Abstract
This study adds to the limited research on play-based therapy programs in special school settings for children with an IQ of less than 70.
In occupational therapy, play is viewed as an important and meaningful occupation for children; even so, occupational therapists face barriers implementing play in schools (Burton & Kuhaneck, 2023). In early childhood education, play is understood to promote a child’s development, learning, and well-being (Larsen et al., 2023). Evidence strongly positions play, particularly pretend play, in early literacy learning (Roskos & Christie, 2013). When engaging in pretend play, children create a shared meaning with others to sustain the play (Whitebread & O’Sullivan, 2012), create stories in the play (Nicolopoulou et al., 2009), and use counterfactual reasoning (Francis & Gibson, 2023). Research on play-based learning has supported a continuum of play from free play to teacher-directed learning, with guided play being more effective for learning than didactic teaching or free play (Zosh et al., 2018). In guided play, a teacher sets up a play situation and the children direct the play (Zosh et al., 2018). Children’s self-initiation of play brings deeper learning and increases quality of thinking (Whitebread & O’Sullivan, 2012).
In Australia, special schools are schools for children with IQs < 70 that use occupational therapists. Many children who attend these schools are neurodivergent and display challenges to self-initiate pretend play, which affects their childhood occupations of player, student, and learner (Phelan & Kinsella, 2014). Research on play in special schools is lagging even though pretend play interventions for neurodivergent young people have been found to be effective in promoting their development (Kasari et al., 2006; Sualy et al., 2011). In Sualy et al.’s (2011) study, 5 of the 6 children with language delay who were in the play intervention strengthened their play skills, whereas 5 children in the comparison group remained at pretest levels. Kasari et al. (2006) conducted a randomized controlled trial with 58 preschoolers with autism in which the children were allocated to one of three groups (joint attention, symbolic play, or control) for up to 6 wk. Children in the joint attention or symbolic play group made language gains, with the joint attention group making more gains that the symbolic play group up to 12 mo postintervention (Kasari et al., 2008). Children in the symbolic play group showed more diverse pretend play ability in parent–child interaction and increased in play ability on post play assessment (Kasari et al., 2006).
O’Connor and Stagnitti (2011) used a group comparison design to investigate a play intervention based on Learn to Play Therapy (Stagnitti, 2021) within a special school setting in Australia. Nineteen children were in the play intervention and 16 children participated in traditional nonplay-based classroom activities. The aim of the study was to investigate differences between the groups in play, language, and social skills of children ages 5 to 8 yr pre- and postparticipation in the play intervention, which ran for 22 wk. O’Connor and Stagnitti (2011) found that children in the play intervention group significantly improved in social interaction skills (47% increase) and social connection (36% increase) as well as language skills and object substitution, which increased by 27% from baseline. Children in the intervention group became more socially connected with their peers, whereas children in the comparison nonplay group displayed more play-disruptive and socially disconnected behaviors (O’Connor & Stagnitti, 2011; Stagnitti et al., 2012).
Learn to Play in schools is adapted from Learn to Play Therapy (Stagnitti, 2021). Learn to Play Therapy is a child-centered therapy in which the therapist, who has observed and assessed the child’s play and understands the child’s play abilities, plays beside the child. The aim of Learn to Play Therapy is to increase the quality and joy of a child’s ability to intentionally self-initiate pretend play (Stagnitti, 2021). Learn to Play is most used by occupational therapists in supporting a child’s participation in play and supporting connection with peers. It is often delivered by occupational therapists working in private practice, hospitals, and community organizations with therapists working in the home, childcare, and nursery school.
For children with low play ability, therapy begins with the therapist engaging the child’s interest and attention in joyful interaction and fun with a toy or play materials. Through joyful interaction with the child, the therapist may then shift to introduce a pretend play activity, such as feeding a teddy bear or having a character dance to music. The choice of play is based on the therapist’s understanding of the child’s play abilities and interests at the time. The aim of Learn to Play Therapy is to strengthen a child’s play capacity so that the child takes over the session, choosing play activities that are meaningful and intentional for the child and supporting the child as an autonomous player. Therapy is at the pace of the child, and parents are encouraged to be involved in sessions as they deepen their understanding of play and of how their child plays. The therapist’s knowledge of the increasing complexity of pretend play and underlying skills, such as sequences of play actions, object substitution, character play, role play, attributing properties, referring to absent objects, and stories in the play, inform the therapist’s observations when working with the child. Specific techniques may be implemented to support the child’s play, such as repeating with variation, modeling play, interacting with the child through a puppet or character, following the child’s lead, or introducing a problem in the play (Stagnitti, 2021). Learn to Play Therapy training includes knowledge of play assessment, pretend play development, importance of pretend play for child development, creating a sense of safety, and building a relationship with the child that is child centered (Stagnitti, 2021). Supervision hours and logged practice hours support the therapist to integrate their knowledge with practice so that the therapist can respond to and support the child in the moment during play.
