Making Sense of the Social,Making the ‘Social Sense’: The Development of Children’s Perception and Judgement of Social Class

Introduction

Few subjects have proven more stimulating to the sociological imagination than that of ‘social inequality’. ¹ In a supposedly interdisciplinary age where research topics increasingly skip across the frontiers of individual disciplines, the study of social inequality remains firmly anchored in sociology’s heartland. While economists or psychologists occasionally stray into the field (Mullainathan and Shafir, 2013; Piketty and Goldhammer, 2014), it is among sociologists that research on inequality retains a privileged status. More than others, we chart its reproduction across generations, trace its evolution over the life course and compare its relative degree across time and space. Apart from delving into the causes of contemporary inequality, we try to uncover the mechanisms that perpetuate its existence and analyse its wide-ranging effects on anything from occupational status and academic performance to mental well-being or family relations.

Few concepts are more foundational to sociology’s sense of disciplinary identity than that of ‘socialization’. In fact, our raison-d’etre as a distinct discipline hinges in no small part on the premise that a sizeable portion of what humans do, say and think cannot be explained by genetic hardwiring or psychological idiosyncrasies, but has its roots in the various social groupings to which they belong. This in turn implies some type of mechanism of transmission through which the ‘culture’, ‘mentalities’, ‘discourses’, ‘practices’ or ‘habitus’ that characterize these groupings end up shaping its individual members. Indeed, even the most ardent advocate of Society as a causal reality sui generis had to concede that ‘society cannot constitute itself unless it penetrates individual consciousnesses and fashions them “in its image and likeness”’ (Durkheim, 2005 [1914]: 35). The sheer diversity of terms that sociologists have coined to conceptualize this mechanism – internalization, introjection, institutionalization, conditioning, incorporation, encoding, emulation and so on – reveals the enduring lack of consensus over what socialization effectively entails. However, that it plays a more or less central role in our analyses of the social world is something that most professional sociologists would concede.

Given the centrality of ‘socialization’ to the logic of sociological explanations and given the popularity of ‘social inequality’ as a topic of sociological study, one could reasonably infer that research on how we ‘learn’ to view and judge the world around us as inherently ‘unequal’ should by now be a fairly crowded field. Curiously however, quite the opposite holds true. Symptomatic of our discipline’s more general disinterest into how and when basic socio-cognitive skills are formed (see Bergesen, 2004; Turner, 2018), contemporary scholarship on class and stratification curiously lacks a developmental perspective. This is all the more remarkable given that the ‘cultural turn in class analysis’ (Savage, 2003) has seen scholars on inequality increasingly relying on ‘cognitive’ concepts; that is, those that explicitly start from agents’ perceptions and conceptions of the social world. Behind the rise of concepts like ‘habitus’, ‘symbolic boundaries’, ‘subject positions’ or ‘meaning making’ lies the growing recognition that problems of class; that is, of the structures of inequality that define the social world, are indissolubly tied to problems of classification – that is, of how actors perceive and judge these structures of inequality. However, while much research on class readily assumes that social actors have the ability to ‘class-ify’, it rarely questions how and when this ability is actually formed.

This article argues that a better understanding of the (re)production of class inequality implies a firmer grasp on the development of the elusive ‘social sense’ (Bourdieu, 1984) that actors use to navigate social space. It does so by shifting the focus to the formative period in which this sense first starts to form, namely early childhood. To this end, this article uses a specifically designed visual method that captures the ways in which young children perceive and judge social status. It does so by asking children to combine a series of illustrated characters into a set of ‘class families’, to attribute an economic and occupational status to these families and to indicate their most and least ‘likeable’ family. This article starts by addressing the curious lack of a sociogenetic perspective on class inequality (section 2). It will then try to unpack Bourdieu’s concept of ‘social sense’ to make it more amenable to developmental analysis (section 3). Such analysis in turn requires methods that are more attuned to the particular challenges of working with very young children than the conventional techniques that are available in the sociological toolkit, as discussed in sections 4 and 5. The article then goes on to present the findings of an initial study which uses this method on a sample of 82 children, aged 5 to 12 (sections 6 to 8). The theoretical and practical implications of its findings are discussed in the concluding remarks.

Class Analysis: For ‘Adults Only’?

While contemporary sociological research on class inequality might lack a developmental perspective, this was not always the case. Interestingly enough, research on the childhood development of status perception dates back as far as the 1950s (see Estvan and Estvan, 1959; Himmelweit et al., 1952; Jahoda, 1959; Stendler, 1947). These early studies showed how a basic understanding of social class develops long before individuals assume their own position in the adult division of labour. While pre-schoolers and children in the first years of formal education still struggle with the concept of social status, the outlines of a ‘class consciousness’ appear to form roughly midway through primary school, with an important shift taking place between the ages of 8 and 10. Even though children at this age still largely lack the vocabulary of social status, they not only start identifying symbolic markers of class – as deduced from pictures of persons, houses, cars, interiors and so on – but also use these markers as a basis for evaluating others (see Estvan and Estvan, 1959; Jahoda, 1959; Stendler, 1947; Tudor, 1971). Elementary conceptions of social hierarchy also develop at this time with children showing the ability to rank jobs in terms of occupational prestige and situate themselves and others on scales of social status (Connell, 1977; Stendler, 1947). Children’s understanding of socio-economic inequalities mirrors their grasp of economic concepts. From age 8 onwards, children gradually learn to view money as a medium of exchange and also start to grasp notions of private property, wage-labour and work-relationships (Danziger, 1958). Crucially, social classes are not only increasingly perceived as features of the world at large, but also shape the way in which children judge their immediate peers. Already in primary school, children show a preference for friends from their own or a higher class position and more readily attribute positive attributes to children from high-status backgrounds, while viewing those at the bottom of the social hierarchy in a more negative light (Estvan and Estvan, 1959; Stendler, 1947). Overall, these early studies showed that our faculty for social differentiation already appears well entrenched by the time we enter adolescence.

