What counts as learning in play? Uncovering patterns in perceptions of South African early educators

ABSTRACT

Early educators’ perceptions of play and learning influence whether they adopt responsive roles to facilitate learning in play. While studies have investigated perceptions of play as pedagogical practice and its perceived relevance for learning, few have also interrogated educator learning perceptions, and hence compared these perceptions. This study explored perceptions among South African early educators by contrasting their interview and questionnaire responses. Findings revealed play perceptions encompassing structured, adult-led activities alongside child-managed play. While play was highlighted as active, joyful and chosen, participants emphasised appropriate, real-world play over children’s self-directed explorations and make-believe. Learning perceptions presented a span that primarily centred on correctness, followed by children understanding and applying lessons. Surprisingly, less nuance emerged for learning in play: rather than viewing play as a context for practicing and improving, respondents saw play as a ‘stage’ where children could demonstrate proficiency. Finally, patterns in play and learning perceptions came out more clearly in interviews than the questionnaire. These findings point to a need for alternative at-scale methods when eliciting educator perceptions, and for supporting educators to develop reflected, nuanced notions of play as practice and children’s learning in play.

KEYWORDS:

• learning through play
• play perceptions
• learning perceptions
• early educators
• mixed-methods

Introduction

Play-based practices refer to a wide repertoire of educator roles and competences to foster young children’s engagement, participation and learning in play (Zosh et al. 2018; Hedges and Cooper 2018; Whitebread 2018; Pyle, DeLuca, and Danniels 2017). For instance, early educators can adopt playful stances with toddlers to assist them during morning leave-taking (Pursi, Lipponen, and Sajaniemi 2018) or inspire young children’s interest in literacy and numeracy through demonstrating every day problem-solving (Colliver and Arguel 2018). When playful activities are combined with a responsive educator role, children’s learning of specific academic skills also improves (e.g. Vogt et al. 2018; Goble and Pianta 2017). Researchers investigating such joint forms of play propose that playful practices can be understood as a spectrum ranging from educators providing for children’s free play with minimal direction to guided forms of play and onto more direct instruction (e.g. Zosh et al. 2018). Importantly, children read subtle cues in their interaction with adults, and to perceive an activity as playful, they need an element of choice and sharing of control (McInnes et al. 2013). This entails responsivity, meaning educators tune into children’s states of being in play, and intentionally meet learning goals without disrupting children’s playful experience (e.g. Jensen et al. 2019).

In several countries, including South Africa, such playful pedagogies are mandated for early education settings; however, this mandate does not always translate to practice. For instance, in policies framing South African early education, play and play-based approaches are deemed part of quality practices, and refer to ‘ … a range of child-centred activities, such as free-play in the fantasy corner or block construction site, and teacher-directed activities such as a story “ring” or other “rings” (NCS Home Language Learning, 2012’, 20). In practice, South African early educators are found to prefer structured, predictable play and view child-led play in more recreational terms (Aronstam and Braund 2015). With their in-depth qualitative study, Shaik and Ebrahim (2015) carefully described the perceptions and professional lenses, which South African early educators used to spot learning opportunities in play contexts, and found that concerns for keeping order and discipline, and for finding evidence of learning that corresponded with the curriculum, were recurring themes. Importantly, these findings are not unique to the South African context.

Studying educator perceptions of play and learning

Across diverse cultural settings, early educator differ in whether and how they perceive children’s play as valuable for learning (Walsh and Fallon 2019; Rentzou et al. 2019; Avornyo and Baker 2018), and in turn, their different perceptions seem to influence the approach they adopt to play as pedagogical practice (Pyle and Danniels 2017; Aronstam and Braund 2015). For instance, in a Canadian study (Pyle and Danniels 2017), one group of early educators saw play as naturally promoting children’s social development and ‘foundational’ skills while academic learning required direct instruction; hence, children in their classrooms spent most of their play time without adult participation; a second group ‘ … embraced play both as its own developmentally appropriate activity and as a pedagogical tool that had the potential to support children’s learning of academic skills'. (2017, 280). This second group of educators engaged intentionally in play and guided children in ways that enriched rather than disrupted their playful experiences. From yet another context, Norwegian researchers noted similar findings: educators with nuanced notions of play as a learning context also engaged with children in play settings to intentionally scaffold and support (Karlsen and Lekhal 2019). In yet other cases, researchers find surprising inconsistencies between stated and enacted play practices (e.g. Wen, Elicker, and McMullen 2011).