O’Connor and Stagnitti (2011) recommended further research on Learn to Play Therapy within special school settings, in particular, longer interventions (more than 22 wk) and with larger sample sizes. The current study aimed to build on previous findings by analyzing the effectiveness of Learn to Play Therapy within a variety of special schools. Because all children and schools were different, this broader study would allow a wider generalization of results.
We carried out a program evaluation of Learn to Play Therapy within three special schools and one specialist school in Australia (specialist schools are for children with IQs < 50), including interviewing school staff pre and post implementation of a 7-mo Learn to Play Therapy block (Wadley & Stagnitti, 2020a, 2020b). The study reported in this article was one aspect of this program evaluation.
Aim of the Study
This article reports the results of the program evaluation based on Gervais’s (1998) program evaluation model, which require measurement of outcomes of the program being evaluated. For Learn to Play Therapy, the expected results were children’s play ability and related areas of language, social and emotional skills, and learning (Francis & Gibson, 2023; Nicolopoulou et al., 2009; Roskos & Christie, 2013; Whitebread & O’Sullivan, 2012; Zosh et al., 2018).
The study had three aims. Aim 1was to establish whether there were any changes in children’s development (pretend play, language, social, and emotional skills) and learning (academic competence) in their first year of special or specialist schooling after participating in Learn to Play Therapy. Aim 2 was to determine the size of the effect of Learn to Play Therapy on a child’s development. Aim 3 was to determine whether pretend play predicts a child’s language, social, emotional, and academic skills. In addition, we also explored whether there were any gender differences in play, language, social, and emotional skills. Differences between genders in language, social, and emotional skills were found in a 4-year longitudinal Swedish study, with girls scoring higher than boys (Nordberg et al., 1991). Lucisano et al. (2021) and Stagnitti (2022) also found differences between genders in pretend play among children ages 3 to 5 yr in Brazil and Australia, respectively. All these studies were conducted with neurotypical children, and there is little research on gender differences among neurodivergent children. We made the following hypotheses: Hypothesis 1: Children will improve in their pretend play, language, social, and emotional skills and academic competence over the 7-mo period. Hypothesis 2: Learn to Play Therapy will have a medium effect on the children’s pretend play, language, social, emotional, and academic skills over the 7-mo period. A medium effect was hypothesized on the basis of research findings by O’Connor and Stagnitti (2011) evaluating Learn to Play in one special school, as well as findings by Kasari et al. (2006, 2008). A small effect was ruled out because these studies found significant positive findings, and a large effect was unable to be hypothesized because research was limited within school settings. Hypothesis 3: Pretend play will predict language, social, emotional, and academic skills.
Method
The current research, which was part of a larger program evaluation, was a pre–post single-cohort design using multiple assessments.
Participants
Children
Thirty-eight children (15 girls and 23 boys) from four special or specialist schools across Victoria, Australia, participated in this study. Children were in their first year of school and were between 5 and 7 yr of age. The mean age of children at baseline data collection was 5 yr 7 mo (SD = 0.46 yr). All children participating in the study had a diagnosis. Of the 38 children, the main diagnoses included intellectual disability (29 children), autism (14 children), and global developmental delay (7 children); many children had dual diagnoses. Parental written consent was given before any child was included in the study, and child assent was sought for the pre- and postassessments.
Teachers and Therapists
Across the four schools, 7 foundation teachers were approached to be in the study. Foundation teachers are teachers of children in their first year of formal schooling. All 7 (5 female and 2 male) foundation teachers consented to participate. All the teachers taught the children and had known the children since they began school, which was 3 mo before the study began. All teachers had been employed at the special or specialist school for more than 12 mo. Four therapists (2 speech pathologists and 2 occupational therapists), who were employed in the schools, were approached to be in this study because they had been working with the children in their schools with Learn to Play Therapy. All 4 therapists consented to be in the study. It is important to note that two of the four schools did not have therapists involved in the Learn to Play Therapy program within their school; therefore, these schools used only the teaching staff in this study. The role of the 4 therapists was to support the teachers in running the Learn to Play program, supporting their engagement and understanding of pretend play in childhood.
Instruments
Child-Initiated Pretend Play Assessment
The Child-Initiated Pretend Play Assessment (ChIPPA; Stagnitti, 2022) assesses the quality of a child’s ability to self-initiate pretend play for children ages 3 to 7 yr. It is a norm-referenced standardized assessment, with a norm sample of 693 children (Stagnitti, 2022).
