to assess the
extent and character of this expansion, Owen Crankshaw (1986) engaged with
contrasting theories offered by Wright and Nicos Poulantzas. We have already
encountered the former. The upshot of Poulantzas’s (1975) position was that all
those not engaged in direct production of surplus value – including, for instance,
routine white-collar workers – should be included as part of a ‘new petty bour-
geoisie’.29 Crankshaw’s (1986) estimate was that this ‘new petty bourgeoisie’
numbered about 35 per cent of wage earners by 1983. According to him, Wright
would have placed all but about 9 per cent of these among the proletariat or ‘semi-
autonomous employees’. The wider significance of the argument was that, if one
accepted the broad (that is, Poulantzian) definition, it helped justify the working
class holding back on its own demands in order to develop a broader alliance.
Actually, many of the ‘semi-autonomous employees’ could have been included as
part of the proletariat. This was specifically true of school teachers, who, by 1980 –
following the massification of African schooling and the transformation of teaching
into a mainly black occupation – were busy converting professional associations
into unions (Crankshaw 1997: 89; Amoako 2012).30 A further problem with the
statistics is that they refer to employees, thus excluding the unemployed, and hence
giving an inflated impression of the proportion of the African population who
could be regarded as petty bourgeois under any definition.
Bonginkosi ‘Blade’ Nzimande (1990; 1991) advanced the debate by distinguishing
divergent interests within the African petty bourgeoisie. For Nzimande (1990: 181–
209), there were four ‘strata’ (his term). The first was the bureaucratic petty
bourgeoisie, the senior functionaries in the Bantustans and townships. Secondly,
there was the civil petty bourgeoisie, that is, state employees (mostly teachers, nurses
and clerks), which was the largest of the four strata. Thirdly, there was a corporate
petty bourgeoisie, which ‘closely identifies with a free market’, was hostile to the
apartheid state and under pressure from the militancy of the working class. Finally,
there was the trading petty bourgeoisie, which included one component that was
closely associated with the bureaucratic petty bourgeoisie and another that was
said to be ‘autonomous’.31 Under the right conditions, the civil petty bourgeoisie,
part of the corporate petty bourgeoisie and the autonomous trading petty
bourgeoisie might ally with the working class. Nzimande’s categorisation had much
to offer. In particular, it took us beyond the horizons of the township, created a
better sense of diversity within the African petty bourgeoisie, and drew attention
20 Peter Alexander
to the impact of the state, as well as capital and labour. However, fundamental
changes have occurred since the study was completed in 1991, and one doubts
whether the ‘African petty bourgeoisie’, in the form described by Nzimande, still
exists in a meaningful way. While one element emerged as capitalists and rulers, a
much larger component is identifiable as workers.32
Recent academic research on the petty bourgeoisie is limited. However, an
interview-based study of black corporate managers by Geoffrey Modisha (2007)
yielded insights that are particularly relevant for our own account. All the managers
lived in the suburbs, but whereas the more junior ones came from trade union
backgrounds, lacked formal qualifications and had parents in traditional working-
class occupations, the more senior had formal qualifications and most had parents
in professional positions. An important experience that most of these managers
had in common, and which distinguished them from their white colleagues, was a
need to negotiate relationships with kin who were generally poorer and living in
townships. One explained: ‘I can’t actually get rich in a short period of time, because
I share the money with members of my extended families because they are not
working.’ Respondents who maintained links with their families living in townships
said things like: ‘People are always looking for employment from me . . . Some
people are crying in front of me.’ By contrast, the managers who came from a
professional background tended to cut ties with their relatives, with one explaining:
‘I struggle to live with those people because of my interpretation of reality.’
Modisha’s analysis recognises the importance of ‘levels’ within corporate
hierarchies, and his contrast between black and white managers shows how people
from dissimilar backgrounds experience the same class location in different ways.
Thus, he adds further complexity to the account of politically salient class
characteristics provided by Bonner, Nzimande and others. In sum, one’s specific
‘position’ within the petty bourgeoisie shapes one’s judgements, but so too does
personal history and the ‘pull’ of external forces (capital, labour and the state). In
the second decade of the twentieth century and in the 1980s, powerful working-
class movements were able to radicalise important sections of the petty bourgeoisie,
but is the working class still sufficiently coherent to have a similar impact today?
Unemployment and class analysis

