The complex reasons for learners’ underachievement in literacy in South African government primary schools are well documented (Fleisch 2008, 2012; National Education Evaluation and Development Unit 2013). However, what is new is the observation that even South African middle class children are falling behind in literacy proficiency in comparison with their peers elsewhere, and that this overall underachievement picture is unlikely to change soon (Fleisch 2012:1). South Africa's learners were assessed in 2006 and 2011 by the Progress in International Reading Literacy Studies (PIRLS), an international comparative study of learners’ performance in reading achievements at their fourth year of schooling. In 2006, the basis of the written test was reading comprehension, with approximately 215 000 learners taking part across 40 countries (45 educational systems) who were expected ‘to engage in a full repertoire of reading strategies, including retrieving and focusing on specific ideas, making simple and more complex inferences and examining and evaluating text features’ (Howie et al. 2007:13). As the performance of the South African Grade 4–5 learners was so low in comparison with similar learners from other participating education systems, the test for South Africa in 2011 was redesigned and presented as a new baseline measure. PrePIRLS is a shorter, easier test for Grade 4 learners; it has less emphasis on higher-order reading skills and is administered in all 11 official languages (Howie et al. 2012:7).

Of particular concern with the PIRLS 2006 findings was the fact that in South Africa none of the African language learners was able to reach the High International Benchmark¹ either at Grade 4 or at Grade 5 level, and that at most 17%–18% of South African learners in only two languages (Afrikaans and English) could be considered competent readers (Howie et al. 2007:28). Therefore, African language groups were not tested in PIRLS 2011, but only learners who were tested in Afrikaans or English² (also only in Grade 5) (Howie et al. 2012). Despite the easier assessment in prePIRLS 2011, only a few (6%) were able to read at an advanced level (Howie et al. 2012:112). It is significant that teachers, reportedly, spend most of their time on basic reading skills instead of on ‘more inferential types of skills’. More complex reading skills are introduced ‘at a much later stage for South African learners than internationally’ (Howie et al. 2012:112). Moreover, ‘[l]earners exposed at an earlier grade tended to achieve higher scores in reading’ (Howie et al. 2012:114). Although there are some exceptions, few teachers were found to use children's literature. Instead, textbooks, workbooks and worksheets have become very popular (Howie et al. 2012). Why is this?

The assessment criteria of PIRLS 2006 were directly aligned with national curriculum policy standards (Howie et al. 2007) and its ‘much-worse-than-predicted’ results (Howie et al. 2012:15) have influenced teacher education, national and regional policies as well as reading intervention strategies. In the struggle to ensure that the poor in South Africa enjoy quality basic education, two significant national interventions have been introduced since PIRLS 2006, which still serves as a critical external baseline for reading literacy achievement at Grades 4–5. These interventions include a new revised national curriculum, namely the Curriculum and Assessment Policy Statement (CAPS). CAPS has been introduced with standardised national workbooks for Grades 1–6 aimed at improving classroom practices in combination with the Annual National Assessments (ANAs), with a focus on learning in Grades 1–6 (Department of Basic Education 2011:5). Since then, the Department of Education claims that the ANAs 2013 and 2014 show improved results, but extreme caution is required: although monitored by the department, anecdotal evidence suggests that teachers can interfere with the process (e.g. by helping the children when doing the tests), teachers tend to teach for the test and each year new tests are designed, thus undermining the possibility of reliable, direct comparisons between schools and years (Department of Basic Education 2014).

The content and the level of the ANAs have been widely criticised (Howie et al. 2012:4), partly because of the low expectations of the children as critical readers. The authors of the PIRLS report claim that ‘[t]hrough classroom and independent reading, learners should become critical and creative thinkers’ (Howie et al. 2012:17). The action research project discussed in this article shows how an approach to teaching and learning called philosophy with children (P4C) offers unique opportunities for the teaching of critical and creative thinking in South African early literacy contexts (Murris 2014a, 2014b). This is important research because the international test results and the latest Department of Basic Education's ANAs (DBE 2011) suggest that Grade 4 learners (especially but not exclusively) in under-resourced South African schools struggle with thinking and reasoning tasks such as comprehension and problem-solving (Zimmerman 2014). Fleisch (2012) observes that teachers rely heavily on recitation and repetition with little scope for meaning-making and understanding. As part of the solution to this monumental problem of reading underachievement, he suggests that teachers should focus on the second common ‘instructional practice’ – teaching reading with graded or level reading schemes (so-called basic readers) (Fleisch 2012).