In short, play and learning perceptions are both puzzling and complex to capture, especially with at-scale methods. In this regard, the present paper extends educator perception research in two ways. First, as indicated with the studies cited, early educators’ stances on play as practice, and whether play has a role for children’s learning, are rarely compared with their notions of learning. Secondly, studies tend to use either qualitative or quantitative methods (Bubikova-Moan, Hjetland, and Wollscheid 2019) rather than identify trends among larger groups of respondents, and qualify these through exploring nuances in-depth. In this paper, findings from two eliciting methods are presented and contrasted: a self-report questionnaire and responses from in-depth interviews. Following insights from these combined analyses, avenues for future development of at-scale perception measures are discussed, along with ways of promoting nuanced educator perceptions of play and learning.

Materials and methods

This paper draws on data from a larger mixed-methods study exploring changes in South African early educators’ perceptions and enactment of play-based practices. The full study followed participants of an in-service professional development (PD) programme, which focused on playful practices in reception grade classrooms (called 'Grade R'). Individual informed consent to take part in the study was collected from all educators. All respondents received an ID to ensure their anonymity, while also allowing for responses to be connected across methods. Ethical approval was obtained from the University of Kwa-Zulu Natal.

Research context

The South African Grade R is a non-compulsory programme for children aged 5 and was introduced to mitigate inequities and vast performance gaps among learners; the full-day programme uses a play-based curriculum, and focuses on literacy, numeracy and life skills (Richter and Samuels 2018). The qualification required to teach Grade R is at National Qualification Framework (NQF) level 4, corresponding to a high school degree or vocational certificate (SAQA, November 2012). Despite efforts to upskill the Grade R workforce, training levels continue to vary greatly among educators (Richter & Samuels, 2018).

Participants

Participants in the present study were early educators in the province Kwa-Zulu Natal, and based in rural and semi-urban areas where the home language is IsiZulu. The full participant group was ninety-six Grade R practitioners. All were women, and varied on the number of children they taught (12–68 children), age (23–64 years) and years of teaching experience (1–30 years). From this group, eight focal participants were purposively chosen to represent a similar diversity in class sizes (16–48 children), age (33–63 years), and experience (2–30 years): Maude, Anele, Fikile, Lisa, Lihle, Liyanda, Thembi and Martha (pseudonyms).

Data collection

Data for the analyses were collected through interviews, including video-stimulated recall (Meijer, Verloop, and Beijaard 2002), and a perception questionnaire adapted from Fisher and colleagues’ original instrument, which uses activities young children might do as items (2008). These methods were chosen for their shared degree of specificity (i.e. concrete practice examples; Limon 2006). The questionnaire was given to all participants in April 2016 (N = 96). The focal participants (N = 8) were visited in May to June and again in November 2016 to record their practice through videos and interviews. For five participants, revisits took place in March the following year to complete their interview data. At each visit, one-hour interviews were conducted using the video recordings from that day. All data were collected in IsiZulu and translated into English by PD provider staff fluent in both languages.

Questionnaire data collection

In their study on play and learning perceptions, Fisher and colleagues (2008) found three profiles among U.S. mothers of young children: 44% of respondents clearly distinguished between structured and unstructured play (Traditional); 45% rated a wide range of activities similarly play-like (All play) and 12% had highly variable ratings (Uncertain). Items in their questionnaire cover unstructured activities children might do (e.g. ‘Throwing a ball’) and more structured activities (e.g. ‘Having a book read to them’). These items were piloted (exempting play with devices, given lack of access in this setting), and additional items were generated by four local trainers, resulting in 11 unstructured and 7 structured play items. Unstructured items followed Fisher and colleagues’ definition of ‘ … requiring imaginative or creative processes, often lacking clearly delineated rules or goals … ’ while structured play items had an ‘inherent goal-oriented structure’ (2008, 309). Respondents rated how much each activity was a form of play and set a base for learning in school on a Likert-type scale ranging from 1 to 6 (e.g. 1 = this is definitely not a form of play/sets a base for learning in school to 6 = this is definitely a form of play/sets a base for learning in school).