Modisha’s interviews point to the anguish and suffering associated with poverty
and unemployment. Thus, he takes us to matters at the heart of South Africa’s
social crisis and to the main focus of current debate around class.
The problem of unemployment did not become apparent until the 1970s. This
was largely because influx controls ensured that jobless Africans were hidden away
in the Bantustans, mainly in rural resettlement slums. Urban unemployment had

been ‘displaced’, to quote Charles Simkins (1981). He and Cosmas Desmond (1978:
vii) state bluntly that from 1960 South African unemployment never fell below
one million. The rate rose steeply from the last years of the decade, dipped in the
mid-1970s, and then rose continuously through the 1980s (Crankshaw 1997: 104–
5).33 Unemployment underpinned the 1970s and 1980s struggle against apartheid in
two ways. Firstly, it provided a critical mass of young recruits for township
resistance. Secondly, the increased burden of unemployment wiped out the benefit
of improved wages, pushing workers to fight for higher pay, leading in turn to
stronger union organisation and heightened class consciousness. According to official
statistics, the expanded rate of unemployment (that is, including discouraged job
seekers) stood at 28.6 per cent in 1994; it peaked at 41.8 per cent in 2002; then went
down, but up again following the slump of 2008/09, reaching 33.8 per cent in early
2012 (a figure that represented 6.9 million adults of working age) (Altman 2003;
Stats SA 2012a, 2011a). South Africa’s level of unemployment is worse now than at
any time under apartheid and is more pronounced than anywhere else in the world.
The unemployed – and the ‘poor’ more generally – highlight a problem for
occupation-based class analysis: how does one ‘class’ people who have no occupation?
In the South African context, where the issue is presented in stark relief, this
conundrum has been addressed by some of the country’s leading sociologists. My
comments here are reinforced in Chapter 4 by Ceruti.
An important contribution to this debate was made by Seekings and Nicoli
Nattrass (2005), who began their analysis with Goldthorpe’s revised scheme, and
modified this to take account of South African conditions. Their main theoretical
innovation was to move from an assessment based on individuals to one based on
households, which they justified by reference to Wright’s notion of mediation.
Everybody in a household was allocated to the occupation of its ‘dominant’ member.
This was modified to take account of household income from wealth and
entrepreneurial activity (so long as this was greater than a government pension).34
This left a residual category that included households in which nobody was
employed and where income from these other sources was negligible. From here,
Seekings and Nattrass (2005) mapped the ‘class structure’, which they presented in
the form of a pyramid with three layers, and then used this model to summarise
changes that occurred with the ending of apartheid. They found that the top layer
became deracialised and better off; the middle, which included the ‘core working
class’, was also generally better off, but had declined in size; and the bottom (which
included the ‘marginal working class’ and the residual category) was mostly worse
22 Peter Alexander
off and had grown in size.35 The approach is responsive to South African conditions,
and as a means of describing significant aspects of social transformation it is helpful.
But if one wants to explore the relationship of ‘classes’ to subjectivities, we need an
account that recognises that people are individuals as well as members of households.
Further, the dominance principle undervalues the size and impact of poverty by
allocating many of the unemployed, for instance, to a ‘higher’ class. Moreover,
what individuals can ‘afford’ will be determined by the composition of the whole
household (particularly its size and number of income earners) and not just the
occupation of its dominant member and income from wealth and entrepreneurship.
Seekings and Nattrass (2005) also took up the issue of whether or not part of
the unemployed constituted an underclass. Their response disaggregated their
residual, defining as ‘underclass’ only ‘households with no employed members
and thus no social capital’ (Seekings and Nattrass 2005: 289). This solution is elegant
but, as with Wright’s (1978: 93) attempt to grapple with the same issue, it is also
problematic.36 Lack of employment in a household is not the same as absence of
social capital – people can be jobless and still have friends, go to church, attend the
weddings and funerals of their extended family, and receive remittances from a
mother or father who works away from home.37 Being part of a poverty-stricken
household is likely to have an adverse impact on life chances, but poverty is
widespread and cannot be reduced to households without any employed members.
In a recent reflection, Seekings (2008a: 23) insists that there is ‘a discrete “underclass”
or lower class separate from the “working class” ’, but acknowledges that this ‘can
be identified by any number of criteria’ and ‘the appropriate label for this class can
be debated’. It would seem, then, that a problem remains. I would certainly not
wish to suggest that this is an easy matter but, from the perspective of the questions
we are asking, it is possible to simply recognise that some people are unemployed.
Difficulties arise when one merges ‘class’ and what are essentially occupation-based
categories.38
An alternative theorisation was advanced by Karl von Holdt and Edward
Webster (2005), who represent their position diagrammatically in the form of an
onion (reproduced in Chapter 4). This includes a core of full-time workers; a ring
of outsourced, temporary, part-time and domestic workers; a further ring of informal
workers; and an outer layer comprising the unemployed (Von Holdt and Webster
2005: 28; see also Webster 2006; May and Meth 2007). This model refocuses attention
on individuals, it shows movement between the different layers and it avoids slicing
off a separate underclass. However, while households are mentioned, they are not
integrated into the model and the two in five South African adults ‘not in the
labour force’ are missing. The strength of the argument – and it is considerable – is
its concern with organisational questions. The limitation of trade unions (specifically
COSATU) in failing to move beyond the core is recognised, and the authors advocate
the importance of community-based movements that mobilise people outside the
nucleus. For ‘a counter movement to truly be effective’, they argue, it should link
workplace and community struggles, adding: ‘in other words, uniting the trade
union movement and social movements in a broad coalition against the destructive
impact of the market on society’ (Von Holdt and Webster 2005: 38). Unfortunately,
their view of organisation is somewhat formalistic, with social movements seen as
the equivalent of trade unions. Mass organisation among the jobless is far more
difficult to sustain than among workers, and successful mobilisation tends to be
short term, thus posing critical strategic and tactical dilemmas.
Von Holdt (2011: 6) has now broadened his analysis, proposing that ‘a rapid
process of class formation’ is underway in South Africa. He points to an important
shift: among black people, in particular, there is a growing gulf between those
who, at the one end of the spectrum, have become part of the ruling class, either as
capitalists (still far smaller in numbers and less wealthy than their white counterparts)
or as part of the governing elite, and, at the other end, those who Von Holdt
describes as the ‘unemployed and precariously employed’. One aspect of this
phenomenon is migration from townships to suburbs, involving, according to a
2008 study, some 12 000 black families per month, in the six largest cities alone
(National Planning Commission 2011).39 An associated process is access to better-
performing schools, which has implications for entry to better-paid jobs (Simkins
2005). But it is difficult to see how these processes amount to the creation of new
classes.
Main findings
We now consider how our own research adds to the literature discussed above.
In Chapter 2 Kim Wale provides an overview of the history of ‘class’ in Soweto.
The township’s origins were as a residential area designed mainly for workers living
in married circumstances. In 1946 the council provided space for the educated elite
in an area known as Dube, and from the late 1950s hostels were built for migrant
workers, adding another component to Soweto’s social mix. While restrictions on
the development of the black middle class led to a measure of ‘class compression’,
the Soweto uprising of 1976 was followed by attempts to improve the position of
better-off urbanites and to lock the new unions, growing apace, into state-sponsored
negotiating procedures. But the plan backfired. Throughout the 1980s, the elite
provided leadership for a community whose school students and unemployed youth
were becoming increasingly militant. Meanwhile, workers advanced a class struggle
24 Peter Alexander
that merged with activity in the township to create a force that was central to the
overthrow of apartheid. In the post-apartheid years, there has been significant class
differentiation, with better-off people generally moving out of the township and
increased poverty occurring within it. Before the end of apartheid, growing numbers
of people were living in shacks, and many people still live in such conditions.
In Chapter 3 Ceruti paints a picture of contemporary Soweto that draws mainly
on quantitative data, much of it gathered as part of our 2006 survey. She shows that
the population of the township was growing, in part because of continuing in-
migration. On a number of indicators, its residents tended to be a little better off
than other South Africans, but the rates of unemployment were higher than the
average for the country. Housing provides the main focus of the chapter. Here the
evidence lends support to findings from other researchers that there is growing
differentiation. A minority live in relatively costly new houses or in substantially
improved older ones, while the majority reside in former council houses that have
not been upgraded, backyard rooms, shacks and hostels. However, Ceruti cautions
against an oversimplified view that equates better housing types with work status
or higher income, showing that other factors contribute to the differentiation,
including household size and when people came to Soweto. Significantly, people
without work live in close proximity to others with regular salaries. Indeed, among
our respondents, those living in backyard shacks were more likely to be receiving
a regular wage than to be unemployed, and the proportion of unemployed was
above average for Soweto among those living in unrenovated former council houses
(which are superior in quality to shacks).
In Chapter 4 Ceruti extends her account onto the terrain of employment and
unemployment. Drawing on the survey, she shows that only 24 per cent of Sowetans
were ‘regularly employed workers’, with a further 7 per cent categorised as petty
bourgeoisie, managers or capitalists. The remaining 69 per cent were either fully
unemployed, engaged in survivalist activities, or outside the labour force by virtue
of being students or pensioners. In these circumstances, an occupation-aggregate
approach to class analysis is unhelpful, because it leaves a large majority of the
population ‘unclassed’, unless, perhaps, they are treated as appendages of people
with jobs (which, as we have seen, creates further problems). Ceruti opts for an
approach that emphasises the ‘community of fate’ that exists between those who
are employed and those without regular jobs. The critical issue is the role played
by households in ensuring survival through the redistribution of resources, with
the effect that the ratio of earners to dependants can have a considerable impact on
experiences of poverty. Ceruti develops Marx’s concept of the labour reserve and
surplus population to propose a dynamic-systemic model of class, where different
experiences of exploitation may be crosscut by common experiences of home life.