However, despite this promising early start, this ‘developmental sociology’ of class inequality was gradually pushed to the margins of the sociological agenda. Even the repeated calls for a ‘new sociology of childhood’ (James and Prout, 2015; Matthews, 2007), which moves away from seeing children as the passive recipients of adult instruction and aims to examine their own conceptions of the social world, have yet to spark renewed interest in children’s conceptions of the structure of that world. It is indeed quite telling that most contemporary research on the development of children’s status perception has been largely taken up by developmental and cognitive psychologists (Elenbaas and Killen, 2016; Sigelman, 2012).

There are at least three reasons why a properly sociological perspective on the development of children’s ‘social sense’ deserves renewed attention. The first is historical. Indeed, to say that the lifeworld of children has undergone drastic changes since these earliest studies would be quite the understatement. As children have gradually lost their role as economic producers (Ariès, 1962), they have been discovered as potentially avid consumers that are now targeted by an industry that generates multi-billion pound revenue streams (Linn, 2004). Scholars in the field of gender and race studies have long noted how the symbolism that pervades the ever-expanding universe of children’s toys, books, television and online-interaction is far from neutral, but actively conveys sexist and racist stereotypes (King et al., 2010). However, some notable exceptions aside (Martin, 2000; Streib et al., 2017), sociologists have paid far less attention to the stereotypical representations of class that populate children’s lifeworlds. Gaining insight into the development of children’s earliest conceptions of social status can help gauge the impact of such stereotypes on their budding ‘social sense’.

A second reason is more internal to science itself and is linked to recent developments within cognitive neuroscience and developmental psychology. Renewed empirical attention to the development of cognitive abilities in children has led to a profound revision of previous theoretical assumptions. As one of the pioneers of this movement puts it: ‘These new findings led to the gradual weakening, and finally the collapse of classical Piagetian theory’ (Meltzoff, 1999: 252). Curiously, the shockwaves of that collapse have left sociologists largely unperturbed and this despite the fact that a number of findings call into question our traditional conceptions of ‘culture’, ‘agency’ and indeed ‘socialization’ itself (see Bergesen, 2004; Turner, 2018). Cognitive abilities that were previously thought to depend on lengthy learning – like ‘role-taking’, the ability to imitate others or identify intentionality – have by now been shown to exist in infants who have experienced only limited forms of social interaction (see Meltzoff, 2007). At the same time, the methodological individualism that continues to characterize much research in this field (see Lamont et al., 2017) means that few studies tackle the question of how we learn how to perceive social categories or groups. Focusing more explicitly on the sociogenesis of our faculty for class-differentiation provides an opportunity for an overdue sociological engagement with the cognitive literature.

A third reason is related to developments within the field of ‘class analysis’ itself. One of the more fundamental shifts within this field has been the growing recognition of the symbolic dimension of contemporary class dynamics. The infamous ‘cultural turn’ in class analysis has in fact expanded the traditional conception of ‘classes’ as being primarily structural features of the social world to include their reality as cognitive constructs, that fundamentally shape the way that social agents come to experience that world. As evident in approaches that define themselves as ‘phenomenological’ (Charlesworth, 2000) or ‘humanist’ (Willis, 2010) or adopt labels like ‘cultural-materialism’ (Lamont, 1992), ‘constructivist structuralism’ (Bourdieu, 1989) or ‘cultural class analysis’ (Savage, 2003), a growing body of work now recognizes the centrality of actors’ everyday acts of classification and judgement to the sociological study of class. However, if a tacit consensus seems to have emerged, on both sides of the Atlantic, that social classes are not only ‘out there’ – defining the objective structure of the social world – but also exist ‘in our heads’ – structuring our subjective experience of that world – then one of the more obvious questions this raises is: ‘How do they get there?’ How do actors acquire this elusive ‘sense of the social structure’ (Bourdieu, 1984: 474)? How do they learn ‘to define and discriminate between worthy and less worthy persons’ (Lamont, 1992: 1) or to master ‘the symbols that provide a basis for action and for understanding a material location in social context’ (Willis, 2010: 7)? While most contemporary work on class and stratification readily assumes that we simply have the cognitive ability to ‘discriminate’ between the persons, properties and practices that make up the social world, the question of how we acquire this ability is rarely the subject of systematic scrutiny.