The ChIPPA is administered in a one-on-one assessment in which the therapist sets up a play space (a sheet thrown over two adult-sized chairs) and the child and therapist sit on the floor in front of the space. The ChIPPA takes 30 min to administer with 4- to 7-yr-old children and 18 min for 3-yr-old children. For 4- to 7-yr-old children, there are two 15-min sessions during which the child is presented with conventional–imaginative toys for 15 min (conventional–imaginative session) and unstructured play materials for 15 min (symbolic play session; Stagnitti, 2022). Each 15 min is divided into three 5-min segments. In the first 5 min, the therapist introduces the toys and invites the child to engage with the play materials. In the middle 5 min, the therapist introduces a second figurine and models (without disturbing the child’s play) up to five play actions, then the therapist is passive again for the final 5-min segment. Throughout the assessment, the therapist interacts with the child in a way to make the child at ease, without instructing or directing the child in any way. The assessment can be scored as the child plays; if needed, the assessment may finish early, depending on the child.
The ChIPPA measures three items. The percentage of elaborate play actions (PEPA; elaborate play) is the percentage of logical sequential actions to total actions and includes attributions of properties and references to absent objects. The number of object substitutions (NOS) is the number of toys or objects to which the child imposes a meaning (e.g., “the box is a cave”). The number of imitated actions (NIA) is the number of times a child imitates the examiner. Further aspects of play and enjoyment of play are scored on the clinical observations form. The ChIPPA has established interrater reliability (intraclass correlation coefficient) of 0.87 to 0.96 (Francis & Gibson, 2023) and stable test–retest reliability (Stagnitti & Unsworth, 2004). The ChIPPA discriminates between preschool children with acquired brain injury and typically developing preschoolers (Thorne et al., 2021).
Clinical Evaluation of Language Fundamentals–Fourth Edition, Australian
The Clinical Evaluation of Language Fundamentals–Fourth Edition, Australian (CELF–4) is an individually administered standardized assessment tool used for the identification, diagnosis, and outcome measurement of language and communication disorders among students ages 5 to 21 yr (Semel et al., 2003). In this study, Level 1 was chosen because it was a measure of general language skills (Semel et al., 2003). This level assesses identifying the language problem and core language and includes concepts and following directions, word structure, recalling sentences, and formulated sentence tasks (Semel et al., 2003).
The CELF–4 is sensitive to language difficulties among children who have a developmental disability, autism, and hearing impairment (Semel et al., 2003). The validity and reliability of the CELF–4 is poorly documented in the literature. Semel et al. (2003) discussed the evidence base on special group studies and the clinical validation for children with disabilities. The diagnostic sensitivities within this group was high, but ongoing research is critical in the applicability of the CELF–4 for clinical groups (Semel et al., 2003).
Expression, Reception, and Recall of Narrative Instrument
The Expression, Reception, and Recall of Narrative Instrument (ERRNI) provides a measure of a person’s expressive language skills and story comprehension skills (Bishop, 2004). The ERRNI is used in clinical settings to evaluate the language skills of children with language delay (Bishop, 2004). Children are invited to look at one of two textless stories the Beach Story or the Fish Story—and view the pictures in the story before telling the story to the therapist. The assessment takes 15 min, and the children’s stories are audio-recorded for scoring. Four scores are calculated: the initial storytelling, recall of the story, comprehension of questions about the story, and mean length of utterances (MLU; Myers & Botting, 2008).
To evaluate reliability, we measured internal consistency using the coefficient α for the ideas and comprehension scores, and we used Pearson correlation to measure the relationship between storytelling, recall, and MLU. Acceptable internal consistency was reached for the ERRNI indices (Bishop, 2004). For concurrent validity, the ERRNI was compared with other comprehension and grammar measures, including the CELF–Preschool (Semel et al., 2003). The strongest correlations were found between the comprehension items, but all other correlations were weak. This provided evidence that the language skills measured on the ERRNI were independent of those measured on alternative language measures (Bishop, 2004).