For Ceruti, Soweto is overwhelmingly a differentiated proletarian township, where
‘proletarian’ is taken to mean ‘those without property’. While differences between
workers and the unemployed do not amount to class differences, differentiation
tends to obscure commonality in the popular imagination.
Chapter 5 rounds off the discussion of class and employment. It does so through
an analysis of ‘underemployment’ undertaken by Wale and myself. The ‘under-
employed’ include two components: the survivalist self-employed and partial
workers. We show that the households of the underemployed tend to be worse off
than those of the unemployed. To explain this, we draw on an Indian aphorism
that ‘the poor are too poor to be unemployed’. That is, for most adults,
unemployment is only possible if a member of one’s family has some source of
income, whether it is a wage or a grant of some kind. The implication is that
government statistics for unemployment – even those based on the expanded
measure – are an unsatisfactory indicator of social well-being. The labour absorption
rate – the percentage of the working-age population in employment – is more useful,
and for South Africa this is the lowest for any major country, at just over 40 per
cent. But, on interrogating this figure further, we find that only 27 per cent of the
working-age population is formally employed.
In Chapter 6 we move the focus to ‘subjective’ aspects of class as they were
reflected in our survey and qualitative research in Chapter 6. The data presented by
Mosa Phadi and Ceruti shows that a very high proportion of Soweto’s adults have
some class awareness. For instance, extrapolating from our survey, 93 per cent of
Sowetans gave themselves at least one class label. However, only 38 per cent accepted
just one label. The main forms of multiple class identity were (1) poor/lower
combined with middle and (2) working combined with middle, both of which are
reasonable responses in the context of what we came to know about class identity
in Soweto. The dominant model was based on three classes, with some kind of
upper, middle and lower (including poor) classes. ‘Working class’ was a
supplementary identity. In qualitative interviews the upper class was pictured as
comprising those with a ‘luxury life’, but in the survey those claiming the identity,
a small minority, were not especially well off and were distinguished mainly by
their self-confidence. The middle-class identity was nebulous. While it represented
those who were neither ‘upper’ nor ‘lower’, the precise definitions varied. The
poor or lower class – the two were used as synonyms – were indeed poorer than
other people, and this included lower levels of educational attainment. Extrapolating
from the survey, 66 per cent of Sowetans claimed a middle-class identity, with
‘working class’ coming second at 43 per cent.
26 Peter Alexander
In Chapter 7 Wale draws on her reading of interview transcripts to develop an

