Constructing deficit data doppelgängers: The impact of datafication on children with English as an additional language

Abstract

In this article, an evaluation of the English early childhood education context reveals children constructed as data. The complex, chaotic and unpredictable nature of the child is reconstituted in numerical form – a form which can be measured, compared and manipulated. Children are reconceptualised as data doppelgängers, ghostly apparitions which emulate the actual embodied child. The focus of early childhood education and care thus moves from child-centred to data-centred education. The author specifically focuses on the impact of this aspect of the performative regime on children who have English as an additional language – an under-researched area in the field. Foucault’s work on governmentality is used as a theoretical lens through which to understand the process of datafication. The author uses a composite child, generated from a number of children from her experience as a teacher, as a starting point for discussion. This reveals children as disadvantaged, as their home languages are no longer used to assess communication skills. Their data doppelgängers are not useful to the teacher as they are unable to demonstrate a Good Level of Development – a key measure of school readiness in English policy. The author argues that in post-Brexit-vote Britain, subtle changes to early childhood education increase disadvantage, promoting white, British culture and thus marginalising those from other cultures.

This article analyses changes in assessment and accountability policy relating to early childhood education and care in England. It uses Foucault’s (1977) work on governmentality as a tool to analyse processes of observation, normalisation and examination which, I argue, lead to the ‘datafication’ (Bradbury and Roberts-Holmes, 2017b) of early years education. This process of datafication is defined by Bradbury and Roberts-Holmes (2017b) in terms of the increase in the use of data in schools, its impact on child subjectivity and its role as a tool for governmentality. Laney (2001, cited in Bradbury and Roberts-Holmes, 2017b: 6) identifies three areas of increased use: volume, variety and velocity. Thus, the amount of data collected, the variety of its use and its processing rate have increased rapidly in recent years. This has led to a transformed educational landscape, in which data plays a pivotal role.

I take the performance measure of the Good Level of Development (GLD) as a policy field through which changes to the use of data occur. This measure identifies children who have achieved a specific range of goals at the end of their early years phase of education at the age of five, and is used to judge the effectiveness of educational institutions in England. The data is gathered gradually over the year, mainly through teacher observation, and is then converted into a summative score for each child. I also discuss the recent development of online algorithmic tracking systems to measure and analyse progress as a feature of datafication (Roberts-Holmes, 2015). Roberts-Bradbury and Roberts-Holmes (2017b) argue that assessment in the early years creates ‘data doppelgängers’ of each child. This term, originally used to describe the online selves created by users of social media, is adopted by Williamson (2014) to refer to the data other created through the use of databases. These shadowy figures are constructed through the process of converting qualitative observational data into numerical form for the purposes of national statistical analysis. The demands of accountability have created a system where this data may be more important than the actual embodied child. Thus, the aim of early years education has become the formulation and maintenance of these data doppelgängers. The doppelgängers create the ‘good data’ demanded by national governments (Millei and Gallagher, 2017: 1528). The embodied child is sacrificed for the doppelgänger as it is the doppelgänger which will be used to judge the school.

My particular focus in this article takes Roberts-Holmes and Bradbury’s arguments as a starting point, applying their theories of datafication to the experience of the child with English as an additional language (EAL). An examination of policy changes reveals this group of children to be particularly disadvantaged. These developments in policy can be viewed as emerging from the political landscape which gave rise to the Brexit vote in Britain.

I use reflections on my experience as a preschool teacher as an initial stimulus for discussion. This takes the form of a vignette – a composite child generated from a number of children I have encountered as a teacher. It focuses specifically on the reception year – the final year of preschool, which, I acknowledge, is only a part of the early years phase. In England, the reception year usually takes place in school, where as previous early years provision is delivered by a range of private and state run providers. The vignette is also influenced by anecdotal evidence from informal conversations with fellow teachers over the past three years. I use this to illustrate the process of datafication, as well as the disadvantage created by current early years assessment policy.

Vignette

In a school reception class in England, five-year-old Mariam plays alone at the art table. She moved to the UK from Pakistan last year. Prior to entering the class, she had been cared for at home by her mother. Mariam speaks Urdu at home but does not speak at all at school. She found starting school very difficult and was distressed when left by her parents. Gradually, she settled and began to watch and play.