Interview data collection

The interviews were designed to elicit participants’ perceptions of play and learning using concrete examples. The main method was video-stimulated recall (Meijer, Verloop, and Beijaard 2002) using videos recorded in their classrooms: one of children playing, and one of an adult-led activity with the educator teaching new content. First, participants’ thoughts and feelings during the play and adult-led activities were elicited; next, they described and compared roles they adopted in the two activities. The last part of the interview sought to emulate the questionnaire and explored two approaches: At the first round of visits, participants watched three clips of children playing in their classroom, gave their opinion on how much each activity was a form of play and learning, and justified their statement. However, this approach meant participants commented on diverse play activities. For all later visits, a card sorting game, which allowed them to judge similar activities, was used instead (Shirilla 2018). The full card deck shows 36 activities children might do (alone, with peers or adults), such as using hula hoops, packing a school bag or playing dressing-up. During four sorting rounds, participants placed all cards into piles: play and not play, learning and not learning. From these piles, they chose cards representing most and least play, and most and least learning, giving reasons for their choices. This approach yielded justifications as data.

Data analysis

Questionnaire analysis

Exploratory factor analysis (EFA) was used to identify underlying factor scales for the perception responses. Data distribution for responses was negatively skewed due to an overweight of high ratings, and consequently, Principal Axis Factoring was chosen (Field 2013). Factor analysis aims to reveal underlying ‘latent variables that cause the manifest variables to co-vary’ (Osborne and Costello 2009, 133). However, 6 respondents had rated whole sections using one number (i.e. rated all activities as 6 for ‘definitely play’). In order to successfully identify factors, these sections were recoded as missing data. A ratio of 10 respondents per variable was adopted; items were retained if they correlated at .3 or above with at least one other item and fell above .5 in the anti-image matrices; factor cross-loadings above .3 (Yong and Pearce 2013). Cross-loadings above .3 (Yong and Pearce 2013) and highly loading variables (de Winter, Dimitra Dodou, and Wieringa 2009) were the final exclusion criteria, leading to six items. Table 1 lists included items, and shows each item to correlate above .3 with at least one other item. Next, direct oblimin was used as the rotation method (Field 2013). Factors were retained if accounting for at least 10% of the variance with eigenvalues above 1, significant factor loadings at .35 or higher, adequate internal consistency and if they could be defined conceptually by the content of their items (Field 2013).

Table 1. Inter-item correlations for included perception items.

Perception item no.	Item 2	Item 4	Item 9	Item 12	Item 17	Item 18
2 Using building blocks or building sets 4 Using empty cardboard boxes for games 9 Using flashcards with words and pictures or with simple math concepts 12 Singing and dancing 17 Making music and rhythms with handmade instruments 18 Retelling a story in their own way or coming up with a story	– .384^a .142 .114 .210 .196	.384^a – .229 .279 .150 .169	.142 .229 – .482^a .419^a .360^a	.114 .279 .482^a – .475^a .470^a	.210 .150 .419^a .475 – .473	.196 .169 .360^a .470^a .473^a –

^aCorrelation is significant at the 0.01 level (2-tailed).

Perception profiles were identified among respondents using cluster analysis. First, four variables were generated to distinguish conceptualisations of play and learning, using mean scores for respondents’ play ratings and ratings of the academic learning value of factors found in the preceding analysis. The potential number of clusters was determined using Ward’s method, a hierarchical procedure where each participant is initially considered a separate cluster, and then systematically linked with similar participants to form larger groups (Landau and Everitt 2004). Results are represented as a dendrogram, which in this case revealed two clusters: one group showed a clear structure with few sub-branches, while a second group had a more complex structure. Next, the full sample (N = 85) was sorted into two clusters using the K-means procedure: this method strives to minimise variability within each cluster and maximise variability between clusters (Landau and Everitt 2004, 308–9). A non-parametric test was used to test the significance of cluster distinctions (Field 2013).