What Fleisch presents is a so-called ‘false dilemma’ (Law 2007:203). This is when an argument is presented that suggests that there are two mutually exclusive choices, when in fact there might be another possible alternative. In Britain, other possible alternatives have come to the fore. For example, in the Cambridge Primary Review – the UK's most comprehensive review of primary education for decades – a case is made for pedagogical changes to include more creativity, imagination and dialogical relationships between learners and teachers (Alexander, Hofkins & Northen 2009:24). The report emphasises classroom speaking skills to promote thinking, reasoning and learning, and states that ‘a radical re-think of … language, oracy and literacy’ (Alexander et al. 2009:24) is urgent. One approach in particular is mentioned in the context of ‘exploring, knowing, understanding and making sense’, namely philosophy with children (P4C) (Alexander et al. 2009:23). In this article we focus on a philosophy with children action research project and interpret children's philosophical responses to a picturebook by award-winning author and illustrator Anthony Browne.

Joanna Haynes (2003) argues how P4C's emphasis on philosophical responses to texts is helpful for literacy because it helps children develop a sense of power as readers. The critical approach makes space for learners to ask their own open-ended questions and to ‘challenge as well as … admire and celebrate the printed word’ (Haynes 2003:29). In P4C sessions also very young children are encouraged to interact with texts at a deep level and the ‘contemplative and learner-led nature of the approach makes an excellent counterbalance to the rapid-pace and focused learning’ (Haynes 2003:30) in skills-based approaches to literacy.

Long before the Cambridge Primary Review, Mroz, Smith and Hardman (2000) concluded that in the context of the highly prescriptive literacy approach in the UK, a focus on subject knowledge and content in the curriculum had been at the expense of good quality communication and cognitive demanding activities rendering students mere passive learners and respondents to teachers’ rapidly paced rhetorical questions (Mroz et al. 2000:387). They suggest that teachers should instead incorporate pupils’ answers into subsequent questions, a process they call ‘uptake’, that is, teachers’ ‘questions should be shaped by what immediately precedes them so that they are genuine questions’ (Mroz et al. 2000:387) – as is, for example, the case in P4C. They contrast this with recitation. This is when teachers use prepared lists with set questions that they use to check children's correct understanding.

Their solution (and P4C) is in line with the now widely accepted social constructivist theories of learning (Green & Murris 2014). Pollard (2008) provides a useful overview of the differences and similarities between behaviourist, constructivist and social constructivist theories of learning. In a social constructivist model, teachers and children work interdependently and (as in the constructivist model) ‘knowledge and skills are constructed through experience, interaction and adult support’ (Pollard 2008:182).

Stanley (2006) recommends that a typical P4C session, which usually takes at least 1 hour, should be timetabled in the schools’ curriculum on a regular weekly basis. She suggests that the first session of the day would yield the best results as this is the time of the day when the children are ‘bright and fresh’ (Stanley 2006:30). A typical P4C session includes certain sequential ‘steps’ (Figure 1).

This basic structure is a helpful guide for the novice and can vary according to the learning context and the pace of the lesson and can be adapted when following the enquiry wherever the children take it. The way in which many student teachers are taught to teach literacy in South Africa is very different from the pace and sequence of a P4C session as outlined below. Firstly, the duration of a typical story telling lesson would be no longer than 15–30 minutes. The teacher decides in advance what the theme and learning content will be. A typical structured lesson procedure would include a short introduction, presentation of the learning content by the teacher, application of the learning content by the learners and direct observation by the teacher to assess the children. In this approach, limited time and emphasis is placed on assisting young children to develop their thinking and reasoning skills.