Mariam has created a picture on the art table. She takes it to the teacher and waits near her. The teacher is working with a writing intervention group and does not notice Mariam. After a while, Mariam walks away. She had expected to work with the teaching assistant in the EAL group today, but the group has been suspended in favour of a maths intervention aimed at children who are almost at the expected level for mathematics.

During snack time, the teacher gives each child a piece of fruit. Today, Mariam feels that she is ready to try to speak English and, in a very quiet voice, says ‘thank you’ to the teacher. This is the first time she has said anything at school and the teacher is clearly very pleased. She hugs Mariam and says how happy she is that Mariam has spoken. She uses her iPad to take a photograph of Mariam. Underneath she identifies her as ‘16–24 months low’ for Communication and Language: Speaking. This grading indicates the expected level for an age band. As such, a score of ‘16–24 months low’ indicates that Mariam is functioning towards the bottom of the expected range for an average toddler. Although the teacher is genuinely pleased to see Mariam’s progress, she knows that this progress will not significantly change her data. She also knows that Mariam’s language at home is probably at the expected level but because she does not speak English, this development will not be measured.

At the end of the day, the teacher updates her online learning journals, uploading Mariam’s observation into her blog. This adds to the emerging but incomplete picture of the child. The blog contains many photographs and observations of Mariam’s non-verbal play but lacks information about her language development. After updating the journal, the teacher opens her class online tracking programme. She amends Mariam’s score for Communication and Language: Speaking from ‘8–20 months high’ to ‘16–24 months low’, highlighting ‘Copies familiar expressions’ as being achieved. Over the year, Mariam has showed very little progress in this area, having moved up only one subcategory. The teacher reflects that, for Mariam, achieving a Good Level of Development is impossible.

Mariam’s position is not unusual. The current English system requires teachers to collect vast amounts of data for each child, and many use similar online systems to record this data. Many speakers of EAL, particularly those who are new to English, are unable to demonstrate their language development and, as a result, are disadvantaged by the assessment system. In order to gain a deeper understanding of the current situation in school reception classes, it is useful to track the key changes in the English political landscape over the last 20 years which have created the conditions of possibility for its present incarnation.

Policy context

Prior to the New Labour government of 1997, early years education held a relatively low status. There was a range of provision available, some of which was within the school setting. Child-centred pedagogy was dominant, with teachers seen as facilitators of learning. Observation formed an important part of this approach, as it enabled the teacher to determine what kind of provision should be given to children (Palaiologou, 2016).

During the New Labour years of 1997 to 2010, the status of early years education completely changed. The early years sector was seen as crucial in terms of developing human capital (Moss, 2014), enabling mothers to return to work and intervening early to ensure future academic success (Pugh and Duffy, 2014). In 2000, the Department for Education published Curriculum guidance for the foundation stage. This related to children aged three to five and provided a set of norms, based on developmental psychology, for children’s education and development. Following on from this, in 2003, a standardised summative assessment for the end of the foundation stage was produced in the form of the Foundation Stage Profile (Qualifications and Curriculum Authority, 2003). In order to complete this assessment, practitioners needed to gather a wealth of observational data about each child. Knowledge of the child that had previously been largely unrecorded was converted, firstly, into writing in the form of observational notes and, secondly, into numerical data. This turning point marked the beginnings of children being constructed as data.

A change of government in 2010 led to changes in the early years foundation stage (EYFS) curriculum. In 2012, the end of early years statutory assessment, the EYFS profile was redesigned, altering the norms against which children’s progress was measured (Standards and Testing Agency, 2012). All communication, language and literacy goals were to be assessed in English rather than the home language. Although teachers were encouraged to evaluate children’s proficiency in their home language, the final judgements had to relate to the speaking of English. This meant that the communication skills of young children who were new to English were not formally assessed, and that communication in English was privileged over all other forms of communication. Prior to this, early levels of communication could be assessed in the home language (Qualifications and Curriculum Authority, 2008), although higher levels of communication, language and literacy were to be assessed in English.