Interview analysis

Thematic analysis with an open approach to coding was chosen to allow for culturally specific aspects of the South African early educators’ perceptions to emerge (Braun and Clarke 2012). The unit of analysis was an interviewee response to the interviewer’s prompt. During video-stimulated recall, participants often made indirect statements about play and learning, referring to classroom events as these unfolded, what they noticed children do and how they interpreted these actions. They also relayed their own thoughts, feelings, doubts, and intentions while teaching. Occasionally, participants made note of their reactions to watching themselves on video. These responses differed from when participants compared their own roles across adult-led and play activities, or were prompted directly for their perceptions. Given this combination of practice descriptions and answers to direct questions, the approach adopted involved two steps of initial coding: sorting the interview data into six compilations, followed by reviewing and note-taking of each compilation. Initial codes for these steps were deliberately kept broad to accommodate nuances in the eight practitioners’ educational beliefs: play perceptions, activity purpose (or learning value), appraisal (of child ability/knowledge), educator role, learner role, and educator-learner relation. This first sorting round ensured that each statement was considered from several angles relevant to teaching as an enactment of beliefs. Hence, some statements featured in more than one data compilation. While reading each compilation separately and in detail, possible codes and interpretations were then noted. After this initial coding process, the next step was theme searching. Here, codes from all compilations were sorted, followed by the labelling of code groups and noting how these pointed to overarching patterns. With each iteration, potential themes grew more nuanced until most consisted of 3–5 codes. Finally, all interview excerpts were re-read and coded in light of the emerging themes. Through this final process, statements on children’s home context and physical well-being were excluded as these did not occur across the whole group of respondents.

Results

Results from the questionnaire analysis

Two factors emerged from the EFA, accounting for 62.16% of the total variance in responses (see Table 2 for item loadings by factor). The first factor had four items emphasising language and vocabulary as the learning goal (variance explained = 42.94%, eigenvalue 2.576). The second factor had two items (variance explained = 19.215%, eigenvalue 1.153), which were child-directed, hands-on activities (i.e. using building blocks or sets and using empty cardboard boxes for games). In a member-checking session with local trainers (Creswell and Miller 2000), Factor 1 was identified as typical ‘ring time’ activities (mainly adult-led, e.g. whole-group reading or a thematic discussion) with language learning as a shared theme (Cronbach’s alpha = .759), while Factor 2 were common ‘choice time’ activities (mainly child-led activities) (Cronbach’s alpha = .564). The cluster analysis was based on respondents’ ratings of Factor 1, Ring time activities, and Factor 2, Choice time activities, as play and as setting a base for learning in school, and resulted in two perception groups (see Table 3). In the first All play (Ap) group (n = 66, 69% of sample), respondents rated Choice time and Ring time activities highly as play-like and foundational to learning in school (means over 5), although this group preferred Choice time activities as a form of play over Ring time activities. This group also rated Choice time activities equally high on play and academic value, while Ring time activities were rated slightly higher on academic value than as play. A second, more Selective (Se) group (n = 19, 20% of the sample) had lower overall ratings (means under 5). At first glance, patterns for this group were similar to the first All play group: Ring time activities were valued slightly more as academic learning than play, just as Choice time activities had comparable ratings on play and academic value. But unlike the All play group, respondents in the Selective group did not rate Choice time activities higher on academic value; their ratings for both activity types were similar. Finally, results from the Wilcoxon rank sum tests showed that the two clusters differed significantly on mean scores for the four variables (cluster means are indicated after p-values in the parentheses): (1) Ring time activities rated as a form of play (p < 0.001; Ap = 5.03, Se = 4.18), (2) Academic value of Ring time activities (p < 0.001; Ap = 5.24, Se = 4.37), (3) Choice time activities as rated as a form of play (p < 0.001; Ap = 5.40, Se = 4.43), and (4) Academic value of Choice time activities (p < 0.001; Ap = 5.39, Se = 4.35). These findings confirmed two distinct perception profiles among the full group.

Table 2. Pattern matrix showing item loadings by factor.

Perception item no.	Factor 1	Factor 2
12 Singing and dancing 17 Making music and rhythms with handmade instruments 18 Retelling a story in their own way or coming up with a story 9 Using flashcards with words and pictures or with simple math concepts 2 Using building blocks or building sets 4 Using empty cardboard boxes for games	.738 .694 .643 .601	.657 .583

Extraction method: Principal Axis Factoring.