Social Skills Improvement System Rating Scales
The pen-and-paper teacher form of the Social Skills Improvement System (SSIS) Rating Scales is a reliable standardized assessment that evaluates social skills, problem behaviors, and academic competence. The SSIS analyzes common social skills across seven subdomains: including Communication, Cooperation, Assertion, Empathy, Responsibility, Engagement, and Self-Control (Gresham et al., 2011). In addition, the SISS assesses five problem behaviors: externalizing, bullying, hyperactivity, inattention and internalizing. The teacher form also assesses academic competence in math, cognitive skills, reading, and motivation (Gresham et al., 2011). Each item on the SSIS is rated on a 4-point scale (0 = never, 1 = seldom, 2 = often, 3 = almost always). Scoring is based on the teacher’s interpretation of the frequency of the social skill (Gresham et al., 2011). The SSIS shows strong test–retest reliability and internal consistency. Gresham et al. (2011) stated that “median scale reliabilities of the Social Skills and Problem Behaviour Scales are in the mid- to upper .90s for every age group on each form” (p. 37). Test–retest reliability is good, with .82 for total social skills on the teacher form, .83 for problem behaviors, and .92 for academic competence (Gresham et al., 2011).
Strengths and Difficulties Questionnaire
The Strengths and Difficulties Questionnaire (SDQ) is a behavioral screening questionnaire for children and adolescents ages 4 to 16 yr (Goodman, 1997) that assesses their psychosocial attributes to evaluate whether these attributes affect their behavior (Kersten et al., 2016). The teachers filled out the SDQ in this study. There are 25 psychological attributes divided into five scales measuring emotional symptoms, conduct problems, hyperactivity–inattention, peer problems, and prosocial behavior.
In a systematic review completed by Kersten et al. (2016), the five-scale structural validity of the SDQ was noted as strong. The systematic review included published studies reporting on the validity and reliability of the SDQ parent and teacher questionnaires for preschool-age children (3 to 5 yr; Kersten et al., 2016). Overall, the review confirmed strong convergent validity and internal consistency for the SDQ (Kersten et al., 2016). The authors of the systematic review addressed the lack of evidence for other psychometric properties including test–retest reliability, cultural validity, and criterion validity, which need to be evaluated in future studies because of the SDQ’s widespread clinical use (Kersten et al., 2016).
Procedure
We obtained ethical approval from Deakin University’s Human Research Ethics Committee and the Department of Education and Training before the study began. The schools that were participating in the larger program evaluation study were invited to participate in this study on child outcomes. Four schools consented to participate. The four schools were in varied locations in Victoria, Australia. Two schools were in metropolitan areas, one school was in a regional area, and one school was in a rural area. The schools varied in size, resource availability, staffing, and management policies. These schools sent out information packs, including consent forms, to parents of children in their first year of school. Parents placed the signed consent forms in a reply-paid envelope in a box in the reception area of the school. Teachers returned their consent form in a reply-paid envelope to Chloe Wadley. Child assent was secured if the child verbally agreed to play and engaged with the toys, or if they nonverbally showed interest in the toys and did not show stressed behaviors such as crying or wanting to go back to their classroom.
Baseline or Predata Collection
Baseline assessments on all 38 child participants were completed over 2 mo by Wadley, who was trained to deliver the assessments. Children were directly assessed with three instruments: ChIPPA (Stagnitti, 2022), CELF–4 (Semel et al., 2003), and ERRNI (Bishop, 2004). The three assessments were randomized using a Latin square design to account for a child’s test order fatigue. If a child was away on the day of assessment, Wadley returned on an alternative day to assess the child. The assessments took between 45 min and 1 hr to complete for each child.
Social skills and academic competence (SSIS) and emotional regulation (SDQ) data were completed by teachers for the children who were in their class. The forms took approximately 45 min to complete for each child.
Play Intervention: Learn to Play Therapy
Children participated in the play intervention, which was Learn to Play Therapy that had been adapted for small groups in a school setting. Children participated for a minimum of 1 hr each week over the 7-mo period within school hours (Term 1 to Term 4 of the school year). Teachers, therapists, and integration aides ran the play intervention based on the principles of the Learn to Play Therapy (Stagnitti, 2021). In addition, 2 occupational therapists and 2 speech therapists supported the teachers in the schools where they were employed.
Before the study, Wadley trained the teachers and therapists on Learn to Play Therapy for 2 days. The training involved educating teachers and therapists on how to assess and recognize a child’s abilities in pretend play, explaining the principles and techniques in engaging children in play, and providing practical support in setting up Learn to Play in a school setting in relation to resources, toys, and space. School staff were trained on the following 10 key play skills and how to integrate them developmentally into their play sessions: ▪ using play sequences ▪ describing and explaining the play ▪ substituting objects ▪ recognizing and describing decentration (using a character in play, such as a teddy bear) ▪ recognizing and facilitating play scripts ▪ joining the child in role play ▪ talking about the play (attributing properties to objects) ▪ referring to absent objects and actions ▪ adding problems in the play narrative ▪ predicting what will happen next
The occupational therapists and speech therapists in the study understood the children’s developmental play levels and supported the teachers by choosing play materials that were suitable for a child’s developmental play level and offering strategies to support Learn to Play key skills (e.g., tracking the play, repeating with variation, modeling). For the weekly play intervention sessions, teachers and therapists set up a play space within the school, such as a room that was not being used. The play activities chosen were based on the children’s play abilities and interests. Although there was flexibility in how schools implemented the play intervention, many schools set up play stations for 3 to 4 children with staff at each station who played beside children and coplayed, facilitated, and followed the children’s lead in play. An example of a play station would be a shop scene in which one child was the shop keeper and two other children would come to the shop to buy items. The teacher or therapist would join in the play supporting the children.
