Effects of Asking Autistic Children Questions During Play

Recherche de Caroline Peters sur l’impact des questions posées pendant le jeu chez des enfants autistes.

Auteure : Caroline Peters – MSc Applied Behaviour Analysis
Université : Bangor University
Année : 2012

Résumé professionnel en français

Les comportements suivants ont été observés : le jeu avec les jouets, les interactions sociales, la stéréotypie, la palilalie, les interactions sociales négatives et les productions verbales (énoncés spontanés et réponses aux adultes).

Pour quatre enfants sur cinq, la condition « Questions » a entraîné :

• une diminution des interactions sociales avec l’expérimentatrice ;
• une baisse du langage spontané ;
• et, pour certains enfants, une augmentation de comportements de stéréotypie ou d’évitement.

Ces résultats suggèrent que, pour de nombreux enfants autistes, les questions fréquentes pendant le jeu peuvent fonctionner comme des demandes ou des contraintes, rendant la situation de jeu moins agréable et diminuant l’envie de communiquer spontanément. À l’inverse, des interactions plus naturelles, basées sur le suivi de l’enfant (commentaires, imitation, jeux partagés), favorisent davantage l’engagement social et la communication.

Cette recherche remet en question certaines pratiques courantes, où l’on multiplie les questions pour « faire parler » l’enfant. Elle souligne l’importance d’apprendre aux adultes d’autres façons d’entrer en relation pendant le jeu : réduire la pression, suivre l’initiative de l’enfant, modéliser le langage, et créer un contexte où l’enfant a envie de communiquer.

---

Full article in English

The text below is the full body of the thesis as submitted for the MSc in Applied Behaviour Analysis (Bangor University, 2012), reformatted for web reading.

Title Page

Effects of Asking Autistic Children Questions during Play

Caroline Peters
Presented for MSc Applied Behaviour Analysis
October 2012 – Bangor University

Abstract

We investigated the possible adverse effects of asking questions during play on the social interaction, toy play, and spontaneous speech of children with autism. Five children (5–10 years old) participated. In an alternating treatment design, we compared three free-play conditions: (1) Questions: the experimenter asked the participant a question on a 20-s variable-time schedule. (2) Comments: the experimenter stated a comment at the same rate. (3) Neutral: the experimenter asked no questions and made no comments. Each condition lasted 10 minutes and was counterbalanced over three 1-hour sessions. All sessions were conducted by the experimenter in the same playroom.

We recorded the following responses: (a) toy play: manipulating, carrying or holding toys appropriately, (b) social interaction: being oriented towards the experimenter and looking at, smiling to, speaking to the experimenter, (c) stereotypy: repetitive movements with body or an object, (d) palilalia: dysfunctional delayed repetition of words or sentences, (e) negative social interaction and, (f) utterances: functional words or sentences. Four out of five participants were less likely to interact with the experimenter and emitted fewer spontaneous utterances in the Questions condition compared to the Neutral condition. In addition, the participants engaged more in undesired behaviours like stereotypy in the Questions condition. A possible function of the child’s stereotypy is to escape from demands. This suggests that questions at this high rate may be aversive for some children, who then avoid interacting with the experimenter.

Keywords: embedded prompting, naturalistic teaching, asking questions, incidental teaching, autism, children, pivotal response training.

Introduction

Effects of asking autistic children questions during play

Young children learn many social and language skills during play (Koegel, Koegel & Dunlap, 1996; Moyles, 1989). Autistic children often lack appropriate play skills. This reduces their opportunities to learn language skills during play. The environment in which children learn language is strongly influenced by the parents, who have many opportunities to teach their children communication skills, pretend play and play with toys (Kaiser, Hancock & Nietfeld, 2000). Using children’s motivation for toys to teach them language skills is an example of naturalistic language teaching (Peterson, 2004).

Teaching language to children with naturalistic teaching procedures

Various naturalistic language teaching procedures have been described during the last 20 years: Pivotal Response Training (Koegel et al., 1996), Incidental Teaching (Hart & Risley, 1974, 1975), Enhanced Milieu Teaching (Kaiser et al., 2000), Embedded Teaching (Daugherty, Grisham-Brown & Hemmeter, 2001; Horn, Lieber, Li, Sandall & Schwartz, 2000). They have been shown to be effective in teaching children language skills (Peterson, 2004). For instance, with Incidental Teaching, children in a school setting have been taught to name objects, then name characteristics of objects, then make compound sentences (Hart & Risley, 1974, 1975). Parents are often very enthusiastic about the training itself and the progress their children make (Kaiser et al., 2000).