What sets P4C apart from other oracy or literacy approaches is that children themselves are allowed to ask their own questions as starting points for the lessons. It is this feature of P4C's practice that teachers struggle with the most, and that often provokes censorship (see Haynes & Murris 2011, 2012). Ironically it is also what learners often report as one of the most stimulating aspects of the literacy lessons (Murris 2014b).

In summary, the argument so far has been that national research and international literacy tests clearly suggest that (relative to other countries) South African Grade 4s’ lack of comprehension skills is substantial. Other research suggests that learners’ cognitive framework can be developed through ‘talk, particularly where pupils are given the opportunity to assume greater control over their own learning by initiating ideas and responses which consequently promote articulate thinking’ (Mroz et al. 2000:386). Therefore, it follows logically that comprehension should be strengthened through literacy approaches that emphasise meaning-making, critical thinking and communication.

Another necessary condition, however, is to do this in contexts children find exciting and rewarding. As Luke, Dooley and Woods (2011) point out, the argument sometimes put forward that children need to learn to decode first before engaging in more complex higher-order thinking does not hold. Or, put differently, that they need to ‘learn to read’ before they can ‘read to learn’ (see e.g. Pretorius 2014). These competencies cannot be separated as cognitive tasks from ‘lived and institutionally situated social and intellectual’ practices (Luke et al. 2011:158). Following Dewey, reading and writing are regarded as pragmatic social actions that need to connect with children's own experiences (Haynes & Murris 2012). Narratives provide good contexts for children to ask their own questions and to initiate classroom activities that demand that they think and reason about texts. Carefully selected picturebooks are now popular resources for philosophical enquiries with children (ed. Costello 2011; Haynes & Murris 2012; Murris 1992; Wartenberg 2009). But using picturebooks for comprehension from a socio-cultural perspective is far from straightforward, and we report back on a research project to draw out some interesting conclusions about the challenge of ‘allowing’ learners to read their ‘multiple worlds’ (Luke et al. 2011:158).

In 2011, Vursha Ranchod started a small-scale qualitative action research project. She decided to focus on young children's philosophical responses to a few carefully selected picturebooks. ³ The major theme of her research project was to implement the pedagogy of philosophy with children (P4C) to explore, identify and understand young children's cognitive and emotional responses to a picturebook, in particular by focusing specifically on the ‘turning point’ (a sudden reversal of circumstances). Action research proved to be the most suitable methodology for her purposes as it embodies the values and principles of practitioner-based research and self-reflective practice. The crux of this method of research underpins ‘an enquiry of the self into the self’ (McNiff 2002:3). An important outcome is to show how one's own understanding of practice has shifted, by showing one's changed thinking over time. Crucial in this cyclical process is to explain how these changes are a direct result of one's investigations into one's own actions (McNiff, Lomax & Whitehead 1996:18). Therefore, keeping a reflective journal systematically was a salient research instrument. According to Ken Winograd (2003), a journal of this nature maximises the use of emotions in assisting data analysis. Winograd suggests that ‘teachers engage in both functional and dysfunctional uses of emotion in their work in schools’ (Winograd 2003:1642). He explains:

The functional uses of emotion tend to alert teachers to problems, so they can effectively take action to address those problems. The dysfunctional uses of emotion reflect situations in teachers’ emotions (especially dark emotions, like anger and disgust) that do not lead to positive action, but, instead lead to the blaming of either self, students, parents, or the system. (Winograd 2003:1642)

Ranchod became an active participant in the research whereby she was encouraged to be critical of her own work and focus on her own functional and dysfunctional emotions. Data were gathered from various sources, namely formal observation of a P4C session using a picturebook conducted by herself with her own daughters; informal observation of two P4C sessions using two picturebooks by a practicing teacher in a Grade 2 class; and informal observation of a publicly available video-recording of a P4C session in a UK primary school (http://www.gallions.newham.sch.uk). The dialogues at home were all transcribed and published (Ranchod 2012:E1–E25). The entire action research cycle was also published (Ranchod 2012:77). In this article, we report back only on Ranchod's research with one of the picturebooks, namely Anthony Browne's Little Beauty (2008).