argument that links the pivotal concept of affordability with different forms of
class consciousness present in respondents’ worldviews. She begins by reminding
us that affordability mediates class as economic difference and class as cultural
difference. Then, using the metaphor of a ladder, she argues that when people look
‘upwards’, they tend to emphasise cultural and individual expressions of class, but
when they look downwards, poverty and lack of work come to the fore. Wale’s
account then moves to a discussion of competing worldviews and their relationship
to the ladder. The implications of her account are contradictory. On the one hand,
class difference is not necessarily seen as something bad: it may be associated with
a system that is basically just, and is thus politicised in a way that reinforces the
status quo. On the other hand, where mobility is stunted, either through economic
impediments or through race, feelings of injustice increase and with them the
potential that class, perhaps alongside race, can be mobilised by a counter-hegemonic
movement. Thus, affordability matters, not only as a concept that permits a holistic
understanding of class, but also as a reality that is pregnant with possibilities for
social change.
In Chapter 8 Phadi and Owen Manda explore how Sowetans’ understandings
of class are shaped by terms available to them in their indigenous languages.
Raymond Williams (1976), discussed by the authors, regarded ‘class’ as a ‘difficult
word’. In South Africa, difficulty is magnified by the interaction of indigenous
terminology, originally constructed in particular rural contexts, with this already
‘difficult word’. Two empirical issues are considered. Firstly, there is no general
term that can be equated with ‘class’ as used in a primarily economic sense. The
nearest counterparts convey implications of standing, status or level. Secondly,
while there are no direct equivalents for specific class terms, several words are
associated with poverty and a number are related to wealth. But ‘middle class’
poses a particular problem because the closest translation in indigenous South
African languages means ‘in the middle’, creating possibilities for numerous
interpretations. This helps to explain why such a high proportion of Sowetans
consider themselves to be middle class. ‘Working class’ creates further ‘difficulty’,
with people tending to translate it as meaning ‘those who work’, thus limiting the
potential for developing a class vocabulary with broad emancipatory implications.
This chapter has wider significance because it reveals the inadequacy of imposing
an English-language template (with all its ‘difficulty’) on a non-English setting,
where people are likely to be framing their worldviews according to different
standards and where a common term – ‘middle class’, for instance – may have
different meanings.
Chapter 9, by Keke Motseke and Sibongile Mazibuko, analyses the relationship

between class and church. This is important because our survey showed that 87 per
cent of Sowetans were Christians (more than the national average) and almost half
the population had attended church in the previous seven days. In terms of affiliation,
the mainline churches (such as Catholic and Anglican) have less support than
reflected in ethnographic accounts reviewed above, with only 26 per cent of the
population associating with them. These have been overtaken by the African
Independent Churches (AICs) (27 per cent of the population), and the Pentacostals
are growing (12 per cent). Firstly, as has been true historically, there is a tendency
for poorer people to be attracted to the AICs and for those who are better off to
associate with mainline churches. Secondly, while the significance of class was widely
denied and in some churches wealthier congregants helped those who were poorer,
class differences were often manifest in clothing, seating position and the kind of
people who took leadership positions. The general proposition is that people attend
a church where they feel socially accepted and where their needs and aspirations
are nourished, and this means that class is implicated in church attendance.
Chapter 10 provides my conclusion. There are two appendices. In Appendix 1,
Ceruti and I describe the methods used in this study. In particular, we provide
details about the quantitative survey: definition of population, questionnaire design,
sampling procedures, data collection, response rate and weighting. In Appendix 2,
Ceruti provides additional information about Sowetans drawn from the survey.
This includes demographics, levels of educational attainment, mother tongue and
political support. Only 9 per cent of the population were members of a trade
union, but 88 per cent ‘felt close’ to the unemployed. Finally, the film Phakathi:
Soweto’s Middling Class is available from the South African Research Chair in Social
Change at the University of Johannesburg and can be viewed on www.youtube.com.
This documentary, made by Phadi and others as part of the project, explores the
reasons why such a high proportion of Sowetans regarded themselves as middle
class.
Summary
The volume’s conclusions are presented in the final chapter. Here I draw together
aspects of the theoretical framework, relevant literature and chapter outlines as a
means of signalling key issues.
Our ‘anchoring question’ was about emancipation – posed in the context of
South Africa’s acute social inequalities and intense but separate struggles by workers
and the poor. The methodology, broadly Marxist in character, aimed at exploring
the relationship between an objective marker of employment relations, aspects of
28 Peter Alexander
social reproduction and various classed subjectivities. In Soweto, a large majority