Alongside these changes, a new definition of the GLD was introduced in 2013 (Department for Education, 2014b). This is a performance measure used for accountability purposes. The percentages of children reaching the GLD are submitted to the Department for Education and used to create national statistics. They are also used by the inspection body in the judgement of schools’ effectiveness. Prior to 2013, the GLD was defined as the percentage of children achieving the expected level in two areas – personal social and emotional development and communication, language and literacy – alongside an average score across six areas of learning. Thus, a child with lower levels of English could still achieve the GLD if they excelled in other areas. The new GLD is a measure of those children who achieve the expected level in five areas: personal social and emotional development; communication and language; physical development; literacy; and mathematics. This change has had a marked impact on EYFS pedagogy, as those children who have a ‘spiky profile’ – who do not perform equally well in all five areas – are now unable to meet the GLD. Supporting these children to achieve a GLD has become an increasing focus for many schools (Bradbury and Roberts-Holmes, 2017a).

A final policy development which has impacted on EYFS assessment is the introduction of a new national curriculum in 2014 (Department for Education, 2013) and its removal of the levelling system which had been in place since 1998. Government-set levels enabled teachers to evaluate children’s learning against standardised norms. The removal of the levelling system, described as ‘assessment without levels’ (Mcintosh, 2015), opened up the market for businesses to create online tracking systems for primary-aged children. Changes to the inspection process meant that progress data was increasingly being used as a primary performance indicator. As no system for tracking children in the EYFS had been provided by the Department for Education, Development matters (Early Education, 2012), a non-statutory guidance document which has since been removed from the Department for Education’s website, was used by many teachers to track children’s progress. The broad phases of development described in this document outlined expected developmental norms for an age-related phase, such as 30–50 months. Broad phases of learning were found by the creators of many online tracking systems to lack the level of detail required to track the minutiae of child progress over time, so each broad phase was broken down into smaller steps to fabricate a more specific norm (Roberts-Holmes and Bradbury, 2016).

Immigration

As part of its 2010 manifesto, the Conservative Party committed to reducing immigration and promoting integration. This has been followed by an increasing focus on developing English language skills in immigrants, with tighter regulations requiring those applying for citizenship to possess English language qualifications as well as the Life in the UK test. The current prime minister, Theresa May, has vowed to make Britain a ‘really hostile environment’ to illegal immigrants (Hill, 2017) and, since coming to office, has presided over seven immigration bills. A recent government report (Casey, 2016) into integration in the UK recommended that promoting English language must be a priority. This was followed by changes to the census, requiring all schools to collect data on English language proficiency for the first time (Department for Education, 2017a). This data was used to identify communities which are not well integrated. In addition, schools are now required to promote fundamental British values. These are defined as democracy, individual liberty, the rule of law, and mutual respect and tolerance for people with different faiths or no faith (Department for Education, 2014a). The current society is a society in which British culture, British values and the English language are privileged – a society that is increasingly hostile to immigrants.

The increase in movement of peoples across the globe in a post-9/11 world gave rise to an increase in nationalism, which could be seen in the rise of nationalist parties such as the UK Independence Party. Nationalism can be seen as a necessary aspect of neo-liberalism. Hoops et al. (2016: 730) argue that, as neo-liberalism demands open markets, nationalism is ‘deployed as glue’ to unite the nation in this global market. This, in turn, has given rise to such initiatives as Prevent (Her Majesty’s Government, 2011). This policy is part of the UK government’s counterterrorism strategy and requires those working with children and young people to report those who they feel are at risk of being ‘radicalised’ and becoming involved in extremist groups. It has been criticised by the National Union of Teachers and the Muslim Council of Britain for discriminating against Muslim children and promoting distrust (Versi, 2015). This is the political landscape which, in 2016, gave rise to the Brexit vote, in which the British public voted to leave the European Union. The public began to see immigration as a key cause of the global economic downturn.

In order to understand these changes to the political landscape and the emerging policy implications more fully, an examination of Foucault’s (1977) work on governmentality is useful. This analysis builds on the work of others, such as Ball (2013, 2017) and Mac Naughton (2005), who use Foucault as a lens to interpret current educational policy. In his work Discipline and Punish, Foucault (1977) provides a commentary on the technologies of power, which he refers to as ‘disciplines’, used to create and control citizens. He specifically discusses education in terms of disciplinary power, which he identifies in the schools of 18th-century France. In part three, ‘Discipline’, Foucault identifies three instruments of power: hierarchical observation, normalising judgement and the examination. I will examine each of these instruments in turn, relating them to current policy and practice in early years education.