For fidelity, Wadley kept in contact with the schools via email and telephone to check in with the teachers or therapists about any difficulties or challenges they were having, the types of play activities the school was implementing, how the staff were engaging the children in the play, and whether the guidelines (see below) were being adhered to. Teachers and therapists were advised that they could contact Wadley at any time if they had questions. The following guidelines for program fidelity were provided to each school: ▪ Learn to Play Therapy must run for a minimum of 1 hr each week. ▪ Play activities are to be chosen by the school, with reference to the Learn to Play manual. ▪ Play activities are the right challenge for the children. ▪ Teachers and therapists use the 10 key play skills for Learn to Play in each session. ▪ Teachers and therapists coplay with the children. ▪ Teachers and therapists provide repetition while playing beside children to assist children to understand the play. ▪ Schools determine how many children and how many play activities are in each session. ▪ Schools choose the physical space for the play intervention. ▪ Learn to Play runs from Term 2 to Term 4.
Follow-Up or Postdata Collection
Follow-up data collection occurred at the end of the school year, approximately 7 mo after the initial baseline assessment. The same procedure for data collection at baseline took place at follow-up for each child. Karen Stagnitti, an experienced occupational therapist, assisted with the postdata collection and was familiar and trained in the administration of the assessment tools. She had no background knowledge on the children. For the follow-up, 32 children were assessed on the ChIPPA and ERRNI, 30 children completed the CELF–4 (lower number because of absenteeism), and 34 SSIS questionnaires and 33 SDQs were returned.
Data Analysis
We analyzed data using IBM SPSS Statistics (Version 22). Because all children in the sample had IQ < 70, the data did not meet the assumptions of normality; therefore, nonparametric statistics were used. The specific analysis methods for each aim are listed below.
Aim 1
We used a Wilcoxon signed-rank test and descriptive statistics to compare the baseline and follow-up results on each of the five assessment measures (ChIPPA, CELF–4, ERRNI, SSIS, and SDQ). The Wilcoxon test is used when there are multiple sets of data and the scores come from the same participants at two different points in time.
Aim 2
To ascertain the impact of Learn to Play (i.e., effect size), we calculated Cohen’s d for small sample sizes based on paired samples. Cohen’s d uses the square root of the average of each measurement (baseline and follow-up) to estimate the effect (Peng & Chen, 2014). Cohen’s d can be interpreted as having a small (d = 0.2), medium (d = 0.5), or large (d = 0.8) effect (Cohen, 2013).
Aim 3
We used a generalized estimation equation (GEE; Liang & Zegler, 1986) to investigate whether play predicted changes in children’s language, social, emotional, and academic skills. We chose this statistical approach because it is used to fit a model for repeated data analysis (Wang, 2014), which in this study included baseline and follow-up over a 7-mo period. The GEE is suitable for samples under 100 and can account for covariates (Wang, 2014) such as time and age.
Further Analysis
We completed a Mann–Whitney U test to determine any gender differences in play, language, social, emotional, and academic skills at follow-up. We chose this test because it allows two groups or more to be compared without the assumption that the values are normally distributed (Stangroom, 2015).
Results
Aim 1 and Aim 2 Results: Changes in Children’s Scores and Effect Size
The results were positive in relation to the changes in child development over the 7-mo period. Tables 1 and 2 present the descriptive statistics, p values, and Cohen’s d for all children assessed at baseline and follow-up for ChIPPA, CELF–4, ERRNI, SISS-RS, and SDQ. Table 1 presents the play and language data. Table 2 presents the social, academic competence, and emotional data. There was a significant difference between the baseline and follow-up scores in elaborate play (PEPA) combined (p = .03) and object substitution (NOS) conventional (p = .047). Increases in the raw scores for all PEPA scores, NOS combined, typical play indicators, and total time playing from baseline to follow-up were noted. Play deficit indictors decreased.