Rotation method: Oblimin with Kaiser Normalisation.

Table 3. Mean (SD) factor 1 and 2 scores and range for the All play and Selective groups.

Perception factors	All play group (Ap) n = 66		Selective group (Se) n = 19
	M (SD)	Range	M (SD)	Range
Ring time (Factor 1) Perceptions of play Academic value	5.03 (.77) 5.24 (.51)	2.71–6.00 4.45–6.00	4.18 (.59) 4.37 (.69)	2.86–5.00 3.00–5.14
Choice time (Factor 2) Perceptions of play Academic value	5.40 (.45) 5.39 (.44)	3.64–6.00 4.45–6.00	4.43 (.45) 4.35 (.42)	3.00–5.00 3.60–5.09

SD = Standard Deviation; M = Mean.

The focal participants’ perception profiles

Reviewing results for the focal participants, five were located in the All play group: Fikile (P3), Lisa (P4), Lihle (P5), Liyanda (P6), and Martha (P8). Three were in the Selective group: Maude (P1), Anele (P2), and Thembi (P7). Overall, mean scores for the focal participants and full group resonated across perception factors (see Table 4), although some distinctions seen at cluster level were less evident at the individual level. For instance, means for the full All play group indicated higher play-ratings for Choice time activities, and this was the case for two out of five focal participants in this group: Fikile and Martha. Lisa, Lihle, and Liyanda all gave similar play ratings for Ring time and Choice time activities. In line with characteristics for the Selective group, Maude, Anele and Thembi had lower ratings overall. Compared to peers in the All play group, they also assigned lower academic value to Choice time activities. However, Maude rated the academic learning value of Ring time activities lower, and Anele gave higher ratings for Choice time activities as play. Considering these results, the most distinctive characteristics of the two perception profiles were overall higher ratings for the All play group and lower ratings for the Selective group. These findings suggested a more enthusiastic and perhaps less discerning All play profile, compared to a Selective profile that might reflect uncertain or more discerning ratings.

Table 4. Overview of focal participants’ (P1-8) group membership (Ap / Se) and ratings of Factors 1 (Ring time) and 2 (Choice time).

Participant no.	P3	P4	P5	P6	P8	P1	P2	P7
Group (Ap / Se)	Ap	Ap	Ap	Ap	Ap	Se	Se	Se
Ring time (Factor 1) Perceptions of play Academic value	4.00	6.00	6.00	5.00	5.00	4.50	4.50	4.75
	5.50	5.50	6.00	5.00	5.50	3.75	4.50	4.75
Choice time (Factor 2) Perceptions of play Academic value	5.50	6.00	6.00	5.00	5.50	3.50	5.00	4.00
	5.00	4.50	6.00	5.00	5.00	4.00	4.00	4.00

Ap = All play group membership; Se = Selective group membership.

Results from the interview analysis

Themes on play perceptions

This thematic analysis included 69 interview statements that considered the nature of children’s play. Three shared themes emerged where child choice (but not self-directed or creative efforts) wove a common thread. All eight participants recognised play’s vigorous, joyful and chosen nature, while five in the group highlighted abundant toys and space as critical for good play (Maude, Anele, Lihle, Thembi, and Martha). Seven noted how children’s play preferences could reveal their aptitudes (all except Anele).

Theme one – play is vigorously active, joyful and chosen. This theme consisted of 39 statements where participants described play as vigorously active; typical forms were running, jumping, dancing and moving the whole body. For some, this strong physical aspect was coupled with an element of expressiveness through music and children talking:

He is playing guitar here, using muscles as he is playing guitar and singing. (Liyanda)

Because here, they are playing; these two of them are chasing each other, as they are chasing each other, there are things that are happening in their blood. It is not the same as child who is sitting still, who is playing with marbles. This one is exercising, there is something happening. (Maude)

I will say it is play very much because children love it; everyone wants to join in when we are dancing. (Anele)

The last statement exemplifies how enjoyment and choice were essential aspects of play, in the sense of children picking a favourite area or joining adult-led play activities they liked. Likewise, the cards chosen to represent play most was an adult and child playing with hula hoops, while the cards representing least play featured passive or distracting activities (e.g. watching television), chores (e.g. combing hair or tying shoelaces), and school-like tasks.