Hierarchical observation

Observation has always been, and still is, central to early years pedagogy. In order to teach the child, the teacher must know the child. This knowing takes the form of observation, as the child at this age produces little ‘work’ which could be assessed (Dubiel, 2016). The young child exists within a dense network of surveillance. Foucault (1978: 98) refers to the infant thus: ‘The body of the child, under surveillance, surrounded in his cradle, his bed, or his room by an entire watch crew of parents, nurses, servants, educators, and doctors’. Surveillance is a necessary part of caring for children. It is essential to keep them safe and is central to the child–carer relationship. The child who is not observed, whose parents do not keep them under constant surveillance, could be constructed as ‘neglected’ – such is the need for observation (Palaiologou, 2016).

Observation in the preschool classroom has intensified as changes to policy have altered the function of observation. The purpose of observation has shifted from being predominantly used to inform future planning to generating a body of evidence for accountability purposes, and has thus moved from formative to performative assessment (Hatch and Grieshaber, 2002). This can be seen in the vignette, where the teacher constantly collects evidence to enter into the child’s blog. Each day, she spends time collating this evidence and using it to update her online tracking system.

The early years pupil is under constant and thorough surveillance. A body of evidence is collected about the child before they enter the class through home visits and liaison with the previous setting. A baseline assessment is usually completed within the first few weeks at school and is entered onto an online tracking system (Palaiologou, 2016). During the day, children are watched, photographed and measured in every aspect of their work and play. Detailed notes are kept on their every moment, measuring development in all seven areas of learning from use of the toilet to knowledge of letter sounds (Palaiologou, 2016). Checklists are used during adult-led sessions alongside tests for specific skills and knowledge, with pictures, writing and mark-making retained and analysed as evidence of progress (Dubiel, 2016). Thus, surveillance is ‘inscribed at the very heart of the practice of teaching’ (Foucault, 1977: 176). It is part of the technology of power, as it enables the teacher to know everything about the child.

In current practice, however, this data is not just used to inform future planning. The data double of the child, constructed through recorded observation, is used to judge the teacher and the school. Failure to produce data which compares well to the national average results in teachers being judged to be failing (Department for Education, 2016). This data is used in the judgement of the school, intensifying the significance of each observation and leading to an increase in the amount of observational data collected.

This intensification of observation is not unique to the English system. Hatch and Grieshaber (2002) explore the changing ways in which observation is used in both Australia and the USA, concluding that both have moved towards observing against standardised norms. Scott’s (2015) poignant essay film Seeing children comments on the emphasis on data collection and data management in a US preschool. Her two contrasting stories of a Korean preschool child echo Mariam’s story. Scott first presents the child as a series of statements, followed later in the film by a description of the significant contextual information missing from the data. Her work draws attention to the limitations of assessment against standardised norms to represent the real embodied child.

Normalising judgement

The purpose of the vast body of evidence collected about each child is to enable the teacher to make normalising judgements. The expected norms for the end of the EYFS are set out in the Statutory framework (Department for Education, 2017b). As there are no current expectations of development prior to this point, many teachers and the producers of online algorithmic tracking systems use the broad phases of learning established in Development matters (Early Education, 2012). By knowing how their pupils compare to the expected level, teachers can plan appropriate activities to meet each child’s needs. Mariam’s language is assessed against the broad phases of learning. She is judged as being at the lowest level expected of a child aged 16–24 months when in fact she is 60 months old.

Expectations based on Development matters (Early Education, 2012) or other tracking systems are also used for the purpose of classification. Even before children enter a setting, the information collected about them is used to rank them into those meeting the expectations and those who are not. Each child is judged against the norm and ranked by their score. The highest-ranking children may demonstrate development and knowledge of an age band above their actual age, while the lowest-ranking children, like Mariam, will demonstrate the development of a much younger child.

