Teaching symbolic play skills to children with autism ... - Springer Link

17 downloads 138 Views 1MB Size Report
play actions at levels similar to that of language-matched (ypical controls. In .... Parents were simply asked to play with the child using the training toys, and were ...
Journal of Autism and Developmental Disorders, VoL 25, No. ~ 1995

Teaching Symbolic Play Skills to Children with Autism Using Pivotal Response Training I Aubyn C. Stahmer 2 University of California, San Diego

Used Pivotal Response Training (PRT) to teach 7 children with autism to engage in symbolic play behaviors. Symbolic play, complexity of play behavior, and creativity of play were assessed. In addition, generalization measures were obtained across settings, toys, and play partners. Interaction with the play partners and comparison with typical controls were also examined. Results indicated that children with autism rarely exhibited symbolic play before training or after a control condition. After specific symbolic play training using PRT, all of the children learned to perform complex and creative symbolic play actions at levels similar to that of language-matched (ypical controls. In most cases the children generalized their play to new toys, environments, and play partners and continued to engage in symbolic play behavior after a 3-month follow-up period. In addition, interaction skills improved after training. Treatment implications for these findings are discussed.

1This research was supported by U.S. Public Health Service Grants MH39434 (P. I. Laura Schreibman), MH28210 (P. I. Robert L. Koegel), and MH10385 (P. I. Aubyn C. Stahmer) from the National Institute of Mental Health. This research was conducted in partial fulfillment of a doctoral degree (University of California, San Diego). The author gratefully acknowledges Laura Sehreibman for her helpful reviews of the article; Sara Berns, Julie Hu, Lisa Koch, and Nicole Palardy for their help with reliability and data collection; Karen Pierce and Danielle Thorp for helpful comments on the final draft of the manuscript; and m o s t especially the families who participated in the project. Portions of this research were presented at the 18th Annual Association for Behavior Analysis: International Convention, San Francisco, California, May 25-28, 1992. 2Address all correspondence to Aubyn Stahmer, Psychology Department, State University of New York at Cortland, PO Box 2000, Cortland, New York 13045. 123

0162-3257/95/0400-0123507.50/09 1995Plenum Publishing Corporation

124

Stahmer

Researchers have suggested that a lack of symbolic play skills may be one of the characteristic features of autistic disorder, and that this behavior might be useful for distinguishing children with autism from children with other pervasive developmental disorders (Atlas, 1987). Although there has been some research examining the nature of the symbolic play deficit in children with autism (see Jarrold, Boucher, & Smith, 1993, for review) there have been few attempts to teach symbolic play to these children. Symbolic play may be defined as play in which: "1) the subject is using an object as if it were another object (e.g., a block is a car; a banana is a telephone), and/or 2) the subject is attributing properties to an object which it does not have (e.g., a toy stove is "hot"), and/or 3) the subject is referring to absent objects as if they were present (e.g., pantomime)" (Baron-Cohen, 1987, p. 140; Ungerer & Sigman, 1981). A sample form of representative play is functional play, which includes play in which objects are used in ways appropriate to their conventional function (BaronCohen, 1987). All nf the c . l l r r e n t r ~ a r c h examining s2,.,mbo!ic,~l~,, ;. ~hi!a . . . . . . ;,1. autism has found that these children perform fewer symbolic play actions and less complex actions than typical children of the same language ability (e.g., Baron-Cohen, 1987; Lewis & Boucher, 1988; Riguet, Taylor, Benaroya, & Klein, I981; Wing, Gould, Yeates, & Brierly, 1977). These symbolic play deficits do not seem to be due to mental retardation, which often accompanies autism. Although many children with developmental disabilities are delayed in their development of symbolic play, studies indicate that there are no qualitative differences between the play of mentally handicapped children without autistic disorder and typical children of the same mental age (Beeghly, Weiss-Perry, & Cicchetti, 1989; Hill & McCune-Nicolich, 1981). Children with autism, on the other hand, tend to show an absence of spontaneous symbolic play regardless of their mental age, indicating some type of play deficit that goes beyond mental retardation in general (Baron-Cohen, 1987). Further research has examined the effects of modeling and verbal prompting on the symbolic play of children with autism. Many researchers have found that children with autism perform more symbolic play actions after modeling than they do in spontaneous situations, however they are still not performing at typical levels (Riguet et al., 1981; Ungerer & Sigman, 1981). Lewis and Boucher (1988), however, found that when verbally elicited or instructed to play symbolically, children with autism perform at levels similar too that of typical children of the same language ability. This research suggests that children with autism may learn to perform symbolic play actions if taught in a paradigm that includes modeling or prompting. . . . . . . . . . . . . . . . . . . . . . . . .