Baseline and Follow-Up Descriptive Statistics and Cohen’s d for Raw Scores of the ChIPPA (PEPA, NOS, and NIA), CELF-4, and ERRNI
Note. ChIPPA = Child-Initiated Pretend Play Assessment; PEPA = percentage of elaborate play actions; NOS = number of object substitutions; NIA = number of imitations; typical play indicators = number of typical play indicators recorded on clinical observations; play deficits indicators = number of play deficits recorded on clinical observations; total time playing = number of seconds the child played within 30-min duration of ChIPPA; CELF-4 = Clinical Evaluation of Language Fundamentals, Fourth Edition; ERRNI = Expression, Reception, and Recall of Narrative Instrument; MLU = mean language utterance. The p values are based on Wilcoxon signed ranks test with matched data.
Includes the conventional and symbolic sessions.
p < .05.
Baseline and Follow-Up Descriptive Statistics and Cohen’s d for Baseline and Follow-Up SSIS (N = 34) and SDQ (N = 33)
Note. SSIS = Social Skills Improvement System; SDQ = Strengths and Difficulties Questionnaire. The p values are based on Wilcoxon signed rank test with matched data.
p < .05.
p < .01.
p < .001.
There was a significant increase between the baseline and follow-up scores in recalling sentences (CELF–4; p = .02). There were large effect sizes for narrative (ERRNI MLU and ideas), whereas forgetting scores decreased. All other raw scores, with the exception of word structure, increased from baseline to follow-up.
For the SSIS social skills, there was a significant increase between the baseline and follow-up scores in social skills with Assertion (p = .001), Responsibility (p = .16), Empathy (p = .002), and total social skills score (p = .022). Raw scores also increased in Communication, Cooperation, Engagement, and Self-Control. There were significant increases between baseline and follow-up in SSIS problem behaviors, including externalizing (p = .036), hyperactivity–inattention (p = .004), internalizing (p = .002), and problem behaviors total score (p = .007). There was a significant increase in scores between baseline and follow-up in SSIS academic competence (p = .012).
There were no significant results between the baseline and follow-up scores on the SDQ. However, raw scores decreased for hyperactivity–inattention, peer problems, and total SDQ. Scores increased for prosocial behavior and emotional symptoms.
Effect sizes were small for SDQ emotional symptom. There were medium effect sizes for SSIS social skills (Assertion) and academic competence. There were also medium effect sizes for problem behaviors total and hyperactivity–inattention, where raw scores increased.
Hypothesis 1 was supported. Children significantly improved in two key play skills: elaborate play (PEPA combined) and object substitution with conventional toys (NOS conventional). NIA scores decreased, with lower scores indicating less reliance on the examiner for play ideas. NIA was interpreted against PEPA and NOS, which were cognitive play skills. As NIA decreased and PEPA and NOS increased, children increased in their ability to self-initiate their play. Typical indicators increased, whereas deficit indicators decreased. Children significantly improved in their expressive language (recalling sentences on the CELF–4). Narrative language raw scores increased from baseline to follow-up. Children improved in their social skills and academic competence. Children did not improve in their emotional regulation as measured by the SDQ; however, scores for prosocial behavior increased, and there was a decrease in hyperactivity, peer problems, and total difficulties scores.
Hypothesis 2 was partially supported, with medium effect sizes found for SSIS Assertion (d = −0.54), ERRNI forgetting (d = 0.59), and SSIS academic competence (d = 0.50). Large effect sizes were found for ERRNI MLU in words (d = 2.72) followed by ERRNI ideas (d = 1.26).
Aim 3 Results
We conducted a GEE with the variables that showed a significant change from baseline to follow-up in Aim 1. Elaborate play and object substitution were the independent variables used at baseline to determine whether pretend play influenced certain areas of development over time. Table 3 presents the GEE where elaborate play (PEPA combined raw score) was the independent variable. Table 4 presents the GEE where object substitution (NOS) was the independent variable.
Generalized Estimation Equation: PEPA Combined (Pre) Prediction on ERRNI MLU, SSIS Total Raw Score, Academic Competence, and Recalling Sentences (Post)
Note. PEPA = percentage of elaborate play actions of the Child-Initiated Pretend Play Assessment; PEPA combined = both the conventional–imaginative play session and symbolic play session at baseline; ERRNI = Expression, Reception, and Recall of Narrative Instrument; MLU = mean language utterance; SSIS = Social Skills Improvement System.
Indicates age in years (continuous data).
p < .05.
p < .01.
p < .001.
Generalized Estimation Equation: NOS Combined Raw Score (Pre) Prediction on Recalling Sentences, ERRNI MLU, SSIS Total Score, and Academic Competence (Post)
Note. NOS = number of object substitutions; NOS combined = both the conventional and symbolic session at baseline; ERRNI = Expression, Reception, and Recall of Narrative Instrument; MLU = mean language utterance; SSIS = Social Skills Improvement System.