Providing a satisfactory answer to this ‘how?’-question is of course a rather daunting affair and one that exceeds the scope of a single article. Locating the source of our faculty for classification has after all animated professional sociology ever since Durkheim and Mauss first broached the issue. However, one way of making this ‘how?’-question slightly more amenable to empirical exploration is by transforming it into a ‘when?’-question. In fact, by first establishing at which point in our cognitive development (i.e. preschool, primary school, adolescence, etc.) we start developing a faculty for social differentiation, it should become easier to identify the various pathways (family, school, etc.) through which this faculty is formed. Before discussing the methodological challenges presented by such an endeavour, it is important to first get a firmer grasp on that elusive cognitive faculty that Pierre Bourdieu (1984) referred to as the ‘sense of social structure’, the ‘sense of position’, the ‘sense of one’s place’ or, more succinctly, the ‘social sense’.

Unpacking the ‘Social Sense’

To speak of a ‘social sense’ is to start from the observation that ‘the perceived world is not an unstructured total set of equiprobable co-occurring attributes’ (Rosch, 1978: 29). Instead, the persons, practices and properties that make up this world possess, in Rosch’s (1978: 29) words, a ‘high correlational structure’ meaning that some of them are seen as having a higher probability of occurring together than others. For instance, when confronted with the following two series, most (British) readers should not only be quite capable of identifying the atypical element in each, but also of linking each series to a distinct social position:

fish and chips – tattoos – EastEnders – Warhol – ITV – football

French restaurants – BBC1 – theatre – West Wing – bingo – modern literature ²

This is what Bourdieu (1984: 241) had in mind when he referred to the social sense as an ‘immediate sense of social compatibilities and incompatibilities’, an intuitive understanding for the ‘things and people that go together’. Far from a generic capacity to ‘classify’ or grasp objects in logical relationships (of similarity, difference, exclusion, etc.), this social sense is also a capacity to ‘class-ify’, namely to simultaneously see these objects as implicated in socio-logical relationships (of equality, subordination, hierarchy, etc.). Relationships in which the classifier herself is of course also implicated and in which she hence is not only cognitively, but also emotionally invested. Far from being mere instruments of perception, social categories are also instruments of judgement and are therefore invested with much stronger moral qualities – that is, those pertaining to notions of ‘good/bad’, ‘worthy/unworthy’, ‘refined/vulgar’ and so on – than, for instance, the categories that make up zoological or botanical taxonomies (see Boltanski and Thévenot, 1983: 652).

This presents us with a picture of the ‘social sense’ as comprised of three distinct dimensions: first, the ability to recognize persons, properties and practices as systematically ‘similar’ or ‘different’; that is, as belonging to the same logical class. Second, the ability to link this (dis)similarity to an (in)equality in social status and hence to see it as expressive of a social class. Finally, the disposition to judge these dissimilarities/inequalities in explicitly moral terms. While conceptually neither of these dimensions can be treated in isolation, it is not altogether obvious that – empirically – they actually develop at the same time. Indeed, simply assuming so would risk projecting adult capacities onto children. It could well be, for instance, that children recognize that a ‘tie’ is more readily associated with a ‘business suit’ than a ‘track suit’ without therefore associating the former with ‘high’ and the latter with ‘low’ social status. Furthermore, even if a child recognizes that ‘bankers’ tend to wear ‘business suits’ and ‘cleaning ladies’ more frequently wear ‘track suits’ and that both in turn signal differences in economic status, it does not necessarily follow that they would also judge the former more (or less) favourably than the latter. Crucially, tracing the early development of this ‘social sense’ does not only imply a conceptual shift, but also a methodological one.

Reflections on Method

One of the heuristic benefits of studying children is that it forces one to think through a number of methodological assumptions that often remain unquestioned when working with adults. The challenge is not that children are somehow radically different from adults, but rather that problems one can more easily choose to ignore when working with the latter, become increasingly insurmountable as respondents get younger. Three issues in particular deserve specific attention. First, working with children complicates the use of techniques that rely strongly on written language, like questionnaires, which effectively exclude children who have not yet attained literacy. Furthermore, such discursive limitations effectively cut across the conventional divide between ‘quantitative’ and ‘qualitative’ research. Methods like interviewing are often equally problematic in that young children in particular often lack the basic language skills to verbalize and motivate their ‘class-ifications’. Indeed, the logocentrism of such methods often ignores the more subtle, embodied and affective idioms in which children convey their judgements of the social world (Nolas et al., 2019). Finally, language-driven methods also risk introducing a ‘class bias’, since there are considerable class differences in children’s capacity to verbalize their actions and ideas (Lareau, 2003; Lignier and Pagis, 2012). A second hurdle is attention span. While often taken for granted, our methods impose considerable cognitive demands on respondents who are expected to make use of a wide range of mental faculties, from rating and ranking, recollecting and introspecting, to verbalizing, narrating and motivating. If ‘respondent fatigue’ (Lavrakas, 2008) is an altogether real issue when working with adults, it becomes exponentially more problematic with children. Capturing and maintaining children’s attention therefore requires methods that appeal to their curiosity, while remaining intuitive enough to avoid emotional frustration or cognitive overload. This brings us to the final obstacle, namely the asymmetrical nature of the researcher/respondent-relation, which places children in a relationship of unequal authority with adult researchers. If ‘acquiescence’ is a well-known source of bias among adults, it becomes even more important when working with children, who are often particularly inclined to ‘please’ adults by providing the desired answer (Crandall et al., 1965). To minimize the risk that the research-relation itself skews the results, additional safeguards have to be built in to the research design. The need for a method that (1) makes minimal use of children’s discursive abilities, (2) is able to capture and sustain their attention, (3) is as non-directive as possible, while still enabling us to (4) systematically capture and compare the classificatory logic(s) behind their judgements of class-status led to the choice for a visual technique.