j~a~j

Aa~

~ u

~ L ~ A a

VV~II, ll

Teaching Symbolic Play to Children with Autism

125

Differences in the play behavior of children with autism during spontaneous and elicited play situations may be due to the pervasive lack of motivation seen in these children (e.g., Schreibman, 1988a). Decreased motivation to perform symbolic play actions may be due to attention difficulties (e.g., Courchesne, 1987; Schreibman, 1988a), a history of task failure (e.g., Clark & Rutter, 1979), competing behaviors such as manipulative play that may be more reinforcing (Lewis & Boucher, 1988), or a combination of these. This lack of motivation is especially evident when teaching difficult and abstract skills such as language and symbolic play. To teach children with autism to perform symbolic play actions, the teacher must provide a context in which the children are motivated to play in this manner. Koegel, O'Dell, and Koegel (1987) have developed a method for increasing motivation, and consequently learning, in children with autism. The program, called Pivotal Response Training (PRT), works to increase motivation while teaching important skills, typically language acquisition, to hard to teach children with autism. Important aspects of training include turn-taking, reinforcing attempts at appropriate responding, frequent task variation, allowing child's choice of activities, interspersing maintenance tasks, and using natural consequences. This type of training has proved very successful in increasing language skills in children with autism (Koegel et al., 1987; Laski, Charlop, & Schreibman, 1988). In the present investigation the principles of PRT were used to teach symbolic play skills to children with autism. The purpose of the present investigation was to: (a) assess the feasibility of teaching symbolic play skills to children with autism who are developmentally ready, using PRT; (b) examine individual differences that affect acquisition of these skills; (c) assess generalization and maintenance of the behavior changes across setting, playmates, and toys; (d) examine changes in interaction skills after symbolic play training; (e) examine changes in symbolic play relative to the play of language-matched typical controls; and (f) control for the effect of interaction with an adult and exposure to toys alone on play skills.

METHOD

Participants Participants included seven male children with autism, independently diagnosed by organizations not associated with this research. Diagnostic

126

S~hmer Table I. Child Characteristics at Pretraining Assessments a

Child

Chronological age

StanfordBinet

Leiter

Justin

4,3

Children with autism 3,0 3,7

58

73

Clark

6,9

2,5

2,11

48

64

Alex

5,11

3,7

4,t0

82

95

Abe

7,2

3,2

4,4

58

73

David

6,1

2,9

3,4

62

68

Eldin

4,11

2,6

2,7

75

111

Brian

6,2

2,5

2,8

48

68

M

6,6

2,7

3,5

62

79

M

3,2

Range

2,5-4,9

PPVT

EOWPVT

Typical children 3,4 2,7-4,10

a PPVT and EOWPVT scores are reported in mental age (years, months). Stanford-Binet and Leiter standard scores are reported.

criteria for Autistic Disorder were those described in DSM-III-R (American Psychiatric Association, 1987). Each of the children had a mental age of at least 2.5 years and had a chronological age of no older than 7.5 years (see Table I). The age constraints ensured that the targeted play task was appropriate to the child's age and functioning level. In typical children, symbolic play skills are well developed by 2 years of age. Therefore, children with autism who exhibited a language ability of at least 2.5 years were chosen to ensure their developmental readiness for the training. The children with autism in this sample, therefore, may have higher language and intellectual abilities than the typical autistic child who may be more severely impaired. The children did exhibit behaviors typical to other children with autism such as lack of social skills, stereotyped behavior, echolalia, and insistence on sameness. Seven typical children matched for expressive language age (Expressive One-Word Picture Vocabulary Test-Revised; Gardner, 1990) with the experimental participants served as playmates for the children with autism. These children were obtained through the subject pool at the University of California, San Diego. These typical children played with the children

Teaching Symbolic Play to Children with Autism

127

with autism once before training, once after training, and once at a 3-month follow-up. In addition, each typical child was filmed playing with the experimenter and the training toys at each time period.

