teaching children with autism three different questions

43
TEACHING CHILDREN WITH AUTISM THREE DIFFERENT QUESTIONS Heather Cramer, B.A. Thesis Prepared for the Degree of MASTER OF SCIENCE UNIVERSITY OF NORTH TEXAS December 2003 APPROVED: Sigrid Glenn, Major Professor Janet Ellis, Committee Member Manish Vaidya, Committee Member Sandra L. Terrell, Interim Dean of the Robert B. Toulouse School of Graduate Studies

Upload: others

Post on 03-Feb-2022

1 views

Category:

Documents


0 download

TRANSCRIPT

TEACHING CHILDREN WITH AUTISM THREE DIFFERENT QUESTIONS

Heather Cramer, B.A.

Thesis Prepared for the Degree of

MASTER OF SCIENCE

UNIVERSITY OF NORTH TEXAS

December 2003

APPROVED: Sigrid Glenn, Major Professor Janet Ellis, Committee Member Manish Vaidya, Committee Member Sandra L. Terrell, Interim Dean of the Robert B.

Toulouse School of Graduate Studies

Cramer, Heather, Teaching Children with Autism Three Different Questions.

Bachelor of Arts (Psychology), May 1995, 39 pp., 5 tables, 3 figures, 14 References

Children with autism often exhibit deficits in question-asking. This study

replicated and extended Williams, Donley, and Keller�s (2000) training package: a

modeling and reinforcement procedure to teach the use of 3 different questions about

hidden objects. Two boys, aged 13 and 12, with primary diagnoses of autism,

participated. A multiple baseline design across questions was used. Both children learned

to ask all three questions: �What�s that?� �Can I see (item name)?� and �Can I have (item

name)?� Question-asking generalized to novel locations, people, and stimulus materials

with minimal additional training. These results support the efficacy of this training

package as an efficient way to teach children with autism to ask questions about objects

in their environment.

ii

ACKNOWLEDGMENTS

I would like to thank Sigrid Glenn and Janet Ellis for their help in all

aspects of this study. I would also like to thank Julio Ruiz for his help in training

the participants and running probe sessions. Finally, I would like to thank the

children who participated in this study.

TABLE OF CONTENTS

Page

ACKNOWLEDGMENTS .................................................................................................. ii

INTRODUCTION ...............................................................................................................1

METHODS ........................................................................................................................12

RESULTS ..........................................................................................................................20

DISCUSSION....................................................................................................................23

FIGURES...........................................................................................................................29

TABLES ............................................................................................................................32

REFERENCES ..................................................................................................................38

iii

1

INTRODUCTION

Conversation is an integral part of human interaction. Minkin et al. (1976)

reported that ratings of conversational skill for typical adolescents and college students

were correlated (.70) with the number of questions they asked. Furthermore, asking

questions yields important information about the environment. Paul (1985) suggested that

such normative data are valuable in identifying specific skill deficits and developing

targets for acquisition. This seemed to be the case in Minkin et al.�s (1976) study,

because teaching delinquent adolescents both to ask questions and to provide feedback

increased ratings of their conversational skill. Children with autism often exhibit

significant language delays (American Psychiatric Association Diagnostic and Statistical

Manual of Mental Disorders (4th ed.), 1994). They frequently exhibit specific deficits in

question-asking (Charlop & Milstein, 1989; Koegel, 1995). As suggested by Minkin et

al.�s study, this likely contributes to conversational deficits and ultimately to their ability

to interact successfully with adults and peers. According to Hung (1977), Postman

(1968), and Rimland (1964), generalization of skills is often a significant deficit for

children with autism. Therefore, if children with autism are to acquire generalized

question-asking, generalization of this skill may need to be specifically programmed.

An emerging body of experimental literature addresses these areas. Taylor and

Harris (1995) used a time delay procedure with modeling and reinforcement to teach

three children with autism to ask, �What�s that?� when presented with novel stimuli

during a picture-labeling task. Time delay procedures involve systematically delaying the

2

amount of time before providing a verbal prompt. As the time until a prompt increases,

answering before the prompt becomes more likely as this produces reinforcement more

quickly. The researchers performed three different experiments. All three experiments

used a multiple baseline design across the three participants: Two 9-year-old children and

one 5-year-old child. Prior to baseline, pre-testing was conducted to generate a pool of

known and unknown stimulus items, and the children were pre-taught a labeling task

requiring them to point to each of four pictures on the table, left to right, as they labeled

each one.