of the adult population is either unemployed, underemployed or not in the labour
force, thus exposing the need for a model that moves beyond occupational
classification. For us, assigning the class position of the ‘dominant’ member of a
household to each member of that household did not provide a solution, because it
is individuals, rather than households, who have class identities and engage in
struggle. Our approach involved developing nine ECs, with these capable of
accommodating the entirety of our sample.
Our questions on identity revealed that more than 90 per cent of Soweto’s
population could define themselves in class terms. Typically, people thought there
were three classes – that is, some version of lower, middle and upper class – but,
when asked about their own identity, the second largest number described
themselves as working class. Initial qualitative research had led us to ask about
identification in a way that was novel in two respects. First, the class labels offered
were derived from interviews rather than preconceived, and, secondly, respondents
could select more than one label. In practice, most people selected at least one such
label (for example, poor and middle class or working class and middle class). One
can hypothesise that an important reason for multiple class identification is that
‘class’ names two distinct experiences. It distinguishes a worker from those who
benefit by exerting control over their labour, but it also distinguishes those regarded
as in some measure better off from those who are less so (commonly producing
three classes). Moreover, the boundaries within these experiences are often unclear.
This can be true of relational class, especially in large corporations and the state,
with their long chains of command and separation of management from ownership.
However, it is particularly the case with hierarchical class, where the signifiers of
difference are varied (such as income, education, housing, clothes, language/accent).
People are making sense of different social realities, and thus identification with
two or more classes is rational. That is, while people recognise the importance of
class (however defined), they often have difficulty aligning themselves with a single
class label because class identities are, genuinely and for good reason, associated
with more than one phenomenon.
There is also an objective basis for class identities. ‘Working class’ was associated
with having a job, and the most popular identity of regularly employed workers
was ‘working class’. By contrast, people describing themselves as ‘lower class’ tended
to be more deprived than those who did not. Popular perceptions of class generally
referenced ‘affordability’, a word used in English that also had mother-tongue
equivalents.40 For instance, the ‘middle class’ were said to be ‘in the middle’ between
those who could afford everything (the upper class, rich, etc. ) and those who were
suffering (the lower class, poor etc.). ‘Affordability’ provides a link from subjective
experiences (including classed culture and verbalised class identities), through income,
to occupation (or lack of occupation). Thus, it connects consumption and
production. In addition, while people mention individual attributes and cultural
associations when they look ‘upwards’ in class terms, they refer to lack of basic
resources when they look ‘down’. This reinforces suggestions in some sociological
literature that the way people define ‘class’ might be affected by their class location,
with those defined as ‘middle class’ mainly concerned with cultural referents and
the ‘working class’ and ‘poor’ emphasising the importance of economic markers.
An important twist in our account of class identity arose when considering the
mother-tongue terms that were available when people were translating class-related
words into English. Notably, this showed that when people use the term ‘middle
class’, they are very often translating from a word that means ‘in the middle’,
which is far more inclusive. This, in turn, helps to explain why ‘middle class’ was
the most popular class identity.
With regularly employed workers tending to identify themselves as ‘working
class’, and pensioners, the underemployed and unemployed more likely to describe
themselves as ‘poor’, it might seem that there is a firm basis for distinguishing a
working class from a poor or lower class. But this is not the position we adopted.
Within Soweto, there was little evidence of antagonism associated with ECs or
class identity, and better-off people were generally sympathetic to those who were
poorer. This reflected life histories and the close physical proximity of workers
and the unemployed. Households tended to re-distribute resources from better-
placed family members to those who were poorer. Kin, friends, neighbours, churches
and so on provided material support in the form of things like food, school fees
and loans, but also, for instance, information about job opportunities. In terms of
accommodation, there were no sharp divisions between the kind of housing occupied
by, for example, the unemployed and the regularly waged. In terms of identity,
most people were not either ‘working class’ or ‘poor’, rather they described
themselves as ‘middle class’ or accepted more than one class label. So, while there is
a gap between ‘workers’ and ‘the poor’, it is mitigated by a range of factors.
In the sense that Sowetans, bar a small minority, are either workers, ‘wannabe’
workers (including subsistence self-employed), retired workers or directly dependent
on workers (including most students), Soweto is predominantly a proletarian
township. However, there is considerable variation. It is a differentiated unity.
Common interests are cut through by different daily experiences of work, survival
and inter-personal relationships, and by the different capacities to mobilise to change
30 Peter Alexander
those experiences. But the long-term interests of workers and the poor are similar,
unlike those of workers and capitalists.
Our research was framed by concerns about divisions between workers and
the poor that existed in the face of massive inequality, and it was informed by
theoretical assumptions and a methodological stance derived from Marxism. How
well does Marxism stand up to challenges posed by the findings and analysis
presented in the chapters that follow? A response to this question is offered in
Chapter 10. Here, though, we can acknowledge that for Marx and Engels tasks of
the kind we have undertaken were not merely about defining and theorising
problems, they were also about helping to chart a solution. Their conclusions
recognised the importance of ‘class’ as part of the struggle for a classless society. We
make no pretence to have produced a book with a different ending, but I hope the
reader finds merit in the evidence and argument we use to get there.
Notes
1. This ratio assumes that the population of Soweto at the time of writing (August 2012)
was nearly 1.5 million, which is a very rough approximation based on figures presented
in Chapter 3 and a measure of subsequent growth similar to that for Gauteng (the
province in which Soweto is located). More reliable numbers will be available once
the results of the national census conducted in October 2011 are published. For the
broader Johannesburg context, see Beall, Crankshaw and Parnell (2002b).
2. In a Cape Town study, where respondents were asked to choose between one of four
classes, only 2 per cent failed to choose one of the four (Seekings 2007a: 14).
3. Wright regards ‘life chances’ as a secondary anchor, reflecting, perhaps, the influence
of Max Weber.
4. There is no intention here to equate Weber’s ‘Class, status, party’ and Bourdieu’s
Distinction (where the concepts of, respectively, ‘status group’ and ‘habitus’ are
explained). The former is vast and unspecific in its reach (though short in length); the
latter is time and place specific (though very long). ‘Class, status, party’ appeared as a
chapter in Economy and Society, first published in 1921, but was probably written
before the First World War according to Dagmar Waters et al. (Weber 2010). Distinction
was not published until 1979 (in French), but it is based on survey data from 1963 and
1967–68 and, according to Richard Jenkins (2002: 74), Bourdieu’s use of habitus dates
from 1967. Bourdieu (1984: xi, 513), himself, drew attention to the fact that his survey
‘measured relatively stable dispositions’ and also to the ‘Frenchness’ of his book.
5. For instance, Engels’s (1969) descriptions of various components of the working class
in The Condition of the Working Class in England, his and Marx’s (1965) discussion of
‘petty-bourgeois socialism’ in the Communist Manifesto, and Marx’s (1954: 617)
references to ‘lower middle class housing’.
6. On production/reproduction I am not arguing anything different from many other