Measures Language Assessments. All language assessments were used as dependent measures of change after training. The Peabody Picture Vocabulary TestRevised was administered to provide a description of the child's receptive vocabulary skills (PPVT, Dunn & Dunn, 1981). The Expressive One-Word Picture Vocabulary Test-Revised (EOWPVT) was administered to assess the children's expressive vocabulary skills. The children's parents completed the MacArthur Communicative Development Inventory (CDI; Fenson et al., 1991). This assessment measured children's language use in the home and looked at expressive and receptive vocabulary, gesture use, sentence complexity, imitation and mean length of utterance (MLU). Functioning Level The Stanford-Binet Intelligence Test, Fourth Edition (Thorndike, Hagen, & Sattler, 1986) and the Leiter International Performance Scale (Leitner, 1979) were administered to provide a general measure of intelligence. V'uteo Measures. The children were videotaped in 14-minute segments before and after training and after a 3-month follow-up period. The children were videotaped, at separate times, with their primary caretaker (6 mothers, 1 father), the experimenter, and a language-age-matched peer. The toys used in these videotaped sessions were the toys used during symbolic play training (see Table II). The children were also fdmed interacting with the experimenter using generalization toys not seen during training (see Table II). Experimenter and parent tapes consisted of 7 minutes of the child playing alone while the adult watched and 7 minutes of the child interacting with the adult. Parents were simply asked to play with the child using the training toys, and were not directed to elicit symbolic play. They were told that there were some "junk" objects (placeholders, see Table II) available to use if the child wished to pretend, and they were aware that the child would be learning to engage in more advanced play during the intervention. During therapist probes the experimenter did attempt to elicit symbolic play during the interaction portion of the assessment. The interaction with the peer consisted of an 8-minute session in which both the children were allowed to interact freely throughout the session. Both children were simply told to "go in and play." Video assessments were obtained both in training and generalization settings. All the following assessments used these video assessments as the source of data.

S~hmer

128 Table IL Settings and Materials

Child

Training setting

Justin

Family room at home containing two sofas, table, television, etc.

Clark

An empty classroom at school

Alex

Family room at home containing sofas, tables, television, etc.

Abe

An empty classroom at school

David

His bedroom at home containing bed, two dressers, and "fort"

Eldin

His bedroom at home containing bed, dresser, and toy shelves.

Brian

Family room at home containing sofas, tables, television, etc.

All children

Generalization setting: a clinic room 2.44 x 4.58m containing a couch, coffee table, and small end table.

Toys

Examples

Training toys

A tea set, an adult male doll, a "Barbie doll" family, Disney Mickey Town Jet Set, Shellcore Little Fix-it Tool set, etc. Placeholder objects included items such as a popsicle stick, a triangular piece of cardboard, screws, a piece of felt, a shoe box,

etc. Generalization toys

A Fisher-Price pool set, a Fisher-Price picnic basket, several farm animals, etc. Several placeholder objects not used during training were included in the set.

Play Behavior. Symbolic play was scored during the interval if the child was engaging in any of the following behaviors: (a) using one object as if it were another object, and/or (b) attributing properties to an object which it did not have, and/or (c) referring to absent objects as if they were present. Play complexity was also scored. Play was considered to be complex if the child performed a sequence of at least three actions related to the same pretend theme. The frequencies of occurrence (in 30-second intervals) of categories of object play and play complexity were obtained from videotaped records. All behaviors were scored during each interval they occurred (i.e., the categories were not mutually exclusive). Creativity of play was scored as well. Play was considered creative if the child performed symbolic play themes not learned during training. Interaction During Play. Tapes were scored in 30-second intervals for the type of response to others' initiations, and the number of initiations made by the child. Responses were scored as either positive, negative, neutral, or no opportunity. In addition, during each interval it was noted whether or not the child initiated toward the play partner.