Theme two – good play is orderly and requires abundant toys. A second theme with 13 statements revolved around conditions for good play. Having an abundance of space and materials, and making sure that children knew how to use these, was seen as necessary:

Each and every one are enjoying playing. It looks like they are having fun and laughing, each one has their own hula hoops and they are not fighting for it with anyone. (Martha)

I felt bad because the shortage of toys causes them to end up using things that are meant for something else to do something else. Can you see, this is a head wrap to use for pretending to be a mom, but she ended using it as a blanket for the baby. (Thembi)

These statements illustrate concerns about lack of toys, space and appropriate use of materials; for instance, what could be seen as inventiveness in the second excerpt (i.e. child using a scarf to wrap a doll) was taken as a sign of shortage, leading the educator to feel ‘bad’. If children could not choose a preferred area, if they fought over toys or used them in unintended ways, practitioners attributed these disruptions to material lack, and focused on solutions such as reminders, removing a toy, or creating more toys from recycled resources.

Theme three – play choices reveal children’s aptitudes. Themes one and two pointed to children’s choice of play areas and materials; the third theme, with 17 statements, added a layer with play choices seen as revealing, not only children’s preferences, but their talents, individual characters, and even future careers:

These children, two of them, they chose themselves to go to the block area. When I came to them, I’m like, okay, this means when they [grow] old, they give me a sense that they will be builders, actually, both of them. They are building a garage to put their cars in it. Oh, it dawned on me that they will be builders, that is why they chose block area. (Martha)

You can see their talents during this time of playing in certain areas. This is where you notice that this one is likes going to block area, that’s when you see maybe s/he [is a] future builder, maybe there at the clinic, that means s/he is a future nurse. (Thembi)

Play moments, as articulated by the practitioners, often resonated with realistic scenes from home life and work, and sometimes with an undertone of gender norms: typically, girls pretended to be mothers and nurses, while boys played with blocks and cars. References to non-realistic play were rare: one practitioner noted a tower built from foam cups as ‘not real’ and another remarked on children taking on a crocodile at the local river. Participants valued play’s lively and dynamic nature; physical prowess, joy and children’s expressiveness were all valued aspects, while more imaginative sides of play were less so. These perceptions were further shaped by constraints, such as material lack, norms, and children’s predispositions. These play themes had no evident hierarchy or progression. The case was different for the focal participants’ perceptions about learning.

Themes on learning perceptions

This analysis concerned signs of learning and what children learned in play. A total of 247 statements (of which 91 referred to learning in play) comprised the four themes: learning as reproducing knowledge, learning as performing correctly, learning as understanding and applying, and learning as empowering children. These perceptions revealed a progression from children as passive to more active learners, even stretching beyond the classroom.

Theme one – learning as reproducing knowledge. With 112 statements, of which 32 referred to signs of children learning in play, this theme was prominent in the data. Here, children demonstrated learning by listening to lessons taught and giving correct answers. Common topics were naming colours, counting, saying the alphabet, days of the week, and months of the year, with some practitioners expressing concern and surprise if children gave unexpected answers:

I was surprised when children forget how many months are in the year. They said a wrong number like, how, because we always say them, every day, and today they forget. (Martha)

It is still the same, I was still there at the creative area. They were drawing. I was happy to see others drawing their families, because we were learning about ‘My Family’. (Fikile)

This emphasis on learning as reproducing content extended to play activities: In the last excerpt, a sign of learning was children overtly relating their play to a daily theme, while other examples included children demonstrating academic learning in play. The next learning theme was characterised by a more active child role in the sense of performance.

Theme two – learning as performing correctly. In this theme, 88 statements (of which 41 referred to learning in play) focused on children’s proficiency with both academic and play activities, and highlighted learning as children performing a given activity well, following the correct approach and according to instructions:

Okay, at this time, I was teaching the learners to learn to place the blocks in the colour order that I am teaching about. (Liyanda)

In this moment, it is still the same as we are continuing saying the parts of the body. I want to see if they are going to do as I am doing. (Fikile)

Learning as performance was also about knowing how to behave in class; children raising their hands was appropriate but smearing each other with paint was not. In play, learning was further recognised as children imitating adult jobs and roles correctly, such as serving tea.