With the aid of a professional illustrator of children’s books we devised four ‘class families’, each composed of a mother, a father, a daughter and a son (see Figure 1). ³ Starting from a field-theoretical conception of class structure as a multidimensional ‘social space’, these families were not only designed to reflect differences along a single status-dimension, but also to distinguish between high economic and high cultural status (see Bourdieu, 1984; Vandebroeck, 2018). The former included family members (drawings 1–4) adorned with stereotypical attributes of economic wealth: a men’s business suit and tie, a women’s suit, jewellery, carefully groomed hair, pretty dresses and dress shirts. The latter (5–8) was composed of family members who represented an ‘intellectual’ or ‘bohemian’ lifestyle as expressed through items like glasses, ‘hipster’ clothing (leggings, beanie), but also subtle emblems of gender non-conformity (long hair for men, short hair for women). The third family (9–12) was designed to represent an intermediate position in social space and is marked by the emblems of what Bourdieu (1984: 246–247) would call a ‘petit bourgeois’ or ‘middle brow’-lifestyle, rooted in a ‘profession of rigour’ and a ‘eulogy of the clean, sober and neat’: button-up shirts, cardigans, closely cropped or braided hair. Finally, the ‘working-class’ family (13–16) incorporates some of the stereotyped emblems of working-class taste and lifestyle, such as the sports cap and sleeveless shirt – famously exemplified in figures like Homer Simpson or Onslow from Keeping up Appearances – or the much maligned ‘football jersey’ and ‘shell suit’ (Charlesworth, 2000: 157; Skeggs, 1997: 86).

OPEN IN VIEWER

Figure 1.

Four ‘class families’ combined by social status and by colour.

The ‘class-variants’ of each family member are all identical in terms of length, posture, skin- and hair colour, but differ in terms of clothing and cosmetics. Crucially, this visual method also enabled us to tackle the asymmetrical nature of the research-relationship. While it is impossible to change this relation itself, it is possible to stack the experimental deck in a way that makes it more difficult for the hypothesized (i.e. adult) classification to be produced. To this end, the various ‘family members’ were colour-coded in a manner that was status-incongruent. Matching the different figures on the basis of similarities in colour produces the most socially heterogeneous composition of the four families (see Figure 1). In fact, while this classificatory game is undoubtedly somewhat artificial, games of this nature are not completely foreign to children. Colours in fact tend to be one of the key organizing principles in the type of classificatory exercises that toddlers and pre-schoolers repeatedly encounter both within and outside of formal school settings (‘Fit the yellow peg in the yellow hole’; ‘Match the red circle to the red apple’). By colour-coding the figures, our study not only allows for an alternative mode of classification that is antithetical to the predicted outcome, but it also allows us to determine at which age children’s use of socio-logical principles of classification (i.e. appearance) starts taking precedence over ‘logical’ principles (i.e. colour).

The use of drawings enables us to control the type of indices children use in their ‘class-ifications’ and to ensure that these vary systematically in terms of class-status. However, it also presents some drawbacks, most notably that they constrain any conclusions on children’s ability to perceive ‘real life’-markers of social status, which would require photographs or video. Instead, drawings limit our discussion to children’s capacity to identify stereotypical representations of class, of the type they would encounter in their toys, books, games and so on. This approach is partly justified by the fact that, as Martin (2000) has argued, it is precisely the caricatural nature of these representations that facilitates the construction of children’s initial systems of social classification. This focus on the stereotypical also partly informed the choice for White, heterosexual nuclear families. Far from being rooted in an unacknowledged ethnocentric or heteronormative bias in the research design, this choice is in fact motivated by the fact that (a) the central focus on class stereotypes led us to limit any additional variation in the characters (reserved for a later stage of research) and (b) the large majority of the fictional characters that populate children’s lifeworlds are still White, heterosexual and part of traditional nuclear families (King et al., 2010).