The use of numbers to classify is an essential aspect of neo-liberal policy. Dahlberg (2016: 125) argues that numerical data is seen as neutral and objective, thus numbers acquire ‘significant power in the public space’. Children must be classified in order for schools to be classified. This, in turn, enables schools to compete with each other, which is seen as a driver for raising educational standards (Winter, 2017).

Foucault (1977) describes the disciplinary power of normalisation as consisting of five aspects: comparison, differentiation, hierarchisation, homogenisation and exclusion. Normalisation compares the individual child to the development of the whole, thus establishing their rank within the group. The statements of the tracking system constitute the behaviour of the whole and thus become a ‘field of comparison’ (Foucault, 1977: 182). Mariam is compared to her classmates and found to be ranked the lowest in the class. Normalisation differentiates individual children, showing how they compare to the average or ‘rule’. Mariam is found to be far below the expected level for her age. It hierarchises children according to how they compare to the expected level. In this way, children are ranked according to ability and often grouped according to rank for the purposes of differentiated teaching (Frederickson and Cline, 2009). The process of homogenisation requires all children, irrespective of culture, race, gender and language, to meet the same norms. They are seen as a homogeneous group rather than diverse individuals whose development reflects their culture. Mariam’s culture and context are not taken into consideration in the assessment process. She is compared with native speakers of a language with which she is largely unfamiliar. Finally, normalisation excludes by measuring who is ‘normal’ and who is ‘abnormal’ or, in the case of the EYFS, has special educational needs. Mariam is identified as abnormal as she has such low levels of English. As her actual communication skills are not assessed, however, any special educational needs relating to language could be overlooked.

The process of normalisation constitutes Mariam’s data as useless to the teacher in two ways. The progress which she has made over the year cannot be measured as it is hidden in Mariam’s head. Her language acquisition has been developing over the year but is not yet visible to the teacher as utterances. This is in line with Krashen’s (1981) input hypothesis theory, in which he suggests that a silent period is a normal part of second language acquisition as the learner processes the new language. In addition to demonstrating no progress, Mariam cannot demonstrate attainment of the norms. With the exception of physical development, the early learning goals which constitute the GLD are dependent on language, and so cannot be met. In this way, Mariam’s doppelgänger is useless.

The norms against which Mariam and children across much of the western world are judged, like numbers, are presented as neutral and objective. They emerge from the developmental norms of psychology. Burman (2017), however, argues that such norms are anything but neutral. Rather, they emerge from an Anglocentric, white, middle-class, monolingual view of children. She notes that Anglo-US developmental psychology textbooks discuss bilingualism in the context of educational disadvantage and the failure to achieve in English language programmes. Bilingual children are expected to meet the norms of monolingual children from the moment they enter school and are constructed as failures when they are unable to do this.

In order to compare, differentiate, hierarchise, homogenise and exclude, a vast body of evidence is needed. Thus, the child is reconstituted in data. The data other, which has been created for each child in the form of this corpus of data, is scrutinised, compared and polished to create the ideal version of the child needed for accountability purposes (Roberts-Holmes and Bradbury, 2016). Hierarchisation is crucial as it identifies those who may possibly reach the GLD. These children then receive more attention. This process is referred to by Gillborn and Youdell (2000, in Bradbury and Roberts-Holmes 2017b) as ‘educational triage’, where children are divided into three groups: those who will meet the expected level, borderline children and the ‘no-hopers’. The teacher then focuses on the borderline children in an attempt to increase the percentage of children meeting the measure. More data is collected for the borderline children, and interventions are planned to try to make them ‘good’. This can be seen in the intervention classes delivered to the borderline mathematics group in the vignette. This phenomenon is not unique to the UK but has also been documented in other countries such as the USA, where high-stakes tests are implemented. Lauen and Gaddis (2016: 127), for example, report finding that for lower-ability children, the focus on ‘bubble kids’, or those who are close to meeting the expected level, led to an increase in the gap between the low-achieving and high-achieving students.