Teaching Symbolic Play to Children with Autism

129

Reliability of Data Recording. One third of each child's sessions were scored by two observers blind to the purpose of the study. All scoring was done via videotaped recordings of each session. Interobserver agreement for developmental level of play and initiations were calculated using kappa coefficients which were calculated for each category according to the following formula: ( P o - P c ) / ( 1 - Pc), where Po is percentage agreement and Pc is percentage of chance agreement. Reliability for symbolic play was 85%; for play complexity, 77%; and for initiations, 71%. Because responses to others' initiations were scored on a continuum (positive, negative, or neutral) the kappa coefficient was not an appropriate statistic. Interobserver agreement for occurrence and nonoccurrence was calculated by dividing the total number of agreements by the total number of agreements plus disagreements then multiplying by 100. Reliability for responses to others' initiations was as follows: overall 87% (range: 70-100), occurrence 80% (range 61-100), and nonoccurrence 87% (range 72-100). Settings and Materials. See Table II for a description. Experimental Design. To obtain detailed information on individual children, a single subject multiple baseline design across subjects was used (Hersen & Barlow, 1976). That is, data were collected for differing lengths of time for each child during pretreatment measures. Measures were obtained for each child before and after treatment and at a 3-month follow-up. The experimental group also served as its own control. A control condition in which the children with autism received PRT for language, not symbolic play, was included to assess whether specific play training was necessary to increase symbolic play and interaction skills, or whether interaction with an adult and toys was sufficient to increase symbolic play in these children with autism. This control condition will be called language training (LT). Procedure Baseline. Baseline measures were obtained before symbolic play training (SPT) or language training (LT) began. The experimenter administered the standardized assessments. Free Play Assessment. Play behavior was recorded in 14-minute segments over several days. The child's parent, a peer, or the experimenter entered one of the experimental settings with the child and proceeded as described above. The free play assessments were conducted at least once in a generalization setting as well as in the training setting. Children in the typical control group and the children with autism while in the LT condition (control condition) also participated in a free play assessment identical to the one described above. The interactant was the experimenter and the training toys were utilized.

130

Stahmer

Symbolic Play Training (SPT). Five of the children with autism (Justin, Clark, Alex, Abe, and David) received SPT first. Treatment sessions were conducted 3 times weekly for 1-hour per session. After baselines of varying lengths were obtained, symbolic play training began in accordance with the Pivotal Response Training (PRT) manual by Koegel et al. (1989). The treatment was modified to use symbolic play as a target behavior instead of language. Thus, the experimenter required the children to engage in symbolic play in order to gain access to the toys. The following parameters were emphasized during treatment: (a) The experimenter presented toys according to the child's selection of a preferred toy (by eye gaze, touching, or verbal request); (b) the toys were varied frequently, according to the child's interest; (c) the experimenter played with the toys and modeled symbolic actions; (d) if the child failed to respond, the experimenter played with the toy and modeled the response again; (e) the reinforcement contingencies were broad so that either the exact correct response (i.e., a symbolic play act) or an approximation was reinforced; (f) the child's response was reinforced with the opportunity to play with the instructional stimulus and with praise; (g) functional play at lower levels was interspersed with symbolic play to ensure success for the child, to increase variation in play style and to reduce the risk of stereotyped play; (h) turn-taking by the experimenter was used to provide exemplars as well as to increase the child's social interaction sells; and (i) as the child improved at symbolic play, he was expected to engage in more complex play. Language Training (LT). Each of the children with autism also received LT three times per week in 1-hour sessions for 8 weeks in the same setting in which they received SPT. Eldin and Brian received LT first, while the rest of the children had LT after SPT follow-up was completed. Video assessments were obtained with the experimenter and the training toys, and language assessments were completed (or follow-up assessments from SPT were used). LT began in accordance with the PRT manual by Koegel et al. (1989). The treatment was identical to SPT except the target behavior was language instead of symbolic play. See manual for a complete description. RESULTS

Symbolic Play and Play Complexity All seven of the children with autism who participated in the symbolic play training showed an increase in symbolic play and play complexity after training. In general, these increases were most apparent

Teaching Symbolic Play to Children with Autism

131

in probes that included the therapist and the training toys, and least apparent in the probes where the child was interacting with a languageage-matched peer. No increases in play behavior were evident after language training. Figures 1, 2, and 3 illustrate individual changes in symbolic play and play complexity. The average amount of symbolic play in which each child engaged during each condition is indicated by the shaded area. s. Justin, Clark, Alex; and Brian engaged in little or no symbolic or complex play before training. Typically these children did not use placeholder objects at all before training, or would use them only with insistent prompting. They could put together play scripts only with repeated suggestions from the experimenter. Abe, David, and Eldin performed some symbolic or complex play actions before training. Typically these actions were not spontaneous, but these children could perform actions with placeholder objects after experimenter suggestions. After symbolic play training all of the children performed more complex symbolic play. The children performed learned actions spontaneously, could now readily take suggestions from the experimenter, and could direct story lines with the toys. After training, an average of 35% of the children's symbolic actions were creative (not learned during training) with a range of 9-56%. Alex and Abe performed especially well, performing very complex scripts, and spontaneously integrating placeholder objects into play. Conversely, Brian performed especially poorly, not generalizing to new situations, and using only learned actions. The characteristic that seemed to make this task difficult for Brian was his high rate of stereotyped play. He tended to perform the same action sequences repeatedly. Symbolic play and play complexity remained high at follow-up for Justin, Alex, Abe, and David. These play skills decreased somewhat at follow-up for Clark, Eldin, and Brian. The children's symbolic play behavior did not increase after LT. Brian and Eldin received language training first, and their symbolic play did not increase after this training. Other children often decreased their symbolic play after LT, probably due to the fact that this assessment period was equal to a 5-month follow-up period. In summary, the children, on average increased their symbolic play behavior substantially, and in a clinically significant manner after SPT. This same increase was not evident after LT. For 6 of the 7 children the symbolic play learned generalized to new toys and to new individuals. Play complexity also increased for all of the children. Five of the children decreased at least some of their play skills at follow-up but these still remained higher than pretraining.