Ranchod reports in her research project how her own two children responded to Gorilla's violent episode. Trained as a foundation phase teacher at a South African university, Ranchod had learned to use storybooks in a didactic manner. Like many teachers she would carefully prepare her lessons with a series of set questions to ensure the children would comprehend the text – with her as teacher in control of what counts as the meaning of the book. Introduced to P4C, she wanted to test the P4C pedagogy with her own two young daughters and compare her questioning skills by observing those of a fellow student in P4C in Grade 2 literacy lessons. In this class, the children wrote down the following questions and comments about Little Beauty:

‘Why didn’t the gorilla switch off the TV?’

‘The gorilla did not have friends because nobody understood him.’

‘Why didn’t the gorilla tell the trueth in the first place?’ [sic]

‘Why did they go to the toilet together?’

‘He got spoilt, but he just wanted a friend.’

‘The movies buged him and brok the TV.’ [sic]

‘Why did he break the t.v. of king kong?’ [sic]

‘How can you give a cat honey?’

‘Why did the gorilla break the zoo tv?’

‘Why wasn’t the cat afraid of the monkey?’

‘How can a cat brak a tv?’ [sic]

‘Why did the gorilla smash the tv?’

‘Maybe he was angry because he saw planes fighting a grela so that's why he broke the tv.’ [sic]

Ranchod herself experimented with what we have referred to earlier as ‘uptake’, that is, she was committed to open up a philosophical space by asking genuine questions. These are questions that are shaped by directly responding to what the children are saying – verbally as well as with their bodies. At home, after reading the story to her two daughters, the youngest asks the following question:

Riya⁶ (aged 6): ‘Can I ask why did he break the TV, because when you are angry you just want to run into your room? Can I ask what film was they watching?’

Then the following dialogue emerges: VR:

‘Yes, you can ask that, well, I can tell you what film they are watching. They are watching King Kong. There's a film called King Kong.’

The transcript shows how Ranchod has moved away from the more common IRE (initiation, response, evaluation) interaction so typical of comprehension exercises in class. ⁷ Teacher asks the questions about a text (initiation), one child responds and the adult evaluates the response (e.g. ‘well done’ or ‘have a look again in the picture’) to check children's ‘correct’ understanding of the text.

In the transcript, we can see how different the adult's role is. Both adult and children are problem-posers as well as problem-solvers. The adult gives information, but tentatively, and the questions she asks are open and inviting to enquire further: Is he indeed angry? Does the red signify anger? In this different ‘deep reading’ approach to literacy, children not only ask or answer the questions that matter, but also learn to ‘question the questions’ (Short 2011:50) and ask second-order questions, as we will see below.

Drawing on her own experiences of anger, Nikita is puzzled about Gorilla's actions and initiates the conversation by asking a question, ‘because when you are angry you just want to run into your room’. Subsequently she makes a tentative link with the movie Gorilla is watching.

Arguably, it is the artwork that strongly communicates here and Riya, who had no prior knowledge of the story of King Kong, picked up on Gorilla's emotions. Rather than condemning Gorilla's violent outburst, she ‘listens’ to what Gorilla is actually saying with his actions. Nikita in turn listens carefully to her sister and responds directly to her. She understands Gorilla's anger because King Kong is not being treated rightly (‘he was feeling very sad, that everybody wanted to kill the gorilla. So then he straight away got up and he got cross and he broke the TV’). She argues that it is morally wrong what is happening to Gorilla and empathises with him. She also sympathises with him and justifies his anger by putting the idea forward that ‘sometimes it's a lot of anger, when you have a lot of, lot of anger, you really want to break the thing and you never want to see it again’.