Marxists (including Wolpe 1972; Cockburn 1977; Harvey 2012).
7. The references for this claim are as follows. By Marx: ‘Critical notes on the article
“The king of Prussia and social reform. By a Prussian” ’, Vorwarts, 7 August 1844;
‘Letter to Ludwig Feuerbach in Bruckberg, 11 August 1844’; and ‘Preface’ to A
Contribution to the Critique of Political Economy, 1859. By Engels: ‘Ludwig Feuerbach
and the end of classical German philosophy’, 21 February 1881. I have used the versions
of these sources available at http://www.marxists.org (accessed 29 June 2011). Engels
reached Paris on 28 August 1844. In passing, it is worth noting that in his preface to
the first German edition of The Condition of the Working Class in England, Engels
(1845) recorded: ‘I have continually used the expressions working men (Arbeiter) and
working-class, propertyless class and proletariat as equivalents.’
8. The recognition that people hold multiple identities is now so well entrenched in
sociological literature that we have been unable to cite an originating source. However,
the basic idea is present in Antonio Gramsci’s (1971: 333) invaluable concept
‘contradictory consciousness’.
9. Combined, these three minorities comprise about 25 per cent of Johannesburg’s
population (Gaffney Group 2012: M289).
10. From 13 844 000 in 2008 fourth quarter to 12 803 000 in 2010 first quarter (Stats SA:
2009d; Stats SA 2010a).
11. Surveys conducted by TNS between February 2010 and May 2012 showed an expansion
in the proportion of the eight metro populations who were ‘unhappy’ with service
delivery, and, in the case of Soweto, the increase was from 49 per cent to 61 per cent
(SA Local Government Briefing 2012: 32).
12. Wright’s approach produced three ‘unambiguous’ classes (bourgeoisie, proletariat and
petty bourgeoisie) and one contradictory class location positioned between each pair
of these (that is, managers and supervisors, semi-autonomous employees and small
employers). Our capitalists were, in Wright’s (1978) terms, ‘small employers’; and his
‘semi-autonomous workers’ were accommodated either as managers, petty bourgeoisie
or regularly employed workers. The unemployed, pensioners and students do figure
in Wright’s (1978: 91–6) broader framework, but not as part of his main scheme.
13. Bourdieu (1984: 175) regards as ‘homologous’ the associations that exist across different
‘areas of practice’. These homologies ‘are all homologous to one another because they
are all homologous to the structure of objective oppositions between class conditions’.
According to Bourdieu, this alignment of classes across different areas of practice is a
feature of class habitus.
14. In practice, status order was operationalised mainly by ranking jobs (which turns out
to be correlated with class positions).
15. A 2006 survey by Future Foundation (2006) found that 53 per cent of Britons regarded
themselves as working class and 43 per cent as middle class. The study had discouraged
what it called ‘shyness’ by asking: ‘Would you call yourself working class, middle
32 Peter Alexander
class or upper class?’ Even so, it complained that there was ‘confusion’ among people
about which class they belonged to (for example, bank managers who said they were
working class) and it decried a tendency towards the ‘muddle class’ [sic]. Even so,
people who said they were middle class had twice the level of savings and three times
the level of investments than those who said they were working class. Unfortunately,
the sample was small – only 1 000 people for the whole country – but the figures were
similar to those of the British Household Panel Study.
16. This is not to say that a high level of working-class identification exists everywhere.
For instance, it has been reported that in Japan, about 90 per cent of people regard
themselves as middle class (Savage 2000: 35), while for India, Deshpande (2003: 129–
31) suggests that ‘middle class’ is a desirable identity to which even those who are
‘ineligible’ lay claim. It is possible that, as in South Africa, language and translation
play at least some part in the construction of class identity.
17. The General Social Survey poses the following question: ‘If you were asked to use one
of four names for your social class, which would you say you belong in: the lower
class, the working class, the middle class, or the upper class?’ According to Hout
(2008: 29), for the years 2000–04 the question received a 99.4 per cent response rate and
yielded the following replies: lower, 5 per cent; middle, 47 per cent; working, 44 per
cent; upper, 4 per cent. Hout adds that when the exact same question was asked by
Richard Centers in 1949, 49 per cent self-identified as working class and 45 per cent as
middle class. It is remarkable how little US class identifications have changed over
more than half a century.
18. Similarly, Deshpande (2003: 140) makes the point that ‘the middle class more than
any other is defined by its ownership and control of cultural capital’.
19. We might reasonably be criticised for underestimating the importance of southern
literature in this introduction. However, the purpose here has been to consider the
most influential texts. Separately, in 2009, we held a conference on ‘Comprehending
Class’ in Johannesburg. This produced three volumes of papers and led to a special
issue of the South African Review of Sociology, which included, in addition to two articles
on South Africa (Phadi and Manda 2010; Moodie 2010), papers on Brazil, India and
China (respectively, Santos 2010; Dhawan 2010; Chan 2010).
20. Van den Bergh was not always consistent. At one point, he envisions the possibility of
a revolution led by the black proletariat, but adds, perspicaciously, that this was unlikely
‘to result in rapidly rising living standards for the African masses . . . [and] the
intelligentsia probably stands most to gain through change’ (Van den Bergh 1965:
210).
21. Seekings (2009: 878) makes the further point that key figures either emigrated (including
Kuper) or died. Sandwiched between the state and the movement, the environment
was harsh for liberals, Weberians included.
22. I have argued against the free/unfree thesis elsewhere (Alexander 2006b). This distinction
is not grounded in Marxism, and the account assumes, wrongly, that all black workers
were like slaves or serfs.
23. This history was strongly influenced by Thompson (1968). Attempts to apply his
notion of ‘class experience’ – which mediated consciousness and exploitation – could
have made a valuable contribution to the debate (Alexander 1999: 32). The distinction
between Marxists and Weberians – or, at least, between some Marxists and some
Weberians – can be drawn too sharply. For example, the Weberian John Rex (1986:
66–7) distinguished between black workers and white workers in terms of a ‘differential
relationship . . . to the means of production’, with the former ‘unfree’ and the latter
‘[part of a] community . . . able to defend real privileges’. This was only a shade different
from the Marxism countered by Wolpe (1976: 200–3; Rex 1986: 68).
24. Unfortunately, Wolpe had blind spots. In particular, because he tended to be dismissive
of history and loyal to the main wing of the liberation struggle, his analysis of
conjuncture was impaired. His new thinking fed into a book, completed in February
1987, that reached a pessimistic conclusion about the possibility of ending apartheid
rule, yet eight months later, and before the book was published (Wolpe 1988), he was
himself involved in negotiations that would bring apartheid to an end (Alexander
2007: 114).
25. It is not always clear whether Mayer is writing about 1965 or 1976 or both, but here he
is referring to 1965.
26. Mayer’s monograph, which exists only as a manuscript, was not available to us when
we completed our main fieldwork, so similarities between his findings and our own
are all the more remarkable.
27. Weber’s (1924: 181) economic classes consisted of ‘people [who] have in common a
specific causal component of their life chances, in so far as this component is represented
exclusively by economic interests’. These economic classes were aggregated into four
major social classes: the working class, petty bourgeoisie, technicians and lower managers,
and propertied elite (Breen 2005: 32; Crompton 2008: 33). Seekings (2009: 868) notes:
‘[E]ven Weber himself tended to elide class and status in practice.’
28. The role of the petty bourgeoisie has been a continuing interest of Bonner’s (see Bonner
and Nieftagodien: 2008). For a history of the black petty bourgeoisie, see Alan Cobley
(1990).
29. Although not discussed here, Poulantzas’s work provided a point of departure for
many South African Marxists, including Rob Davies (1979). In the 1980s, Wolpe (1988)
came to see value in Poulantzas’s approach, but was highly critical of Davies’s
interpretation.
30. According to Seekings and Nattrass (2005: 104), ‘In greater Soweto . . . there were
only eight secondary schools until 1972. By 1976 there were twenty, with three times
as many students as in 1972. By the end of 1984 there were fifty-five [schools].’ Transvaal
African teachers had established their first association by 1920, and held a strike as far
back as 1944 (Amoako 2012).
31. Nzimande (1990: 197) reports that there were 3 000 small shopkeepers in Soweto in
the late 1980s.
34 Peter Alexander
32. For an appraisal of the relevance of Nzimande’s categories in post-apartheid South