132

Stahmer

100-

Pre SPT

Post SPT

Foll_ow_-upSPT

Post LT

8060.,~,Syrnbolic , Play

4020Z

Justin

Play Complexity

0-

r tx.1 Z ,.r er

Clark

o o 0 z

Alex

MP

GT

T T T

T TM

PG

T

PMG

T T

SESSIONS Fig. 1. The percentage of 30-second intervals containing symbolic play and play complexity during baseline, post symbolic play training (SPT), 3-month follow-up, and post language training (LT), for children who received SPT and then LT. Generalization probes in which a peer is the interactant are noted by a P, probes in which the interactant was the child's mother are noted by an M, probes in which the interactant was the child's father are noted by a D, and probes in which the generalization toys were used are noted by a G. In probes denoted by a T, the interactant was the therapist and the training toys were used. The shaded areas denote the average amount of symbolic play in the therapist probes for each condition.

Teaching Symbolic Play to Children with Autism

Pre SPT

133

Post SPT

Follow-upSPT PostLT rt__

i T

Z

_.Q i.r.l

z

9 z

David

TT

TTTT

P TGGG

TTMGMPT

TG P TMT

SESSIONS Fig. 2. The percentage of 30-second intervals containing symbolic play and play complexity during baseline, post symbolic play training (SPT), 3-month follow-up, and post language training (LT), for children who received SPT and then LT. See Fig. 1 caption for legend information.

Comparison of Children with Autism and Typical Children The symbolic play of children with autism was compared to that of typical controls matched for expressive vocabulary. Recall that 5 of the children received SPT first and 2 of the children received LT first. The 2 children who received LT first were not assessed at periods equivalent to the other 5 children because of the 8 weeks of LT they received first. In addition, receiving LT first may have somehow effected SPT or vice versa. Because of this it was reasoned that it would be inappropriate to include these 2 children in any of the statistical analyses. The results of the analyses on play type, unique actions, and play complexity showed the same pattern of performance. Each measure was analyzed according to the proportion of intervals in which the child engaged in the behavior. In each of the analyses of variance (ANOVAs), the

Stahmer

134

Pre SPT Post LT

100.. t

Post SPT

pla

z

o

N U

z

Follow-upSPT

Symbolic 40.

Eldi~

20. 0. TGTT

TTTPTT

TMT

MT

~100_

z

80. Play

~ 60.

Complexity

Brian

1

40~

20~ r

-

i

o. TT

TP

MTT

TT TT T TT SESSIONS

TGGMP

T GPMT

Fig. 3. The percentage of 30-second intervals containing symbolic play and play complexity before language training (LT), after LT and before symbolic play training (SPT), post SPT, and 3-month follow-up for children who received LT then SPT. See Fig. 1 caption for legend information.

between-subjects variable was group (autism, typical). The within-subjects factors were time period (pre, post, follow-up) and the behavior to be measured (i.e., play type [spontaneous symbolic play, elicited symbolic play, functional play]; unique actions [unique symbolic actions performed during play]; play complexity [spontaneous, elicited]). Each of the analyses showed a main effect for time period which was qualified by a Group x Time Period interaction. Follow-up one-way analyses of variance for each measure indicated that the groups differed significantly only during baseline measures. Children with autism engaged in significantly less play behavior before training than did typical children on each measure. In addition, the behavior of the children with autism increased over time. Tukey HSD (p < .01 or .05) analyses were performed on each measure and in every case the children with autism engaged in significantly less behavior before training than they did at either