Procedure and Sample

The classificatory exercise was presented to the children as a game in which they had to help reunite four families that had got lost at an amusement park. First, the drawings of the fathers were placed on the table and children were asked to look at these drawings carefully. The experimenter subsequently placed the drawings of the mothers on the table and asked the child to match the parents. The same was done for the sisters and the brothers. Family members were introduced in rounds to avoid cognitive ‘overload’ that could result from asking the youngest children in particular to simultaneously manipulate 16 figures. This also enabled us to more easily probe children’s motivations for the combinations they made, which they were told they could revise at any point during the exercise. After combining the families, the researcher presented the child with four money bills of varying sizes (see online Appendix A.1) which they were told represented ‘a lot’ to ‘very little’ money. The children were asked to distribute these bills between the families and motivate their choices. After this, all the cards were removed, except for the fathers. The researcher then named four occupations – ‘banker’, ‘postal clerk’, ‘teacher’ and ‘truck driver’ – and asked the child to assign these occupations to a particular father-figure. These occupations aimed to capture differences in high economic, high cultural, intermediate and low social status, respectively, while at the same time remaining intuitive enough for children to understand. However, given that the 5-year-olds in particular have only a limited occupational vocabulary – which is moreover heavily slanted towards manual occupations (Jahoda, 1959: 171) – so that even terms like ‘banker’ and ‘postal clerk’ might be unfamiliar, the researcher provided limited elaborations when children indicated difficulties with these terms (e.g. ‘works in a post office’, ‘works in a bank’). The same procedure was followed for the mothers, but here the occupations were ‘doctor’, ‘teacher’, ‘sales clerk’ and ‘cleaning lady’. Occupations were introduced after the attribution of money in order to avoid that knowledge of occupations would influence children’s perception of economic status. They hence determined the latter purely on the basis of appearance. The final stage of the experiment aimed to determine the third dimension of the social sense, namely the disposition to use status-markers in order to judge others. To this end, right after grouping the various families and before they were asked to distribute the money bills, children were asked to indicate the family they would ‘like to live with’ and most definitely did ‘not want to live with’. Throughout the exercise, the experimenter smiled and verbally encouraged the child regardless of the particular classifications they made.

The analyses presented here are the result of a pre-test of this method on a sample of 82 children from 5 to 12 who were recruited through a ‘snowball’-sampling procedure. The fieldwork itself was conducted from February to April 2018 in three cities in Belgium. ⁴ Some of the experiments were conducted in the context of ‘after school care’-programmes, others were carried out at the homes of children known, directly or indirectly, to the researcher. ‘Letters of consent’ to participate were obtained from children’s parents or legal guardians prior to each experiment. To facilitate developmental comparisons, we divided the children in three age-groups: 5 to 6 (n = 23), which in Belgium comprises children in the final year of preschool and the first year of primary; 7 to 9 (n = 29) which is the period that previous research identified as the point at which our ‘social sense’ first starts developing and, finally, 10 to 12 (n = 30), marking the end of primary school and the onset of adolescence. In addition, the experiment was also conducted with 23 adults from different age-groups (18 to 63) and class-backgrounds. Their responses provide the ‘developmental benchmark’ that serves as the comparison for children’s classifications. Despite the non-random sampling strategy, the final survey not only proved well balanced in terms of gender-composition (41 girls and 41 boys), but also in terms of the parental class of the children: 35 children came from a ‘working-class’-background (i.e. had at least one parent performing manual labour), while 47 hailed from ‘middle class’-backgrounds. Unfortunately, given the small sample, it did not prove possible to produce statistically meaningful comparisons by class (and gender) for each age-group. In what follows we therefore mainly compare children in terms of age and make only incidental reference to the role of class where possible.

A Sense of ‘Difference’

The first part of the experiment asked children to ‘re-unite’ the different families. Given the inherent difficulties in presenting (and indeed analysing) the 256 (4⁴) possible combinations that children could produce, Table 1 limits itself to showing the proportion of every age-group that matched the parents (f-m), the parents and daughter (f-m-d) and the entire family (f-m-d-s) according to the combinations presented in Figure 1 (i.e. on the basis of both status and colour). Needless to say, this places considerable constraints on the results, since we only take into account the combinations predicted by the research design and do not allow for minimal variations of these combinations. The classifications of the adults show that agreement over how the families fit together is indeed far from perfect. Interestingly, adults agreed most strongly about the ‘extremes’ of social space – in this case the ‘economically dominant’ family (1–4) and the ‘working-class’ family (13–16). The classification of the ‘high cultural status’ (5–8) and ‘middle-class’ (9–12) families provoked more disagreement, although the majority of adults matched the parents in the way proposed by the research design. As suspected, colour played little role in the classification of adults who fully grasped the socio-logical nature of the experiment.

OPEN IN VIEWER

Table 1.

Combination of figures by social status and colour.

		‘Upper-economic’			‘Upper-cultural’			‘Middle’			‘Working’
	N	f-m (%)	f-m-d (%)	f-m-d-s (%)	f-m (%)	f-m-d (%)	f-m-d-s (%)	f-m (%)	f-m-d (%)	f-m-d-s (%)	f-m (%)	f-m-d (%)	f-m-d-s (%)
Age
5–6	23	52	35	22	48	9	0	44	18	9	39	18	4
7–9	29	69	35	31	83	28	4	62	17	4	59	41	24
10–12	30	93	73	60	67	10	3	67	13	10	73	67	47
Adults	23	87	83	78	61	26	26	70	30	22	91	65	57
		Green			Yellow			Blue			Red
	N	f-m (%)	f-m-d (%)	f-m-d-s (%)	f-m (%)	f-m-d (%)	f-m-d-s (%)	f-m (%)	f-m-d (%)	f-m-d-s (%)	f-m (%)	f-m-d (%)	f-m-d-s (%)
5-6	23	13	9	4	22	9	9	26	13	9	35	22	17
7-9	29	14	10	10	10	7	4	21	14	7	10	3	3
10-12	30	0	0	0	10	0	0	7	7	0	13	0	0
Adults	23	0	0	0	17	0	0	4	0	0	0	0	0

Notes: Figures in bold indicate the strongest association for a particular age-group. f = father, m = mother, d = daughter, s = son.