In naturalistic language teaching procedures, adults often pose questions to get a desired verbal response from a participant. For instance, when the participant shows that he wants a ball on a shelf, the adult asks the participant: “What is that called?” When the participant says “ball”, the adult will give him the ball (Hart & Risley, 1974, 1975). Questions which are asked to teach a participant language skills or to prompt a verbal response from the participant during naturally occurring activities are also called “embedded prompts” (Horn et al., 2000).

Several studies show that children can make large gains in the acquisition of language during these naturalistic language-teaching procedures (Peterson, 2004). However, some studies indicate that embedded prompting may have aversive effects on children’s play behaviours (Hart & Risley, 1974, 1975; Heal, Hanley & Layer, 2009; Heal & Hanley, 2011). The data recorded by Hart and Risley (1974, 1975) show that the children request toys less frequently towards the end of the incidental teaching study. Heal et al. (2009) and Heal & Hanley (2011) found that children touched toys less when embedded prompts were used.

Possible adverse effects from asking questions during play

Hart and Risley (1974, 1975) argue that the children did not avoid the incidental teaching situation compared to the play situation without incidental teaching. They infer this from the fact that the children preferred requesting the toys used for incidental teaching rather than playing with the freely available materials. A weakness of both studies is that the incidental teaching toys were different and possibly more interesting than the free toys. In addition, because the children could always play with the free toys, they may have been more interested in the incidental teaching toys because of their state of deprivation with respect to these toys.

Furthermore, in both their studies, a downward trend can be seen in the percentage of requests from the children to their teachers. This may be caused by the fact that the children had to answer increasingly more questions before they got access to the toys. This may have made the incidental teaching procedure increasingly aversive for them. At first, little response effort was necessary to get the item, but later on much more had to be done in order to get the toys. In addition, it is possible that for some children this response requirement eventually became too difficult and therefore contributed to the decrease in requests.

Heal, Hanley and Layer (2009) compared three strategies to teach six children colour-name and object-name relations in a multi-element design. Each strategy included a free play situation but differed in the amount of teacher directedness. The purpose of the study was to see which teaching strategy was preferred by the children and most effective in teaching the target skills. The children were taught to name animals or colours in Spanish.

In Strategy 1, the teacher first showed and labelled every target item once. She then played with the participant and praised him every time he labelled the animal or colour correctly. If the participant labelled the animal or colour incorrectly the teacher did not correct him. In Strategy 2, the teacher immediately started the participant-led play session and used embedded prompts to teach the participant the target name relations. When the participant showed interest in a target item, the teacher asked, “What colour is that car?” without interrupting his play. The teacher then corrected the participant if his answer was incorrect and provided praise if his response was correct.

In Strategy 3, the teacher first started to teach the participant the target relations during a brief direct instruction session. She held up a colour or animal card and asked: “What colour [animal] is this?” If the participant was correct, he received praise and tokens; if he was wrong, the teacher corrected the participant. After this brief session, she played with the participant and used embedded prompts as in Strategy 2. Each strategy was associated with a coloured card, which was presented to the children before they entered the playroom. After the children had experienced all three teaching strategies during forced-choice blocks, the teacher conducted free-choice blocks where the children could choose the teaching strategy.

The results showed that the children preferred a strategy in which they received direct instructions (Strategy 3) over a strategy where embedded prompts had been used (Strategy 2). However, the children may have preferred Strategy 3 for other reasons: they additionally received conditioned reinforcers (tokens) for correct responses, they had more learning opportunities and as a result were more successful in naming the target relations. The children made proportionally more errors in Strategy 2 and had a lower rate of positive reinforcement. All these conditions may have led them to avoid Strategy 2 (Carbone, Morgenstern, Zecchin-Tirri & Kolberg, 2007).

The most interesting finding from this study is that the authors note that one participant touched the toys used for embedded teaching less across time in Strategy 2. The authors hypothesize that the delivery of questions while the children were playing might have interrupted their play.

In a follow-up study, Heal and Hanley (2011) investigated the effects of embedded prompts on the play behaviours of one child. Embedded prompts were only used in Strategy 2, in which they asked the child to name the target relation when she touched a target toy. Only during Strategy 2 sessions did the percentage of intervals during which the child played with the toys quickly decrease to zero. Their conclusion was that the embedded prompts had functioned as a punisher for the child’s play.