Teaching Symbolic Play to Children with Autism

135

posttraining or follow-up measures. No Behavior x Group interactions were evident, indicating that each behavior in all the analyses increased in similar manners (e.g., the children with autism had increases in both spontaneous and elicited play behavior). The results for each of the analyses are displayed in Table III. Each behavior is also listed separately to allow a more detailed view of the results. Correlational Data

Correlations were performed between several of the assessments for the children with autism to assess the relationship between play and language and elicited and spontaneous play. Several interesting signific a n t correlations were found. The children's scores on both the EOWPVT and the PPVT correlated significantly with elicited symbolic play after symbolic play training (r = .902, p < .05; and r = .824, p < .05m, respectively). In addition, spontaneous symbolic play correlated with sentence complexity as m e a s u r e d by the CDI b o t h at pretraining assessments (r = .926, p < .01) and at posttraining assessments (r = .859, p < .05). Symbolic play, then, may be related to both vocabulary use and complexity of language. Both overall symbolic play levels and spontaneous symbolic play levels correlated with play complexity (r = .851, p < .05; r = .859, p < .05, respectively). These correlations indicate that the children who engaged in high levels of symbolic play also engaged in complex play. This suggests that the children were not simply performing repetitive, simple play acts but instead used their symbolic play skills in complex play. The level of symbolic play in which the children engaged during posttraining assessments correlated significantly with the children's spontaneous symbolic play (r = .877, p < .05). This might be useful in interpreting results from other research in which only elicited play results are reported. Elicited symbolic play, then, may be a good predictor of spontaneous symbolic play behavior. Symbolic play behavior before training did not predict success in training nor did IQ score. Scores on the StanfordBinet and the Leiter did not correlate with any outcome measures. Interaction with Others

Overall, the children with autism engaged in better interactive behavior after symbolic play training than they did before training. Change scores for these behaviors are illustrated in Figures 4 and 5. The children with autism who tended to be the best interactants (i.e., made the most initiations and responded most positively) were the same children who

34.6 69.4

52

70.4

36.6 67 56.8

53.5

Post

41.6 68

54.8

58,4

42.8 55.2 50.6

49.5

Follow-up

22.8 63

42.9

77.6

14.2 51.2 65.6

43.7

Pre

25.6 65.2

45.4

67.4

13.2 57.2 67.4

42.5

Post

Control

25.6 75.6

50.6

65.8

65.6 64.8 65.8

49.8

Follow-up

a Italicized information is for description only; because there were no interactions individual statistics were not performed. bSignificantly different from both other time periods and control group, p < .01.

1 7.6

4,3b

Play complexity

Spontaneous complexplay Elicited complexplay

59.8

% Unique actions

1.3 8.8 40,2

16.90

Play behavior

Spontaneous symbolicplay Elicited symbolicplay Functional play

Pre

Behaviora

Autism

Table IlL Means Table for Behavioral Results

Teaching Symbolic Play to Children with Autism

137

Changes in Positive Responses to Others' after Symbolic Play and Language Training

5:

100 90 80 70 60 50 40 30 20 10 0 -10 -20 -30 -40

9

Symbolic Play Training

[] Language Training

Overall

SymbolicPlayF'u-'st Language Training First Group

Fig. 4.

Change scores illustrating the children's changes in their positive responses to others' initiations after SPT and LT.

learned to engage in very high levels of symbolic play (specifically, Alex and Abe). For all of the children positive responses increased substantially, while initiations remained low even after SPT. Interaction skills did not increase after language training. In fact, in most cases interaction decreased after LT. The decrease was probably due to the longer follow-up period (for children who received SPT first) or because of the change in response requirements (for children who received LT first).

DISCUSSION Results of this investigation indicate that children with autism who have sufficient language skills can learn to engage in symbolic play at levels similar to that of language-matched typical children. These results imply that other children with autism who have expressive language abilities of about 2.5 years or more may also benefit from this training. For children who do not have that language capacity this intervention may not be developmentally appropriate. In addition to the changes in symbolic play several other findings were apparent: (a) the children with autism learned to perform creative symbolic play; (b) their play complexity increased; (c) interaction skills improved during play after symbolic play training; and (d) data indicate that specific symbolic play training was necessary to promote changes in these skills.

Stahmer

138

100

Changes in Initiations to Others' after Symbolic Play and Language Training

90 80 70 60

9 Symbolic Play Training [] Language Trai.nJ.ng

50 40 30 20 10 0 -10 -20 Oveza]l

Symbolic PlayFn-st LanguageTmiaingFrost Group

Fig. 5. Change scores illustrating the number of initiations the child made to the play partner after SPT and LT.