This proved to be different for the children, especially the 5 to 6-year-olds, of whom 10 to 20% produced at least one family of the same colour (i.e. three members or more). At the same time, the importance of colour as a classificatory principle appears to decline quite rapidly with age. Only about 10% of the 7 to 9-year-olds and virtually none of the oldest children in our sample matched at least one family on the basis of colours. More importantly, even among the youngest, colour was by no means the dominant principle of classification and their combinations suggest that a considerable number of the 5 to 6-year-olds already understood the ‘socio-logical’ nature of the exercise. Indeed, irrespective of age, the number of children who combined the figures similar to adults – that is, socio-logically – was on average twice as large as those who combined purely on the basis of colours. Already half of the youngest children matched the parents of the ‘economic’ (1–2) and the ‘cultural upper class’ (5–6), while 40% linked both ‘working-class’ parents (13–14). In contrast, less than a quarter matched the ‘working-class’ mother to the ‘wealthy’ father or the ‘working-class’ father to the ‘wealthy’ mother, despite the fact that the latter were both colour-coded green. This budding social sense appears to become considerably more refined during the first years of primary school. Close to 80% of the 7 to 9-year-olds matched the parents of high ‘economic’ and high ‘cultural’ status, respectively, while 60% did so for the ‘middle-’ and ‘working-class’ parents. On the other hand, sociologically ‘improbable’ combinations already prove much rarer at this age, with only 10% matching the ‘working-class’ mother to the fathers of ‘high economic’ or ‘high cultural’ status or the ‘working-class’ father to the ‘rich’ and ‘cultured’ mothers. Virtually none of the children produced such a-sociological combinations. Instead, they not only matched the figures of the ‘working-class’ and the ‘rich’ parents, but also matched the daughters and sons in a highly similar manner to adults. Like the latter, children also had an easier time identifying the extremes of social space (i.e. the ‘high economic’ and ‘working-class’ families), than the figures designed to express an ‘intermediate’ social position or one with ‘high cultural’ status. Nevertheless, despite being given little to no clues as to the socio-logical nature of the exercise and even when offered an a-sociological mode of classification, a sizeable portion of the 5 to 6-year-olds already showed an ability to group the figures in a similar manner to adults. The question remains to what degree children understood their groupings as actual ‘classes’; that is, as indicative of unequal positions in a status-hierarchy.

A Sense of ‘Inequality’

We first tried to answer this question by asking children to assign money bills of various sizes to gauge their ability to rank the families on a scale of economic status. Figure 2a presents their attributions of the ‘richest’ and ‘poorest’ family. To facilitate comparisons between the different age-groups without using percentages (which are slightly deceptive given such a small sample), we present these results in the form of ‘bubble grid charts’. The radius of each circle on such a chart is proportionate to the number of children within an age-group that made that particular selection. The larger the circle, the stronger the association or, in this case, the more often the child included that particular figure as part of the ‘richest’ or ‘poorest’ family. Figure 2a shows that the 5 to 6-year-olds not only prove capable of assigning the money, but already did so in a manner that mimics the adults. Especially in their assessment of the fathers, the youngest children already seem aware that business suits point to wealth, while sporting caps and sleeveless shirts are associated with limited financial means. Interestingly, there was less consensus among the youngest children as to the economic status of the mothers with each drawing being equally likely to be included in the ‘poorest’ family. It is difficult to determine whether this is due to the fact that children more readily encounter male representations of economic status, more often see women portrayed in domestic rather than professional roles or whether our drawings were somehow less differentiating for women. What is clear is that such gender-differences in the economic classification of adults become less pronounced as children get older, while the distinctions they draw between the various class families become more outspoken. Indeed, a majority of the 7 to 9-year-olds attributed wealth to the figures of the ‘economic upper class’ (1–4), while few did so for the ‘working-class’-figures (13–16) which were, instead, disproportionately seen as having the least money. Just a few years into primary education, children already seem well aware that dresses, smart shirts and high heels signal wealth, while sneakers and track suits are more readily associated with little financial means. Finally, these associations appear firmly entrenched by ages 10 to 12, when children assign money in a manner that is virtually indistinguishable from adults.

OPEN IN VIEWER

Figures 2a and 2b.

Attribution of ‘richest’ and ‘poorest’ family members (a) and ‘most liked’ and ‘most disliked’ family.