I felt happy because s/he knows when s/he is here, in the fantasy area, they have to act. To do kitchen stuff since they are placed here as toys. S/he could see the cups and used them to make tea with them and serve them in the tray. (Thembi)

While these two learning themes emphasised giving the right answers and performing appropriately, the next two themes shifted towards learning as understanding and mastering skills well enough to apply them in the classroom and in children’s own lives.

Theme three – learning as understanding and applying. In this theme, 30 statements described learning as children understanding lessons taught and applying their knowledge in a practical sense (of which 18 referred to learning in play, specifically). Often, responses referred to real-life situations and to connecting ideas:

It was that they should have an open mind, it was helping them to have an open mind. Because I was asking ‘what did you build? Tell me, what did you build?’ So, the child can have an open mind, maybe, they mustn’t, if they see a car on the road, instead of seeing something that is running, they can see it’s a car and what it does. (Lisa)

Learning as making real-world connections was likewise present in statements about play; this was both in an exploratory and a practical sense, with children applying new knowledge in play or grasping how something worked through using play materials:

Okay, these two boys, as we were learning about farming, we are learning about ploughing, what they are doing here. The other one, when I was talking to him, he said he is drawing farmers and the other one told me that he is drawing things that you find in the garden, that are planted in the garden. Things that he was drawing, as he drew mealies, apples and a pear. (Lihle)

In this learning theme, the scene of children applying their understanding remained the classroom itself. For the final learning theme, practitioners’ purpose extended beyond the classroom, as they aimed to equip children to cope.

Theme four – learning as empowering children. In the interviews, concerns for dangers in the townships and communities weighed on the respondents’ minds: natural hazards, drinking or swimming in contaminated water, but also dangers at home – fire hazards and reckless requests from family members. Some concerns were inspired by the curriculum, while others came from their own, and occasionally painful, life experiences. The 17 statements in this final theme focused on equipping children to stay safe and stand up for themselves:

(…) I had in mind to explain that when an adult sends you to do something, you have to respect them, but you have to tell them that at school, they said cigarette and alcohol, we must not be sent to get those, and they must know that. (Maude)

Some practitioners took this empowering learning theme even further, with children becoming agents of change. Their expressed aim of teaching was that children could encourage their own families to make positive choices, such as planting food. Unlike for the previous themes, no specific form of learning in play emerged for the empowering theme.

Discussion

The introduction to this paper described how early educators can adopt numerous roles in play with benefits for young children’s learning and development, along with findings that practitioners differ in their perceptions of play and learning. For instance, in their systematic review of studies investigating views of play-based learning among early childhood practitioners, Bubikova-Moan and colleagues (2019) found a pattern of most educators seeing play as contributing to children’s holistic development, and laying a foundation for later learning, along with some scepticism of play’s relevance as a learning context. More importantly, the researchers noted a second layer, namely, whether educators’ stances acknowledged the need for ‘ … pedagogical flexibility and variation … ’ to meet children’s needs (Bubikova-Moan, Hjetland, and Wollscheid 2019, 10). This critical point chimes with recent findings that taking more responsive roles in play requires educators to hold nuanced notions of children’s learning in play, and of play as pedagogical practice (Pyle, Poliszczuk, and Danniels 2018; Aronstam and Braund 2015). Training and support programmes offer key opportunities to promote more reflective stances among practitioners in early education (Walsh and Fallon 2019). For this effort to be successful, however, reliable and at-scale measures are needed, which allow for groups among participants to be distinguished – both in terms of initial stances on play and learning, and how nuanced and reflected these perceptions are. Ultimately, such measures could pave the way for flexible programmes that can adapt to participating practitioners’ needs for support and guidance (Opfer and Pedder 2011).