After play training, children with autism performed objectively as well as typical children with similar language ability. Not only did their elicited play increase but the children with autism spontaneously performed complex symbolic play actions while playing alone as well. This indicates that they did not require adult instruction to use these new skills. Correlational results and comparisons with typical children suggest that although children with autism do not engage in spontaneous symbolic play at their language age level before SPT, they may be taught to perform symbolic actions at a level similar to that of their language age. Children with better language skills engaged in the highest amount of symbolic play after training. When provided with increased motivation to use symbolic play actions and the tools to learn symbolic play the children with autism performed more similarly to other developmentally disabled populations (Beeghly et al., 1989; Hill & McCune-Nicolich, 1981). Social validation of these changes in play skills would be a useful measure in the future. After symbolic play training, interaction skills increased as well. The children began responding more positively to initiations made by adults during the play sessions. Some of the problems children with autism have when interacting with others, then, may be due to a lack of skills. The children may not have known what types of behaviors were expected of them, and therefore, responded more negatively to initiations associated with difficult play actions. In the present investigation children who engaged in high levels

Teaching Symbolic Play to Children with Autism

139

of symbolic play improved their interaction skills more than children who did not play as well after training, and interaction skills did not increase after language training alone. The increase in interaction is a promising finding for programs that aim to integrate these children into more typical classrooms. Although the children responded better to adults, they still did not respond well to peer initiations. Peers are typically less tolerant of poor play skills than adults (Oke & Schreibman, 1990). Additionally, the children with autism did not initiate interaction very often after training. Because initiations and responses are so important to integration with typical children, specific initiation and response to peer programming might be required along with play training (Krantz & McClannahan, 1993). The children in the current investigation generalized their play relatively well to new situations. Although the children performed best when engaged with the therapist and the training toys, most of them engaged in symbolic play in new settings, with new toys and with their parents. This may have been due to the high .number of examples the experimenter provided for the children during PRT (Stokes & Baer, 1977). The fact that the children performed complex symbolic actions with the generalization toys indicates their understanding of the actions. They developed new stories and actions with novel toys. The children with autism did not generalize as well as to play with a peer. To combat this problem, typical children might be included in training to familiarize the children with autism with more typical play behaviors. Another concern that arises when teaching complex skills to children with autism is the often stereotyped nature of their interactions. Most of the children in the present investigation learned to be creative and spontaneous in their play. They developed new play themes not suggested by the experimenter, and could engage in complex symbolic play with novel toys. Overall, the children were quite flexible in their play, and did not appear to mind variation or interruption in their play themes. They used as many unique play actions as did the typical controls. This flexibility and creativity may be due in part to the large number of examples used in PRT and the flexible and varied nature of the program. One of the children, Brian, did have difficulty with stereotyped play, which seemed to interfere with his learning. It is unclear why this occurred, however, it is important for teachers and parents to take note of any insistence on sameness in a child's play routine, or failure to generalize. Again, children with the best language skills were the most creative and spontaneous during play.

140

Stahmer

This project was designed to assess the feasibility of teaching symbolic play to children with autism with sufficient language ability. Now that it has been established that they can learn this skill, generalization and maintenance can be programmed into the training situation. The children who were the best "players" generalized to new situations better than the children who were not as competent at symbolic play, implying that a certain level of competence at the task may be needed to ensure generalization. Training may-need to continue for longer than 8 weeks for some children. Teaching parents, teaches, and peers to engage in symbolic play training with children with autism would be useful as well (Schreibman, 1988b). Using typical peers to teach play might also ensure that the play is indeed typical, and would provide more creative exemplars. Pivotal Response Training has proven a useful tool for teaching the complex social skill of symbolic play. This type of training may provide children with autism of appropriate language ability a motivating, somewhat predictable context for learning complex skills that generalize to new contexts. The training is flexible, and allows the therapist to require more difficult responses as the child progresses. Teaching play skills to children with autism may be important for integration into typical classroom situations, and could be used to increase interaction skills as well. Future research should attempt to assess the use of PRT for increasing play and interaction with peers, and concentrate on maintaining the behavior change over time.