In summary, research indicates that parents play an important role in their child’s language development (Hart & Risley, 2003; Kaiser et al., 2000). An important element of many naturalistic language-teaching procedures is the use of embedded prompts. These techniques can benefit children’s language and communication skills but may also have aversive side-effects on play and social behaviour for some children. The present study aimed to investigate more specifically the effects of questions on children’s play and social behaviours across different play conditions.

Method

Participants

Participants were five boys with autism who had been referred by their parents. They were selected based on their availability. All children had language delays and deficits in social play skills. Developmental age equivalents for mand, tact, independent play and social play skills were assessed using the Verbal Behavior Milestones Assessment and Placement Program (VB-MAPP; Sundberg, 2008).

Experimental design

To investigate the effects of asking questions during play we used an alternating treatment design (Barlow & Hayes, 1979). Three free-play conditions were compared: Questions, Comments, and Neutral (no questions or comments). All sessions were conducted by the experimenter in the same room and with the same toys. In a one-week period, each participant had three play sessions. Each session was divided into six 10-minute periods. To minimize sequence effects, the conditions were counterbalanced within and across sessions.

Experimental setting

All sessions were conducted in a playroom with four child-sized tables and chairs and large cupboards with a wide variety of toys (e.g., Lego, plastic animals, cars, puzzles, books, Play-Doh, pretend play materials). The same toys were available during all sessions.

Free-play conditions

Before starting the free-play session, the participant had not played with any toys (including computer games) and had not watched any videos for at least two hours. Parents were received in the playroom and the experimenter told the participant that she was going to play with him for one hour. She briefly showed the participant which toys were available and explained that the play session would be filmed and that the parents would return after one hour.

In the Questions condition (Q), the experimenter asked the participant a question on a variable-time (VT) 20-second schedule. Questions varied depending on the situation. When the participant was not engaging in any play or when the experimenter did not know what the participant wanted to play with, she asked: “What do you want to play with?”, “Do you want to play with …?” or “Shall we play with the …?”. When the participant touched or reached for a toy, the experimenter asked questions like: “What is that?”, “What colour is that?”. When the participant was performing an action or engaged in an activity without toys she asked questions such as “What are you doing?”. When the participant made a doll perform an action she asked: “What is that?” or “What is he/she doing?”.

In the Comments condition (C), the experimenter commented on her toy or on what she was doing on a VT 20-second schedule. For instance, when she had a car she said: “I am driving”, “I have a red car”, “That car is going fast!”, or when she played with pretend food: “I cut the vegetables”.

In the Neutral condition (N), the experimenter did not ask the participant any questions and did not make any comments. She still followed the participant’s lead, played alongside him, and used nonverbal behaviours (e.g., making car noises) but refrained from verbal prompts or comments.

In all three conditions, the experimenter followed the participant’s lead. When the participant engaged in play and/or interacted with the experimenter, she smiled, sang, or responded favourably (e.g., giving a requested toy, answering questions posed by the child). If the participant engaged in inappropriate behaviours such as stereotypy, the experimenter continued with what she was doing as if nothing had happened. At the end of the session, the experimenter told the participant that the play session was finished and that it was time to go home with his parents.

Response definitions

Responses were divided into six categories:

• Toy play: manipulating, carrying or holding toys as the manufacturer intended (including non-symbolic and symbolic play).
• Social interaction: being oriented towards the experimenter and looking at, smiling to, speaking to or gently touching the experimenter (including prompted and spontaneous interaction).
• Stereotypy: repetitive movements with the body or an object with no apparent consequence beyond the movement itself.
• Palilalia: delayed repetition of words or sentences.
• Negative social interaction: behaviours indicating that the child wanted to leave the room or make the experimenter go away.
• Utterances: functional words or sentences emitted by the participant (prompted or spontaneous).

Measurement and inter-observer agreement

All sessions were videotaped. For toy play and social interaction, 10-second momentary time sampling (MTS) was used and data were reported as percentages of intervals. For stereotypy and palilalia, 10-second partial-interval (PI) recording was used. Utterances were counted per 10-minute period and converted to rates per minute.

A second observer independently recorded responses during 30% of each one-hour session. Inter-observer agreement (IOA) was high across all response categories, with mean agreement above 90% for toy play, social interaction, stereotypy, palilalia and utterances.

Results

Percentages of intervals with spontaneous social interaction, toy play, stereotypy, and palilalia, as well as rates of spontaneous utterances, were calculated for each participant and each condition.