To see whether children also have a more specific awareness of status, we asked them to assign different occupations to the figures of the fathers (‘banker’, ‘teacher’, ‘postal clerk’ and ‘lorry driver’) and the mothers (‘doctor’, ‘teacher’, ‘shop clerk’ and ‘cleaning lady’). Despite the fact that the drawings made little explicit reference to occupations (i.e. uniforms), Figure 3 shows that children proved quite capable of linking them to different jobs in a manner that closely mirrors adults. The 5 to 6-year-olds already seem to recognize that a suit is more common among bankers than truck drivers, that the latter more often sport sleeveless shirts, which are almost never worn by teachers. Again, the extremes of social space – in this case the banker and the lorry driver – are most readily identified by the children. Comparing their economic and occupational attributions also shows that children seem to grasp that different jobs are linked to different financial means. The drawings that were most readily selected as banker and lorry driver were also most often part of the ‘richest’ and ‘poorest’ family respectively. Between the ages of 7 and 9, this sense of the social extremes also seems to extend to ‘intermediate’ positions, with nearly two-thirds of this age-group identifying the ‘teacher’ and the ‘postal clerk’ in a manner that was similar to the adults. Finally, the way in which the oldest children (10 to 12) attributed jobs to the fathers is, again, virtually indistinguishable from that of the adults.

OPEN IN VIEWER

Figure 3.

Occupational attribution of parents by age-group.

The occupational attributions of the mothers again proved somewhat less straightforward. On the one hand, more than two-thirds of the youngest children agreed that track suits and sneakers are much more likely to be worn by ‘cleaning ladies’ than by ‘doctors’, ‘teachers’ or ‘shop clerks’ and this despite the fact that the drawing (14) contained little explicit reference to the job of cleaner. On the other hand, there is reason to believe that the youngest children in particular quite actively look for such occupational stereotypes. This is suggested by their classification of the mother of ‘high cultural status’ (6). While adults (and a number of the 10 to 12-year-olds) identified this figure as the most likely ‘teacher’, the youngest children in particular more often classified her as the most typical ‘doctor’. When asked why, they indicated that they treated her light yellow cardigan as a white doctor’s coat (girl [5]: ‘Because she is wearing a white coat. . .’; boy [6]: ‘Doctors wear clothes like that and stuff . . . I was at the doctor once and they were wearing white and had shoes like that . . .’). This suggests that at this young age children already display an elementary understanding of the stereotypical symbols associated with particular occupations. Nevertheless, the attributions of the 10 to 12-year-olds were in fact already more aligned with those of the adults who saw drawing 2 as the doctor and drawing 6 as the teacher. Interestingly, our analyses show little evidence that this budding sense of social differentiation is itself socially differentiated. While the youngest children of middle-class backgrounds produced occupational classifications of the fathers that were slightly more in line with those of adults, any ‘cognitive head start’ they might have, has virtually disappeared by the age of 7 to 9. Moreover, the occupational classifications of the mothers revealed no class differences in children’s classifications at any age.

The results from these first stages of the experiment already present quite a significant departure from previous studies in situating the genesis of the social sense much earlier in children’s cognitive development. The fact that a sizeable portion of the 5 to 6-year-olds already appear capable of using appearance to not only match individuals, but also to deduce an economic and occupational status, contrasts with the abovementioned claims that this sense only really starts developing between the age of 8 and 9. Even when we limit the comparison to studies using a visual methodology, our results are at odds with conclusions like ‘first grade children show little awareness of the symbols of class so far as one can judge by their ability to rate pictures of jobs, recreation, clothing and homes’ (Stendler, 1947: 92); ‘the first-grader recognizes the physical differences between the figures, but does not know the behavioural meaning of these differences’ (Tudor, 1971: 475); that ‘stereotypic cognitive and behavioural distinctions between the [. . .] class levels were not being systematically made until the children were in their second year of school’ (Mookherjee and Hogan, 1981: 97) or that ‘not a single first grade child show[s] an awareness of wealth or upper-status living’ (Estvan and Estvan, 1959: 91). Especially when one takes into account that the visual material used here was at once more limited than these studies (i.e. it only focused on dress and appearance, whereas previous research often included status objects like houses, interiors or cars), but also more demanding of children in that it asked them to identify quite specific occupations, rather than broad differences in status (i.e. ‘low–middle–high’), it is indeed quite striking that even the youngest in our sample already demonstrated the outlines of a ‘class consciousness’.