Africa, see Roger Southall (2004).
33. For a review of the literature on apartheid unemployment, see Nattrass and Seekings
(2005).
34. ‘Dominant’ is a non-gendered concept developed by Goldthorpe and others (Edgell
1993: 47).
35. By 2008, there had been further shrinkage in the size of the ‘core working class’ (that
is, unskilled and semi-skilled manual workers outside agriculture and domestic work),
and continued growth in the size of the bottom layer (Seekings 2011: 538).
36. For Wright (1978: 93) the temporarily unemployed could be equated to the reserve
army and did not pose a ‘special problem’, but the permanently unemployed were
‘more problematic’, and, as far as I can tell, the problem was never resolved, other
than by regarding them as an ‘underclass’.
37. Moreover, the authors’ own evidence contradicts their assumption. They include a
table based on a 2000 Cape Town survey which shows that the main way people got
their first job was through a friend or a relative in a different household (45 per cent),
while getting it through a household member happened much less often (19 per cent)
(Seekings and Nattrass 2005: 285).
38. Ben Roberts (2001: 110), who is open to the possibility that underclasses exist, makes
the observation that, at least in Britain, ‘as yet the best reason for having an underclass
. . . is to maintain the (sociological) purity of the working class’.
39. The National Planning Commission cites the UCT Unilever Institute as its source.
40. For instance, the isiZulu ngiyazikhona, meaning ‘I can afford’. Significantly, this word
has a second meaning: ‘I can take care of myself.’ Thanks to Trevor Ngwane for
clarifying this matter.