REFERENCES American Psychiatric Association. (1987). Diagnostic and statistical manual of mental disorders (3rd ed., rev.). Washington DC: Author. Atlas, J. A. (1987). Symbol use by developmentally disabled children. Psychological Reports, 6L 207-214. Baron-Cohen, S. (1987). Autism and symbolic play. British Journal of Developmental Psychology, 5, 139-148. Beghly, M., Weiss-Perrry, B., & Cicchetti, D. (1989). Structural and affective dimensions of play development in young children with down syndrome. International Journal of Behavioral Development, 12, 257-277. Clark, P., & Rutter, M. (1979). Task difficulty and task performance in autistic children. Journal of Child Psychology and Psychiatry, 20, 271-285. Courchesne, E. (1987). A neurophysiological view of autism. In E. Schopler & G. B. Mesibov (Eds.), Neurobiological issues in autism (pp. 285-324). New York: Plenum Press. Dunn, L M., & Dunn, L. M. (1981). Peabody Picture Vocabulary Test-Revised: Manual: Circle Pines, MN: American Guidance Service. Fenson, L., Dale, P. S. Reznick, J. S., Thai, D., Bates, E., Reilly, J. S., & Hartung, J. P. (1991). Technical manual for the Mac.Arthur Communicative Development Inventories: Preliminary version. San Diego, CA: San Diego State University. Gardner, M. F. (1990). Expressive One-Word Picture Vocabulary Test. Novato, CA: Academic Therapy Publications.

Teaching Symbolic Play to Children with Autism

141

Hersen, M., & Barlow, D. H. (1976). Single case experimental designs: Strategies for studying behavior change. New York: Pergamon. Hill, P. M., & McCune-Nicolich, L. (1981). Pretend play and patterns of cognition in Down's syndrome children. Child Development, 5~ 611-617. Jarrold, C., Boucher, J., & Smith, P. (1993). Symbolic play in autism: A review. Journal of Autism and Developmental Disorders, 23, 281-307. Koegel, R. L., O'Deil, M. C., & Koegel, L. K. (1987). A natural language teaching paradigm for nonverbal autistic children. Journal of Autism and Developmental Disorders, 17, 187-200. Koegel, R. L., Schreibman, L., Good, A., Cerniglia, L., Murphy, C. & Koegel, L. K. (1989). How to teach pivotal behaviors to children with autism: A training manual Santa Barbara: University of California, Santa Barbara. Krantz P. J., & McClannahan, L E. (1993). Teaching children with autism to initiate to peers: Effects of a script fading procedure. Journal of Applied Behavior Analysis, 26, 121-132. Laski, K. E., Charlop, M. H., & Schreibman, L. (1988). Training parents to use the natural language paradigm to increase their autistic children's speech. Journal of Applied Behavior Analysis, 21, 391-400. Leiter, R. G. (1979). Leiter International Performance Scale: Instruction manual Chicago: Stoelting Company. Lewis, V., & Boucher, J. (1988). Spontaneous, instructed and elicited play in relatively able autistic children. British Journal of Developmental Psychology, 6, 325-339. Oke, N. J., & Schreibman, L. (1990). The effects of social initiations on an autistic child's social behavior: Generalization and collateral behavior change. Journal of Autism and Developmental Disorders, 20, 479-497. Riguet, C. B., Taylor, N. D., Benarogya, S., & Klein, L. S. (1981). Symbolic play in autistic, Down's and normal children of equivalent mental age. Journal of Autism and Developmental Disorders, 11, 439-448. Schreibman, L. (1988a). Autism. Newbury Park, CA: Sage. Schreibman, L. (1988b). Parent training as a means of facilitating generalization in autistic children. In R. H. Horner, G. Dunlap, & R. L. Koege (Eds.), Generalization and maintenance: Lifestyle changes in applied settings (pp. 21-40). Baltimore: Paul H. Brooks. Stokes, T. F., & Baer, D. M. (1977). An implicit technology of generalization. Journal of Applied Behavior Analysis, 10, 349-367. Thorndike, R. L., Hagen, E. P., & Sattler, J. M. (1986). Guide for administering and scoring and Stanford-Binet Intelligence Scale (4th ed.). Chicago:Riverdale. Ungerer, J. A., & Sigman, M. (1981). Symbolic play and language comprehension in autistic children. Journal of the American Academy of Child Psychiatry, 20, 318-337. Wing, L., Gould, J., Yeates, S. R., & Brierley, L. M. (1977). Symbolic play in severely mentally retarded and autistic children. Journal of Child Psychology and Psychiatry, 18, 167-178.