‘Discriminating’ Classes

However, even if children prove quite capable of ‘class-ifying’ the drawings on the basis of appearance, it still does not automatically follow that they also use such differences to evaluate others. To assess the latter children were asked – after constructing the various families – to indicate which family they ‘would like to live with’ and which family they ‘definitely did not want to live with’ (Figure 2b). Even though slightly more children struggled with this part of the exercise (choosing multiple families or none), the majority of the 5 to 6-year-olds expressed a preference and in a manner that is not unrelated to the perceived status of the families. For instance, the father they most readily identified as the ‘banker’ and part of the ‘wealthiest’ family also most often appeared as part of the most ‘likeable’ family. Conversely, next to none of the youngest children preferred the ‘lorry driver’ (and to a lesser extent the ‘postal clerk’) and, instead, more often chose them as the father they ‘would not want to live with’. The evaluations of the mothers again proved less straightforward, especially among the youngest. While the mothers they selected as part of the ‘richest’ family were also more likely to figure in their favourite family, the opposite did not hold true. Indeed, the figures most often selected as either the doctor, the teacher or the cleaning lady were all equally likely to be selected as part of the ‘least favourite’ family. Crucially, even the tendency to equate ‘rich’ with ‘most likeable’ seems to become less pronounced as children get older with the same number (or even slightly more) of the 10 to 12-year-olds selecting the ‘middle-class’ parents (9–10) as part of their favourite family than the ‘rich’ parents (1–2). When asked why, some children expressed stereotypical views of the ‘rich’ parents as being overly ‘strict’, ‘cold’ or ‘careerist’: ‘Cuz these parents [1–2] look like, it looks like they’re always busy with their work’ (Lea, girl, 11) or ‘I didn’t choose these ones [1–2], because sometimes when parents are like posh, then you can’t get your clothes and stuff dirty or they’re often more strict’ (Zia, girl, 11). Interestingly enough, the idea that high social status is linked to emotional severity and ‘coldness’ has also been found to animate class stereotypes among adults and the fact that we observe them among young children, suggests such stereotypical conceptions are inculcated from a very early age (Durante et al., 2017).

However, if a preference for higher status becomes less pronounced with age, quite the opposite holds true for the ‘least favourite’ family. As children get older, they increasingly seem to dislike the ‘working-class’ parents (13–14) and daughter (15) as well as the two sons (8 and 16) that they ranked as part of the poorest family (see Figure 3). Conversely, none of the figures of the ‘economic upper class’ (1–4) are selected as part of their most disliked family. It is worth reminding the reader that at this point in the experiment, children had not received any cues about the social status of the figures, nor were they in fact made aware of the socio-logical nature of the experiment (the attribution of money bills and occupations only came afterwards). Again, our results do not reveal a clear link between such judgements and children’s own social status. Working-class children proved equally (or sometimes more) inclined to select the upper-class-figures as part of their favourite families and did not show a stronger preference for lower-status-figures. In fact, out the 26 working-class children who answered this particular question, only one 5-year-old girl selected the ‘upper-class’ father and mother as part of her least favourite family. The fact that their attributions of status nevertheless clearly aligned with their judgements (compare Figure 2a and 2b), and this already among the youngest age-group, suggest that the ability to identify cues of class position develops hand-in-hand with the disposition to judge such cues and the categories we deploy to ‘discriminate’ between class positions hence appear as normative from the outset.

Conclusion

When do we first become aware that the social world we inhabit has its own distinct structure and order? When do we learn that the persons, practices and properties that constitute this world are not just ‘different’, but also ‘unequal’? The preceding pages suggest that we start doing so at a remarkably early age and, indeed, at a younger age than previous research has indicated. While the majority of existing studies situate the development of our ‘social sense’ after enrolment in formal education, our results indicate that children as young as 5 and 6 already demonstrate a rudimentary faculty for social differentiation. Even when presented with very schematic representations of class differences, that were explicitly constructed to allow for alternative modes of classification, the youngest children in our study already proved capable of matching these representations, linking them to a distinct socio-economic status and judging the various ‘class families’ in a manner that was clearly linked to such status-differentials. That they did all of this on the basis of a highly circumscribed experience of the social world is perhaps all the more remarkable. Despite its seemingly empiricist character, this article’s attempt to answer the simple question of when actors first develop a ‘social sense’ nevertheless presents some important theoretical and practical challenges.

First, it tends to complicate theoretical accounts that privilege the role of direct experience of class divisions in the construction of the perceptual schemata that constitute the ‘class habitus’. The fact that young children already demonstrate mastery of such schemata, on the basis of a limited social trajectory and, indeed, even before lengthy formal education and the development of literacy, suggests other sources than the direct experience of class inequality and cross-class interactions. Instead, it lends support to approaches that locate the source of our faculty for ‘class-ification’ in the schematic representations of social space that pervade children’s cosmologies through the toys, books, screens and games they encounter (see Martin, 2000; Streib et al., 2017). Second, the early internalization of class stereotypes like ‘track suits = lazy’ or ‘cleaning = bad’ also complicates practical strategies aimed at countering the symbolic violence that such stereotypes enable in later life. As demonstrated in the case of ethnic and sexual prejudice (Bigler and Liben, 2006), the younger the age at which children internalize racist and sexist stereotypes, the less susceptible they are to counter-stereotypical messages, since the logic of schematic processing tends to filter out or distort any information that disconfirms the original schema (‘black = beautiful’; ‘cleaners = essential workers’). The fact that even the ‘class-ifications’ of the youngest children in our study displayed such stereotypical reasoning suggests that the logic of ‘class racism’ (Bourdieu, 1984) forms no exception. If the ‘cultural turn’ in class analysis is indeed serious about asserting the crucial role that processes of perception, classification and evaluation play in the everyday (re)production of social inequality, then this requires us to get a much firmer grasp on the origins and early development of the cognitive structures that inform our routine judgements of ‘class’. By delving into the ‘primitive forms of classification’ that children use to navigate social space, this article hopes to have taken a cautious first step in that direction.

Supplemental Material