During Experiment 1 the children were taught to ask, �What�s that?� when

encountering one unknown picture placed in the field with three known pictures. The

question was modeled immediately during initial training. Then the time before modeling

was systematically increased by 2 s following trials with accurate responses. Once

previously unknown stimuli were labeled correctly, they were removed from the set.

Generalization probes were conducted in the school kitchen, with a novel person, using

objects rather than pictures. All three children rapidly acquired question-asking in the

presence of unknown pictures, which generalized to the novel conditions. Experiment 2

included pre-and post-tests of expressive and receptive labels. All three participants were

able to learn new labels after hearing these labels stated in response to, �What�s that?�

Finally, in Experiment 3, objects were randomly placed throughout the school building.

Questions were taught using the same time delay procedure as in Experiment 1.

3

Participants learned to ask, �What�s that?� with no further prompting when required to

pause in front of unknown items.

Modeling and reinforcement procedures that do not involve time delay have also

been used to teach children to ask questions. Koegel, Camarata, Valdez-Menchaca, and

Koegel (1998) used modeling and reinforcement procedures to teach three children with

autism to ask the question, �What�s that?� These researchers also used a multiple baseline

design across participants: two 3-year-old children, and one 5-year-old child. Baseline

data on the frequency of asking, �What�s that?� were collected both at the clinic and in

the child�s home. Following pre-teaching of the imitative response, �What�s that?� the

children participated in two 30-min sessions per week at the clinic. During initial training

sessions, highly preferred items, chosen based on pre-treatment videos, were hidden in an

opaque bag. The question, �What�s that?� was modeled, and the experimenter reinforced

imitative or independent responses by providing the label of the item and access to the

item. As the children began to spontaneously ask the question, the experimenter began to

replace highly preferred with neutral items. Finally, the bag itself was faded. In the final

stages of training, unhidden neutral items were placed on the table in the training room.

The experimenters measured both frequency of question-asking and labeling of stimulus

items pre- and post-intervention. All three children learned to ask this question,

maintained this skill, and asking the question generalized to a new setting (home) and to

new stimulus items. Furthermore, the children acquired an average of six new labels per

week during this training.

4

Hung (1977) taught four children with autism, aged 8, 9, 10, and 11, to ask the

questions, �What is �for?� and �What is �.doing?� using modeling and reinforcement.

This researcher collected pre- and post-intervention data for each participant. During

baseline instances of these questions were consequated with praise and tangible

reinforcers, but no prompting was provided. The study was conducted during a 3-week

summer camp. Following baseline 45-min training sessions were conducted daily.

Specific stimuli were used during training. Three picture cards were trained individually

to criterion, then gradually pictures were randomized, and finally pictures, live actions,

and objects were randomly presented. Generalization training was also provided. The

participants were brought to the door of a �shop� adjoining the classroom. Questions

were reinforced with a token and the answer to the question. Prompting involved pointing

to an object and then modeling a question about that object. If questions were not asked

spontaneously in the �shop� and prompting techniques were ineffective in this setting,

classroom training was continued. If question-asking did generalize to this setting, this

procedure was repeated in the gymnasium and along a wooded trail. All participants

learned to ask questions in the classroom and in generalization training sessions, although

the frequency of asking questions varied. Raising the economic value of tokens

contingent on asking questions in non-training time increased question-asking during

those times. This study underscores the fact that a powerful reinforcer is necessary for

both the maintenance and the generalization of question-asking.

5

Yamamoto�s (1987) study suggested one variable that may affect whether or not

children attend to the answers that are provided to their questions. He taught three

children with autism to say, �I don�t know. Let me know,� in response to unknown

stimuli. All three children acquired the ability to make this request. When reinforcement

was made contingent on using the information provided in response to the child�s request,

two of the children learned to use the information and all three children�s question-asking

generalized to new stimuli and settings. This study highlights the importance of

functionality in the maintenance and, possibly, in the generalization, of any skill.

Esbenshade and Rosales-Ruiz (2001) made reinforcement contingent on

answering questions throughout the course of their study. They used modeling and

reinforcement to teach a 5-year-old boy with autism to: 1) ask the question, �What is

that?� and 2) to use the provided information to answer questions regarding an item�s

label. Following pre-testing to generate a pool of known and unknown stimulus items,

these experimenters collected baseline data on the child�s performance on five different

tasks in two different settings. During training, the child participated in five 10-trial

sessions each day. Two pairs of items, each containing one known and one unknown

item, were presented each session. The question, �What is that?� was modeled in the

presence of unknown items. If the participant imitated or spontaneously asked the

question, the experimenter provided the label of the item. The participant was then

immediately asked, �What is this?� about the same item. Praise was contingent on

correctly labeling the item.