Class in Soweto

Uploaded by

Copyright:

Class in Soweto

Uploaded by

Document Information

Copyright

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Class in Soweto

Uploaded by

Copyright:

Class in Soweto

Peter Alexander, Claire Ceruti, Keke Motseke,

Photographs fall between pages 84 and 85

1 Affordability and action: Introduction and overview 1

2 Historical introduction to class in Soweto 35

3 Contemporary Soweto: Dimensions of stratification 55

4 A proletarian township: Work, home and class 96

5 Underemployment: Too poor to be unemployed 127

6 Models, labels and affordability 142

7 Perceptions of class mobility 164

8 The language of class: Confusion, complexity and difficult words 190

9 Class and religion: Denominations in Soweto 210

Appendix 1: Methodolgy 249

Appendix 2: Sowetans in the Classifying Soweto survey 259

Affordability and action

The book is innovative in a number of respects. With 69 per cent of adult

Theoretical and methodological framework

or merely empirically relevant. These six questions are about distribution of

narrowly on class subjectivities. (3) Formulations that clearly distinguish between

any survey-based project). Secondly, we administered our questionnaire and

Some relevant international literature

In response to culturalist approaches, Goldthorpe and Tak Wing Chan (2007;

. . . when we look at things like risk of unemployment or the long-term

legitimacy to arguments on the injustice of domination by a rich minority and, by

Some views of class in South Africa

Structure of class relations and the white working class

. . . the criterion of ‘propertylessness’ is insufficient to establish the class

to be stereotypical of white workers (Simson 1973); another argued that a petty

. . . while at one level classes must be conceived of as unitary entities,

Class in the townships

African petty bourgeoisie

Unemployment and class analysis

in the Bantustans, mainly in rural resettlement slums. Urban unemployment had

experiences of exploitation may be crosscut by common experiences of home life.

In Chapter 7 Wale draws on her reading of interview transcripts to develop an

Chapter 9, by Keke Motseke and Sibongile Mazibuko, analyses the relationship

social reproduction and various classed subjectivities. In Soweto, a large majority

6. On production/reproduction I am not arguing anything different from many other

32. For an appraisal of the relevance of Nzimande’s categories in post-apartheid South

You might also like