For four of the five participants (Will, Leonard, Barry, and Howard), spontaneous social interaction and spontaneous utterances were consistently higher in the Neutral condition than in the Questions condition. Across these four participants, the average percentage of spontaneous social interaction in the Neutral condition was 32% (range 21–41%) compared to 24.5% (range 14–31%) in the Questions condition. The average number of spontaneous utterances in the Neutral condition was 5.8 (range 1.3–9.5) per minute and decreased to 4.65 (range 0.7–7.7) in the Questions condition.

Two of these four participants (Will and Howard) engaged in stereotypy most often in the Questions condition. Will engaged in stereotypy during 15% of the intervals in the Questions condition compared to 11.9% and 13% in the Comments and Neutral conditions respectively. Howard engaged in stereotypy during 1.38% of intervals in the Questions condition, compared to 0% and 0.27% in the Comments and Neutral conditions.

There was no consistent effect of questions on toy play across participants, although some children showed lower levels of toy play in the Questions condition during parts of the study.

One participant, Sheldon, responded very differently from the others. He showed the highest levels of social interaction and spontaneous utterances in the Questions condition and the lowest levels in the Neutral condition. He did not engage in stereotypy at all in the Questions condition, whereas stereotypy occurred at low levels in the Comments and Neutral conditions. For Sheldon, the Questions condition appeared to enhance social engagement and appropriate speech.

Across all participants, responses to questions were much more frequent than responses to comments. Out of 180 questions per child, total responses (correct and incorrect) ranged from 124 to 155, whereas responses to the same number of comments ranged from 6 to 62.

Discussion

The purpose of this study was to examine the effects of asking questions during play on the social interaction, spontaneous speech and toy play of children with autism. We hypothesised that questions could be aversive for some children, leading to reduced interaction and increased avoidance behaviours. The results partially support this hypothesis.

For four of the five participants, questions were associated with less social interaction and fewer spontaneous utterances compared to the Neutral condition. For two participants, questions were also associated with increased stereotypy. These findings are consistent with the idea that for some children with autism, frequent questions during play function as demands and may be experienced as aversive. Stereotypy, in this context, may serve an escape function, helping the child to avoid demands.

However, one participant (Sheldon) showed the opposite pattern: questions improved his engagement and speech, and stereotypy was lowest when questions were used. For this child, the questions may have provided structure, predictability or attention that increased his participation.

We also included a Comments condition to separate the effect of questions from the effect of adult speech in general. For some children, comments were less aversive than questions and were associated with more spontaneous utterances and toy play. Nevertheless, most participants did not respond frequently to comments, which suggests that many children with autism have been more extensively taught to answer questions than to respond to comments. This may become problematic in natural environments where peers primarily use comments rather than questions in conversation.

The heterogeneous patterns of responding observed in this study highlight that the effects of questions depend on the individual child’s history and motivation. Some children, particularly those with a history of escaping from demands via stereotypy, may find question-heavy interventions aversive. For others, questions may provide a helpful framework.

Several limitations should be noted. First, the type of toys available and chosen by the participants influenced their behaviour, and this was not strictly controlled. Second, the Questions condition involved a relatively high rate of questions (approximately three per minute), and it remains unclear whether lower rates would have similar effects. Third, we did not systematically manipulate the difficulty of questions; many questions were easy for the children to answer, and more difficult questions might have been even more aversive.

Despite these limitations, the study adds to the growing body of evidence suggesting that embedded prompts, particularly frequent questions during play, may have unintended negative effects on play and social behaviour for some children with autism. Practitioners and parents should be cautious about over-using questions and should consider alternative strategies such as following the child’s lead, using comments, and modelling language without placing constant demands.

Future research should examine which types of questions are most likely to be aversive, how question difficulty interacts with the child’s language skills, and how to best balance questions, comments, and nonverbal interaction in naturalistic teaching procedures. It would also be useful to investigate how to teach children with autism to respond appropriately to comments, not only to questions, in order to support more natural conversations with peers.

References

The full reference list is available in the original thesis document. Key authors cited include Hart & Risley, Heal & Hanley, Kaiser et al., Koegel et al., Carbone et al., Sherer & Schreibman, Sundberg, and others working in the field of naturalistic language teaching and autism.

Note: Appendices (e.g., detailed response definitions, list of toys, sequence tables, consent form) can be added on this page if you wish, but they have been omitted here for readability on the web.