6

When the child labeled previously unknown items three consecutive times, two

new pairs were presented. This procedure continued until he independently asked, �What

is that?� on the first presentation of two consecutive unknown stimuli. The participant

learned to ask, �What is that?� in the presence of unknown stimuli and not to ask in the

presence of known stimuli. Additionally, the participant learned 18 new labels during the

course of the study.

In the Esbenshade and Rosales-Ruiz study, generalization of question-asking was

assessed more thoroughly and systematically than in prior studies. First, generalization of

question-asking was assessed using novel stimuli within the training task. Then

generalization was assessed with a different person in a different setting.

Generalization was also assessed using four novel tasks. Asking the question,

�What is that?� upon encountering an unknown stimulus item generalized to the

following tasks: 1) a labeling task in which four items were placed on the table and the

child was asked to tell the experimenter what he saw, 2) a task in which pictures were

spread out in front of the participant and he was asked to, �Pick one,� and then was

expected to label that item, and 3) a task in which an item was held up in front of the

participant as the experimenter said, �Look (name of child).� Within these three tasks,

question-asking also generalized when person and setting were simultaneously changed.

Question-asking did not generalize when the participant was handed an unknown

item and asked to either, �Give this to (person),� or �Put this on (location).� The

experimenters used three techniques to train generalization of question-asking in this

7

final task. First, they used stimuli from the training task. In other words, the participant

was handed an item that had been used during the training task, and instructed to, �Give

this to (person),� or, �Put this on (location).� This technique did not enhance

generalization. Next they inserted the question, �What is that?� before giving the

instruction. In other words, the experimenter 1) held up an item, 2) asked, �What is that?�

and then immediately, 3) instructed the participant to either, �Give this to (person),� or,

�Put this on (location).� This technique was also ineffective. Finally, the experimenters

asked, �What is that?� after giving the instruction, but before the participant began to

respond. This was somewhat effective promoting generalization. Following the use of

this technique, the participant learned to ask, �What is that?� when the experimenter

presented this task, but not when his caregiver did.

Le (2002) compared vocabulary acquisition for four children with autism when

labels were provided in response to self-initiated questions versus when labels were

provided in the absence of any question. For three of the four children, vocabulary

acquisition was faster for labels that were provided in response to their questions.

In summary, researchers have examined a variety of procedures to teach children

to solicit the names of objects in their environment by asking such questions as, �What�s

that?� �What is� for?� and �What is�doing?� or by responding to a mand to emit an

unknown response with, �I don�t know. Let me know.� Most researchers used multiple

baseline designs across participants. Participants have ranged in age from 3 to 11. Time

delay, modeling and reinforcement have been used in various combinations to

8

successfully teach mands for labels. Several studies have found increases in vocabulary

resulting from these requests for labels.

Researchers generally have taught these questions in contrived settings, although

Koegel et al. (1998) were able to teach question-asking in a more naturalistic setting.

While this may have aided generalization, this setting may have been less efficient in

terms of the number of learning opportunities that could be presented. On the other hand,

the use of very specific and limited stimuli, as in Hung�s (1977) study may impede

generalization. Generalization of question-asking has been examined using a combined

change in person and place (Esbenshade & Rosales-Ruiz, 2001), place, person, and

stimuli (Taylor and Harris, 1995), change in setting (Koegel et al., 1998; Hung, 1977),

change in stimuli (Koegel et al., 1998; Yamamoto, 1987; Esbenshade & Rosales-Ruiz,

2001) and across tasks (Esbenshade & Rosales-Ruiz, 2001). However, generalization

across all person, place, and stimuli has not been systematically examined. Additional

training to promote generalization has been required in some situations.

Williams, Donley, and Keller (2000) extended this literature by teaching two 4-

year-old children with autism three functionally distinct questions about objects that were

out of their view: �What�s that?� �Can I see it?� and �Can I have it?� using a treatment

package involving multiple presentations of hidden objects, modeling, and reinforcement.