The normalisation process can be seen as an aspect of the neo-liberal system, which is explored by Moss (2014). He argues that the story of quality and high returns dominates education. The purpose of education in the neo-liberal world is to develop human capital. In order to measure the returns of the investment in this human capital, norms must be set against which children are measured. The focus of education thus becomes determining whether a child is ‘normal’ or not, rather than what they can actually do. This can be seen in many developed countries, where measures of ‘school readiness’ are assessed at the end of the early childhood education and care phase (Organisation for Economic Co-operation and Development, 2017). This normalising function, Moss (2014) argues, determines the type of citizens produced by the state, transforming not only the systems for creating human capital, but also the subjectivities of individuals.

The examination

Foucault (1977) describes the examination as the combination of normalisation and surveillance. It is the means by which children are known. They are observed and classified through the process of examination. Examination makes each child into a case to be studied, an object to be measured and investigated, a scientific specimen. It also links knowledge with power as the teacher increases her knowledge of the child through the process of examination and, as a result, alters the power relationship. Examination is a process which makes the invisible visible. Everything about the child must be known, and examination is the process through which this visibility is established. In terms of the EYFS, examination reflects both the medical examination in terms of the close scrutiny of each pupil and the educational examination in terms of assessing the knowledge of children.

Children are examined on a daily basis to identify what they know. Evidence is collected from observations, photographs, video recordings, conversations with parents, data from previous settings, tasks, activities and a whole array of surveillance techniques (Palaiologou, 2016). It is then used to reveal the soul of the child, making their thought processes and development visible to the teacher. It is referred to in schools as ‘assessment’, the process of ‘knowing and understanding children’ (Dubiel, 2016: 8).

This daily form of examination is then used to demonstrate progress, the key tool used by the Office for Standards in Education, Children’s Services and Skills (Ofsted) to measure the performance of a school. The purpose of this body is to inspect and regulate services that care for children and young people. In order to do this, data is collected and analysed by the school and presented for inspection. Qualitative observational data is converted into numerical data in order to demonstrate progress over time. Many schools use online tracking systems to do this (Dubiel, 2016). Algorithms are used to identify how many children meet the expected level and how much progress has been made (School Pupil Tracker, 2017).

The examination can also be seen in the early years in terms of summative assessment. In England, this takes the form of the EYFS profile (Standards and Testing Agency, 2016). This examination can be described as a high-stakes test as the results of the test are used to judge the performance of the school (Hutchings, 2015). The data from the EYFS profile is mined by local authorities and the Department for Education, and forms part of the data set given to Ofsted to support its judgements of school effectiveness (Department for Education, 2016).

Research into high-stakes testing indicates that it does indeed raise test results (Hanushek and Raymond, 2005). However, the negative impact of high-stakes testing on children and teachers is also well documented. Knowledge is found to be restricted to the knowledge specifically required by the test, and teaching is found to be narrowed, focusing only on the test (Amrein and Berliner, 2002; Koretz, 2008, cited in Hutchings, 2015). Hutchings (2015:11), in her report for the National Union of Teachers, lists the negative effects of high-stakes testing on pupils and teachers as being ‘less creative teaching; a narrowing of the curriculum; a focus on borderline students at the expense of others; pupil anxiety and stress; and temptation to both pupils and teachers to “game the system”’.

Foucault (1977: 189) describes examination as a process which ‘introduces individuality into a field of documentation’. The surveillance and normalising judgements of the examination must be recorded, thus placing individuals in a ‘network of writing’. For each EYFS child, like Mariam, there exists a vast body of written and numerical data. This mass of documents acts to ‘capture and fix’ the child (Foucault, 1977: 189). The chaotic, unpredictable nature of the child is converted into fixed data, which can then be measured and compared. It captures the child as they are then fixed in time, unable to change the version of themselves formulated in the data. The use of online data analysis systems intensifies this process as it produces a ‘machine-readable’ (Williamson, 2014) child. The huge quantity of qualitative data recorded in blogs and profiles is converted into numbers and inserted into the system. This datafication constitutes the child as an individual, a ‘describable, analysable object’ (Foucault, 1977: 190). The purpose of this is to individualise the child. It pins down and fixes the individual, making them predictable and manageable – what Foucault (1977: 193) calls ‘the calculable man’. Thus, individualisation creates a version of the child determined by the examination and created through data – a data doppelgänger.