Contemporary philosopher Solomon (1993:132–134) argues that there is a clear link between emotions and the moral dimension of a situation. Emotions, he claims, are not only ‘evaluative judgements, but also constitutive judgements: emotions do not just find interpretations and evaluations of the world, but they construct them’. Riya clearly felt something for Gorilla and when he became angry the emotion gave her information about the moral dimension of the situation. Therefore, while she is thinking about Gorilla, she reasons that Gorilla is right and justified to be angry and as a result to break the television set. Nikita comprehends the injustice of Gorilla being blamed, thus her thinking and emotions express moral and political awareness. She clearly made a judgement: an evaluative and constructive judgement about why he is so angry.

The children are comfortably disagreeing with each other, testing and building on each other's ideas. The conversation between mother and daughters here could be regarded as a good example of how Haynes (2008) describes ‘caring thinking’. ⁸ She explains that caring thinking ‘involves caring enough to make the effort to hear what others are saying and developing the capacity to see each point of view’ (Haynes 2008:46) as well as ‘caring for the paths to truth and justice and this includes caring for the imagination as well as for logic’ (Haynes 2008:46). Here, Haynes challenges the idea that emotions are ‘just’ feelings, and would probably agree with Solomon that they can be important sources of information in the pursuit of truth, which is clearly evident in the transcript. Both adult and children are asking genuine, probing and open-ended second-order questions, as we will explain below through a further analysis of the example.

Riya challenges her mother by asking her a second-order question: how does she know about the movie? The children are pursuing the significance of the colour red in the picturebook and do not accept easy answers when Ranchod supports a deep reading of this text when encouraging them to answer why the page is red. It is a good example of responsive listening, or what Stanley calls ‘listening with the brain’⁹ – she responds to their expression of wonder about the use of this particular colour: ¹⁰ VR:

‘Why is the page red, why can’t we use another colour. Why?’

In this part of the transcript, Ranchod clearly supported her children to read the pictures in a sophisticated open-ended manner and as such did justice to the text's intricate complexity. However, the facilitating role of adults in P4C is far from straightforward, and it profoundly challenges deep-seated assumptions about what is involved in comprehension (Haynes & Murris 2011). This is evident in the following transcript:

Ranchod (2012:49–52) critically reflects on her own role as educator in the above episode. Riya clearly felt that the gorilla should not have broken the TV. She felt that a better way of controlling his anger would have been rather to switch off the TV or leave the room. Nikita, on the other hand, proposed that Gorilla was justified in breaking the TV because he was experiencing a lot of anger. When analysing the dialogue, she realised that her line of questioning closed off a learning opportunity to explore this issue further. Nikita had picked up from her mother's tone, body, facial expression, or possibly even the way she had structured her sentence, the moral message that it is wrong to break things when angry. Nikita challenges this when she gives a reason that justifies the Gorilla's anger and his violent outburst when she states, ‘because you don’t want people killing you in always … that wanting to kill you. That's not a nice thing also’. However, Ranchod steered the discussion to Riya's suggestion on how we could instead control or manage this anger. A further alternative provided by Riya and further prompting by her mother leaves Nikita with little option (considering the power differences between mother and daughter) but to change her mind and agree that controlling one's anger is better than expressing it in destructive ways. Ranchod comments that although she does not believe that destruction is necessarily an acceptable action, she does now believe that she was too quick to conclude that an alternative way of managing one's anger is suppressing or controlling it. What she had learnt from the project was that emotions can be seen as providing information about the moral dimensions of a situation (Solomon 1993). As a result, an emotion such as anger needs to be explored first, before it should be managed or controlled. She comments: ‘I could have asked the children to draw anger, to talk about what really angers them. How do they express themselves when they are really angry?’:

Instead of using this as an opportune moment to discuss, ‘What does it mean to be really angry?’, or ‘When is anger justifiable?’, I had failed to problematise the concept of anger. Through my questioning, body or facial expression I had unconsciously swayed Nikita to change her mind about breaking things when you are really angry and through persuasion she eventually provided the ‘right answer’, i.e. ‘I would switch off the TV rather’. (Ranchod 2012:51)

She also comments:

Facilitating this P4C session was not easy. This evidence was proof to me that although I yearn to be a ‘risk-taker’, my practice suggests otherwise. In other words, I was not being honest about my practice. A ‘risk-taker’ would have helped children to open up an enquiry about the meaning of this concept of anger, what it means to their lives and whether it is ever justified. I found it very challenging to keep the discussion flowing between the three of us in a meaningful way. While I tried to encourage my children to discuss how they felt about this enquiry, I found myself facilitating the discussion in the direction that favoured my own beliefs about anger and how it should be controlled or managed. (Ranchod 2012:50–51)

The research project has profoundly influenced not only how Ranchod teaches, but also the way in which she theorises about practice. She has noted a substantial shift in her thinking as a result of the cyclical and reflective nature of the research process. She found herself constantly ‘at odds’ with the non-linear character of the process and describes how she felt lost in a chaotic sense of uncertainty and insecurity as she struggled to make sense of her own emotions and those of others when she was being exposed to a very different and new pedagogy.

Winograd's (2003:1642) distinction between ‘functional’ and ‘dysfunctional’ uses of emotion mentioned earlier proved helpful. The uncertainty she had felt helped her to improve her own teaching practice and critically, but also constructively, she started to reflect on areas that needed improvement and how awareness of the power emotions can have in obstructing or making the most of a teaching and learning situation. Her daughter had empathised and sympathised with Gorilla's anger and taught her mother that at the age of eight she had understood and comprehended the injustice of the situation and had managed to communicate this in argument form to her mother and sister.

The anecdote is a good example of a ‘blurring’ of the emotion-cognition binary: emotions and cognition are often indistinguishable (Sharp 2007), and in fact are ‘constructive, moral indicators and intelligent responses to situations’ (Haynes & Murris 2012:87); ‘emotion is a kind of thought’ (Nussbaum as cited in Lipman 2003:266); emotions are judgments (Nussbaum as cited in Lipman 2003; Solomon 1993) and emotions are a ‘way of knowing’ (Zembylas 2007:297). These cognitive theories of emotion have far reaching implications for education, because instead of regarding emotions as a barrier to learning that should be managed and controlled, emotions are part and parcel of the process of truth- seeking – a truth that includes the moral and political dimensions of a situation (of which emotions are good indicators). Of course, understanding the reasonableness of Gorilla's anger does not justify his destruction of the television. But children are capable of making such distinctions and of being involved in thinking processes that explore different, more constructive responses to the emotions, and finding solutions that take Gorilla's anger more seriously.

In P4C, children are allowed to ask the questions about a story. Listening and speaking are not driven by the questions formulated by the adult in isolation (without ‘uptake’). In P4C, teachers have to listen to and respect children's own ideas about stories, initially through the questions they ask and subsequently in the discussions about their own ideas. Policies that insist on learners answering, rather than also asking questions, are responsible for the entrenched IRE practices in class (Brodie 2007), whereby teachers’ questions are mainly closed and rhetorical.

In South Africa, the National Curriculum Statement Grades R–12 ‘aims to produce learners who are able to: identify and solve problems and make decisions using critical and creative thinking’ (DBE 2011:5). Although thinking and reasoning needs to be taught in the foundation phase, the documents do not mention the place of emotions in thinking and the texts are silent about the use of picturebooks for teaching thinking and reasoning.

We conclude that this action research project shows how very young children can express deep understanding of a complex narrative when an educational environment is created that allows a philosophical and imaginative play with ideas and emotions that is taken seriously as adding to the pool of knowledge about a text. Classroom contexts are governed, more or less explicitly, by institutional power structures. The schools system and the curriculum are imbued with certain values about the creation and dissemination of knowledge. Literacy in the curriculum is framed within a context where teachers not only teach but also control the behaviour of learners. Who asks the questions in class and what kind of questions are important factors in who decides which meanings count in literacy. On the whole, the closed and rhetorical questions that educators tend to use create a learning environment in which children's astute and insightful ideas remain invisible. In contrast, when children are free to ask their own questions about a story, their profound and often imaginative representations of their understanding provoke burning questions for educators and policy-makers about comprehension; what it is, the contexts in which we teach it and how we should test it.