Indicates age in years (continuous data).
p < .05.
p < .01.
p < .001.
Hypothesis 3 was supported for pretend play predicting language, social, and academic skills but not emotional regulation. Object substitution (NOS) was highly significant in predicting expressive language (p < .001), narrative (ERRNI MLU; p =. 015), social skills (SSIS total score; p < .001), and academic competence (p < .001). Elaborate play (PEPA combined) and time were predictive of social skills (total SISS score; p < .001). Elaborate play (PEPA combined) was predictive of academic competence (p < .001).
Further Exploration
We used a Mann–Whitney U test to determine whether there were any gender differences in play, language, social, emotional, and academic skills. There was a significant difference between girls (M = 91.5, SD = 60.5) and boys (M = 37.9, SD = 44.1; p = .025) for the post-PEPA combined raw score, with girls having higher scores in their elaborate play skills compared with boys. There were no significant differences between girls and boys in language, social, emotional, and academic skills.
Discussion
The purpose of this study was to determine whether Learn to Play Therapy implemented in a special or specialist school setting for neurodivergent children affected the children’s development and learning (including play, language, social, emotional, and academic competence). Over the 7 mo of the play intervention, children did improve in their self-initiated pretend play, as evidenced by increases in elaborate play and object substitution ability and a decrease in imitated actions. Typical indicators of play increased, and play deficits decreased. The play intervention directly targeted these abilities. Because there was no control group, changes cannot be attributed directly to the play intervention; therefore, we focus this discussion on the influence of elaborate play and object substitution on other skills.
Elaborate play indicates the complexity of a child’s ability to sequence play ideas and impose meaning on the play through property attributions and absent objects (Stagnitti, 2022). In this study, elaborate play ability at baseline with time was found to be influential to social skills total score at follow-up. This suggests children needed time to develop their elaborate play and social skills, which took a number of sessions of Learn to Play before changes in play and social ability occurred for children with multiple diagnoses (Stagnitti, 2021). Lin et al. (2017) found that elaborate play was associated with increases in the quality of theory of mind for children with autism, which implied increases in the quality of social interactions.
Elaborate play and object substitution on the ChIPPA are related to counterfactual reasoning (Francis & Gibson, 2023). Counterfactual reasoning is cognitive processing. Learn to Play focuses on pretend play, which requires a child to use their cognitive skills to imagine and think of ideas. In our study, children’s ideas for the textless stories in the narrative language assessment showed a large effect. Children developed the ability to impose meaning on objects and engage in various play scripts (stories in the play). The changes in this study were reflected through the larger study results, where therapist and teacher observations noted that “their [the children] imagination skills just exploded” (Wadley & Stagnitti, 2020a, p. 330). Francis and Gibson (2023) argued that there was a plausible role of imaginative representation capacity to explain similarities in cognitive mechanisms of thinking such as pretend play and counterfactual reasoning. As staff were scaffolding pretend play in this study through coplaying, children had the opportunity to develop their cognitive skills through problem-solving and using their imagination to include absent objects and actions. This was also reflected in the larger study’s results through teacher and therapist observations, with children listening, following instructions, and sharing more (Wadley & Stagnitti, 2020b).
Object substitution reflects a child’s ability to step outside reality and use symbols in play (Stagnitti, 2021). Object substitution had a greater influence than elaborateness of play in this study (see Tables 3 and 4). NOS was highly influential of social skills, expressive language, academic competence, and narrative skills. In other research, object substitution ability in preschool had been found to predict narrative ability in the same children 4 yr later (Stagnitti & Lewis, 2015) and predictive of expressive and receptive language in children in their first year of school (Stagnitti et al., 2020), which corresponds with the findings from this study. In a meta-analysis of pretend play and language across 35 studies (N = 6,848), the association of language and pretend play was robust (Quinn et al., 2018). Within the larger study, teachers and therapists also noted changes in “language, communication and turn-taking” (Wadley & Stagnitti, 2020a, p. 330) among the children after participating in Learn to Play and observed changes in object substitution, with one teacher noting that children were using more symbols in play (“using a box as a frypan”; Wadley & Stagnitti, 2020a, p. 330).
Social–Emotional Changes
The social–emotional changes post play intervention were mixed. On the SISS, frequency scores and teachers’ scores reflected a significant increase in externalizing, hyperactivity–inattention, internalizing, and total problem behaviors. On the SDQ, there were no significant differences between pre and post play intervention. Raw scores for emotional symptoms and conduct problems increased in contrast to positive changes found for prosocial behavior and a decrease in peer problems. Changes in the latter two areas are consistent with social and play changes noted earlier. Total difficulties score on the SDQ decreased after the play intervention.