The present study lends insights to this effort in two ways. First, the interview analysis on the South African participants’ perceptions revealed more nuanced notions of learning on its own than of learning in play. Specifically, four themes emerged on signs of learning: children reproducing content, performing activities appropriately, understanding and applying lessons, and learning as empowering children to cope. The progression of these themes mapped onto findings in the field of epistemological beliefs, for example, from notions that knowledge is certain, absolute, and imparted by an authority to viewing knowledge as tentative, evolving, and constructed by the learner (Hofer and Bendixen 2012). For children’s learning in play, however, participants in the study focused on child proficiency with curriculum content, language and physical abilities. In this sense, play became a ‘stage’ for children to demonstrate proficiency with adult- and curriculum-prescribed content, rather than an engaging context for exploring and practicing according to children's interests. Even when learning as understanding and applying was noted in relation to play, participants highlighted practical application over children’s mastery and self-directed explorations. Further to this point, realistic play was emphasised by the group. This is noteworthy in that scaffolding of children’s meaning-making and imagination form central arguments for play as a learning context (Zosh et al. 2018). In the present study, notions of ‘child choice’ rarely extended to agentic forms of participation: children’s opportunities to pursue preferred activities were underscored, but not their choices in what to do and how. Instead, participants had firm notions of how play materials should be used, including what roles a toy entailed. Together, these findings suggest an untapped potential for promoting responsive educator roles through surfacing and challenging perceptions among Grade R practitioners in South Africa.

The second way in which this study adds to the current literature pertains to methods. In their review, Bubikova-Moan and colleagues (2019) located 62 studies researching early educators’ perceptions of play-based learning; most studies employed qualitative methods, and in only eight cases, these were combined with questionnaires as an at-scale method of capturing educator perceptions. The present paper investigated patterns in perceptions of South African early educators using both a questionnaire and in-depth interviews. At first, the results resembled Fisher and colleagues’ (2008) original findings with structured, goal-oriented activities in Factor 1 and less structured activities, characterised by imaginative processes, in Factor 2. However, two perception profiles, and not three, were identified in the present study, and while these profiles differed significantly, neither made clear distinctions on degree of structure in play activities. Hence, the tenet of the original questionnaire – to identify practitioners who clearly distinguished between structured and unstructured play (Traditional) from those who rated a wide range of activities similarly play-like (All play) or gave highly variable ratings (Uncertain) – did not hold in this study. The thematic analysis of interviews offered more insights on the intricacies of the South African participants’ perceptions, especially how they interpreted learning in play. These findings emphasise the necessity for combining methods, while calling for alternative approaches to ascertain educator stances at scale, and, in particular, whether these stances are more or less reflected. Given that the present participants’ learning perceptions resembled epistemic beliefs, one avenue is to expand upon methods from this field (see Ferguson and Brownlee 2018; Madjar, Weinstock, and Kaplan 2017). Another is scenario-based methods, for instance, eliciting respondents’ interpretations in a manner framed within recurring dilemmas of play-based teaching (Bubikova-Moan, Hjetland, and Wollscheid 2019; Pyle, Poliszczuk, and Danniels 2018). By presenting educators with real-life dilemmas, researchers might gauge their classroom decision-making over socially desirable, or distal, notions of play and learning.

Conclusions

Unlike for previous studies, strong contrasts were not found for play and learning perceptions across the group of South African educators in the present study: some were hesitant in their questionnaire ratings, and others more adamant, but all recognised play as joyful, chosen and vigorously active in their video-elicited statements about play in practice. A main finding was that practitioners held a range of learning perceptions, from more passive to active learner roles, while their notions of learning in play centred on proficiency over mastery. These findings add weight to arguments that holding nuanced notions of both play and learning are necessary for early educators to discern opportunities to support children’s efforts and learning in play contexts (Hedges and Cooper 2018). Early educators may hold several beliefs about playful practices, which vary in reflectiveness: perceptions where learning is more or less assumed to happen in play (in favour), or deemed an unlikely outcome of play (not in favour); others have reflective stances on learning in play, which recognises a spectrum of playful practices and the importance of adopting a responsive educator role (Walsh and Fallon 2019; Pyle, DeLuca, and Danniels 2017). Hence, a promising route for further research and practice is to explore whether supporting early educators to reflect on children’s learning in play can lead to nuanced stances, attended by more responsive roles in play. This would require methods for eliciting degree of reflectiveness in notions of play as a learning context.

Acknowledgements

The authors warmly thank all educators and trainers for participating, and Yeshe Colliver and Valeska Grau for insightful comments. This work was supported by the LEGO Foundation.

Disclosure statement

No potential conflict of interest was reported by the authors.

Additional information

Funding

This work was supported by The LEGO Foundation.