This data double follows the child through school. Although Mariam may well have been labelled without the data, the increased focus on data strengthens the impact of this labelling process. Before Mariam enters Year 1, her EYFS profile scores, her learning journal, her end-of-year report and the online tracking data will all be used to inform the new teacher. The teacher may well use this information to rank Mariam and put her into a differentiated group. Expectations will probably be low, as the doppelgänger is not the ‘right’ kind. In this way, Mariam’s doppelgänger precedes her into the next class and determines the teacher’s expectations. A distorted doppelgänger showing very low language development will create distorted predictions of progress. Thus, expectations of Mariam’s development will be low, possibly leading to a lack of challenge and underperformance in her future education.

The reconstruction of children in data is not new to the English EYFS, but has become more instrumental in shaping pedagogy and practice following changes to the GLD measure. The current GLD requires children to demonstrate attainment in a narrower range of areas and to meet more challenging targets. This makes it harder to achieve in two ways. The first is that children must achieve equally across five areas; the second is that they must achieve a higher level than previously. This was reflected in a statistical analysis of the EYFS profile score results, which showed a drop from 64% achieving a GLD in 2012 to 52% in 2013, when the new measure was introduced (Ofsted, 2014). The early learning goals for literacy and mathematics became significantly more challenging, with children now being expected to independently write simple sentences and solve mathematical problems using doubling, halving and sharing. The change in the GLD also privileges certain subjects such as literacy and mathematics, but marginalises the subjects which are not measured, such as expressive arts and design and understanding the world.

Research into datafication

As the GLD is a performance indicator, the creation of the ‘right’ kind of data – the ‘right’ kind of doppelgänger – has become increasingly important to preschool teachers in schools. In pursuit of these useful doppelgängers, research reveals teachers staging intervention groups to try to get borderline children to the GLD, particularly in the summer term. These interventions are found to use more direct teaching and a less play-based approach, with an increased focus on English and mathematics (Roberts-Holmes, 2015). This increase in direct teaching is reflected in a recent report from Ofsted (2017) which recommended that in order to prevent children from ‘falling behind their peers’ (4), schools should ‘devote sufficient time each day to the direct teaching of reading, writing and mathematics’ (7). Play, it suggested, was not the best way to teach certain subjects. This move towards direct teaching reflects a view of the child as a recipient of knowledge rather than a constructor of knowledge. It prioritises direct teaching, as it is seen as more efficient, quicker and resulting in short-term performance. In the global race for educational success, constructing knowledge is too slow. It is geared towards ensuring that, in neo-liberal terms, a high return is guaranteed for the investment in human capital (Moss, 2014).

Research into child development suggests that in order to develop effective social and communication skills, children need to play. Play enables children to collaboratively create meaning, which leads to a ‘deep understanding of self and others’ (Brock et al., 2014: 290). This kind of ‘deep understanding’ cannot be transmitted from the adult to the child in a direct-teaching approach. Jarvis (2018) goes further, drawing on evidence from neuroscience which indicates that children under the age of seven have incomplete neural pathways. Thus, they are unable to process new information which does not relate to existing concepts in the child’s memory. Through play, children are able to build up knowledge at their own pace, relating it to their existing understanding. Direct teaching, on the other hand, adds new knowledge which may not fit with existing knowledge. Jarvis (2018) uses a useful analogy of a wardrobe with insufficient hangers. If a child does not have the right hangers on which to build knowledge in the brain, the clothes will fall to the floor of the wardrobe in a jumbled mess. The increase in the use of direct teaching as an approach to support children who have not met the GLD can therefore lead to poor-quality learning experiences and can deny children the learning through play which they need in order to develop appropriately.

Within the constructs of the importance of play, however, are notions of the ‘normal’ child with ‘normal’ development. Olusoga explores these culturally defined notions of ‘normal’ play, analysing a teacher’s statement that ‘a lot of children come to school unable to play’ (Brock et al., 2014: 39). Clearly, there is a norm of play in the mind of the teacher which is not demonstrated by the children from ‘deprived backgrounds’ who do not speak English in her class. The teacher’s concept of normal play is not neutral, but culturally biased, based on classed, raced and gendered ideas about what play should be. Grieshaber and McArdle (2010) also problematise the notion of play, questioning whether play is normal, fun and natural, or leads to learning. They contend that a culturally mediated version of play is promoted in settings. Childhood practices manufacture a version of the ‘natural child’ participating in ‘natural play’, and any deviations from this norm are regarded with suspicion. Thus, even the EAL child’s play is measured against Anglo-American, white, middle-class, monolingual norms (Burman, 2017).