The negative findings on the SISS (increases in externalizing, hyperactivity–inattention, internalizing, and total problem behaviors) and SDQ (increase in emotional symptoms and conduct problems) could be because the SISS and SDQ were rated by teachers at the end of the school year. This was the first full year of formal schooling for the children. School children are usually tired by the end of their school year and may exhibit behaviors such as defiance and emotional outbursts (We Are Teachers, n.d.). Hammarberg and Hagekull (2006) found that in a sample of 370 children age 6 yr over a school year, boys were more prone to negative changes in problem behaviors than girls.
Gender Differences in Elaborate Play
Girls had significantly higher elaborate play combined scores at follow-up when compared with boys in this study. There was no significant difference between girls’ and boys’ scores at baseline. These results reflect similar findings from samples of neurotypical children, which found that girls developed pretend play skills earlier than boys (Li et al., 2016; Lucisano et al., 2021; Stagnitti, 2022).
Limitations and Strengths
The strengths of this study included the involvement of four schools, which allowed for a range of geographical regions and variations in staff and school culture, and hence more confidence to suggest that results may be generalized to a wider population of children who attend special schools. The variety of assessments conducted was a strength because it allowed information on a range of developmental areas.
Researcher bias was a limitation. To limit our bias as well as that of the teachers, we ensured that teachers in the study were not aware of the results of the play and language assessments, and we did not analyze the SSIS or SDQ questionnaires until after all the baseline and follow-up data were collected. As Wadley trained the staff, monitored the fidelity of the study protocols, and conducted the baseline assessments, Stagnitti randomly assessed 50% of the children at follow-up. Stagnitti had no contact with any of the schools, did not know the children or their potential for play ability, and had not viewed any baseline assessments.
This study was a single cohort study with no control group. This meant that it was not possible to determine whether the changes in child development and learning could be attributed solely to Learn to Play Therapy or to other factors within the school program. To account for this limitation, the design of the research study triangulated several sources of data across more than one site.
Recommendations for Future Research
There is limited research on play-based curriculums in special schools. We recommend that future research include a randomized control trial within special schools to compare the Learn to Play Therapy intervention with traditional curriculums. Building on this research, we recommend further exploration into the role of occupational therapists in supporting teachers to implement Learn to Play Therapy.
Implications for Occupational Therapy Practice
This study was part of a program evaluation for a play intervention in special or specialist schools and was conducted in three stages (the current study was Stage 2). In Stage 2, children’s development was monitored over 7 mo for pretend play ability, language, social skills, emotional regulation, and academic competence. Stage 1 analyzed the understanding of pretend play in childhood development and the value of teachers and therapists within these schools. This initial stage included focus groups from the same schools that participated in this study. Findings revealed that schools that had access to an occupational therapist were more confident in play assessment, choosing appropriate developmental toys for children, and encouraging self-initiation of play (Wadley & Stagnitti, 2020b). Additional data collection in Stage 3 of this program evaluation (following from this current study) revealed that occupational therapists’ role within the schools was highly valued to support teaching staff in coregulation strategies, how to prompt play ideas, and supporting children through role play (Wadley & Stagnitti, 2020a). This study has the following implications for occupational therapy practice: ▪ Occupational therapists can support teachers with coregulation strategies and support students’ engagement through play. ▪ Focusing on children’s pretend play ability at the beginning of a school year could be influential to language, social, and academic competence over the school year. ▪ The results of this study confirm previous research and concur with staff perceptions of the results of the play intervention and child outcomes.
Conclusion
In this study, changes in the children’s development over the course of Learn to Play Therapy were positive. Pretend play was found to be influential to language, social skills, academic competence, and narrative development among a group of special school children. The results form part of a wider program evaluation investigating the implementation of Learn to Play Therapy in special schools. The results reflect the changes that teachers and therapists saw in the children across the course of the study, reported through focus groups and questionnaires in Stages 1 and 3 of the larger program evaluation (Wadley & Stagnitti, 2020a, 2020b). The findings support the role of play-based learning in special schools. Further research is required to explore the occupational therapist’s role in supporting play-based learning in a school setting.
Footnotes
Acknowledgments
We thank the children, families, and school staff who provided their time to participate in this research. Without their interest and willingness to implement Learn to Play Therapy, we would not have had the opportunity to complete this research and explore how Learn to Play can support children with their development. Thank you for allowing us into your school settings.