In addition to changes in pedagogy, researchers found evidence of teachers fabricating doppelgängers, which reflected their own prejudices as moderation procedures demanded that the data must ‘look right’. Bradbury (2011) found that teachers were changing the data to reflect their expectations, as well as making professional judgements that were highly influenced by deficit discourses about groups such as EAL children. There is considerable evidence of teachers ‘gaming the system’ (Hutchings, 2015: 36), creating the doppelgängers which will portray them in a good light and result in the school being ranked as ‘good’, rather than attempting to support the child’s progress through the assessment system (Hutchings, 2015; Roberts-Holmes, 2015).

EAL children are particularly disadvantaged by the current system. Their communication is not measured and therefore has no value, as only communication in English counts. When a national baseline assessment for the start of reception was trialled in 2015, language development was again assessed only in English. Bradbury and Roberts-Holmes (2016) identified disadvantage for EAL children as a key concern, as knowledge of these children’s development could not be assessed accurately. Their deficit doppelgängers are more shadowy than their English-speaking counterparts’, as an incomplete data double is created. As EAL children struggle to adapt to a new language and sometimes a new culture and place, their needs are neglected in favour of those whose doppelgängers will be more useful to the school. This can contribute to increasing marginalisation of children from immigrant families.

The important role of what Foucault (1977) calls ‘exclusion’ or identifying those children with special educational needs is also hindered by the creation of data doppelgängers. Changes to the assessment of language have meant that only English language development is measured. Language development may not be measured at all in non-English speakers like Mariam, which could lead to children in need of specialist language support not being identified. This is a well-documented concern which is explored in depth in Cline and Shamsi’s (2000) literature review. Children with special educational needs at all levels of education were found to be overlooked at times as their language issues were misinterpreted as a result of their bilingualism rather than a specific learning difficulty. Cline and Shamsi also found that the opposite could occur, with children categorised as having special educational needs when, in fact, they did not display any evidence of specific learning difficulties. An example of this, explored by Bligh and Drury (2015), is a child mistakenly diagnosed as being a selective mute when, in fact, they were displaying the expected behaviour of a child in the silent period of second language acquisition.

Conclusion

The demands of accountability have created a system where the data doppelgänger may be more important than the actual embodied child. The emphasis on pupil progress, where data is the key indicator used to judge the child, teacher and school, creates a landscape in which anything that cannot be measured is irrelevant. Thus, the aim of the EYFS has become the formulation and maintenance of these data doppelgängers. The embodied child is sacrificed for the doppelgänger as it is the doppelgänger which will be used to judge the school. The impact of the creation of data doppelgängers remains with the child throughout their education. Before they even enter the following year group, the data doppelgänger is already there, constructing the child in the mind of the teacher.

The creation of data doppelgängers has privileged certain children, while the doppelgängers of others are deficit doubles. The system disadvantages children with EAL as they are classified as ‘no-hopers’ and therefore receive less attention than the children who are almost at the GLD. Their progress is hidden as their early stages of second language acquisition prevent it from being measured. Their development is measured against culturally unfamiliar norms which actively discriminate against young bilinguals. Their data is incomplete as their language development is not assessed, which can lead to special educational needs being missed or mistakenly identified.

The datafication of early years education has the capacity to lead to increased knowledge of the child and therefore more accurate planning for progress. The use of data to judge the status of the school, however, has repositioned assessment as a control mechanism to produce the ‘right’ kind of child and the ‘right’ kind of teacher. With such high stakes, assessment becomes more important than teaching, and data becomes more important than children. Changes to the early childhood education system have led to children with EAL being particularly disadvantaged, positioning them at the margins of the class and limiting their capacity to learn and develop. This may be symptomatic of an increasingly nationalistic society in which the English language is used as a tool to exclude and marginalise those from other cultures.