Because they were teaching multiple questions, these investigators were able to use a

multiple baseline design across questions. This allowed more definitive statements to be

made about the effectiveness of the teaching package for each individual participant.

9

Baseline and training sessions were conducted in one room of the children�s

homes. A variety of toys were hidden, one each in many small boxes. The experimenter

sat across from the child and opened the boxes one at a time. If no questions were asked,

the experimenter modeled, �What�s that?� Imitation or spontaneous questions were

reinforced with the label of the item and access to the item within the box for 20 s. When

the question was correctly imitated on two consecutive sessions, the prompt was faded to

a partial verbal prompt. For example, the prompt was first faded to, �What�s?� and then

to �Wh.� When a participant asked the question unprompted within 3 s of box

presentation on all trials presented in two consecutive sessions, training of the next

response form, �Can I see it?� was introduced. In this condition, �What�s that?� was

followed by the experimenter saying the name of the toy. If the participant asked no

additional questions, the experimenter then modeled, �Can I see it?� If the participant

correctly imitated or spontaneously asked, �Can I see it?� she was given access to the

item. Finally, the response form, �Can I have it?� was trained. In this condition, the

participant was told the item�s name if she asked, �What�s that?� and was shown the item

if she asked, �Can I see it?� She was given the item only if she correctly imitated or

spontaneously asked, �Can I have it?� Prompt fading in these two conditions was

implemented in the same way as in training, �What�s that?�

Generalization probes were conducted immediately after each session in the living

room with the child�s mother. In other words, setting and person were both different from

training conditions during generalization probes. Box presentation and contingencies of

10

reinforcement were identical to training sessions. However, prompting never occurred in

the generalization probes. Both children successfully learned to ask all three questions

and this skill generalized across listener and setting when they asked their mothers the

questions in a different room of their homes.

The current study seeks to extend this growing body of literature. Training

procedures replicated those of Williams, Donley, and Keller (2000), thus examining

whether Williams, Donley, and Keller�s (2000) results will extend to older children in a

center-based program for children with autism. As in the previous study, the current

training package included modeling and reinforcement, but did not use time delay. As the

acquisition of three different questions was targeted, the use of a time delay procedure

might increase the probability of response errors (e.g., asking to see the item before

finding out what it is). These types of errors could be more likely because questions

taught in the present study, �Can I see (object name)?� and �Can I have (object name)?�

are topographically similar.

To ensure that the identity of the items was unknown, items were hidden within

boxes or bags throughout the course of the study. Even if the participants knew the names

of some items, they would not know which item was hidden on any given trial.

Therefore, it was necessary to ask the question, �What�s that?� to obtain the name of the

item on every trial, regardless of prior knowledge. Hiding the objects also safeguarded

against differential responding based on individual �values� of each item for the

participant. In other words, the participant could not predict, from trial to trial, what item

11

was hidden and, therefore, the value of learning the object�s label should have been

constant across trials.

As in Williams, Donley, and Keller�s (2000) study, a multiple baseline design

across questions was used. Williams, Donley, and Keller�s (2000) treatment package was

modified in two important ways for the current study. First, to maintain the reinforcing

value of information obtained from asking, �What�s that?� once the other questions were

introduced, children in this study were trained to use this information (i.e., the object

name) to ask the next question, �Can I see (object name)?� Second, generalization probes

were conducted before training sessions each morning and systematically assessed

generalization to different settings, novel containers, and novel adults.

12

METHODS

Participants

Two male students, ages 13 and 12, with primary diagnoses of autism, attending a

center-based program for children with autism spectrum disorders, participated in this

study.

Apparatus

All baseline and teaching sessions took place in each student�s classroom at the

center. Generalization probes also took place at the center, but occurred in other rooms

throughout the building.

Sixty small boxes were used with each participant. Each box varied from the other

boxes on at least one dimension (e.g., size, shape, color, decorations), and was labeled

with a piece of masking tape. A handwritten number between 1 and 60 was on the bottom

of each box. To determine the order of presentation for the boxes, the primary

investigator placed 60 typed numbers in a bowl and drew out one at a time until all

numbers had been selected. This procedure determined the order of presentation for the

next three sessions. One toy (e.g., a miniature Etch-a-Sketch®) was hidden in each box

presented. A total of 76 toys were rotated among boxes for Participant 1, and a total of 97

toys were rotated among boxes for Participant 2. No toy was placed in the same box more

than once for either participant during the study. During baseline and teaching sessions,

these small boxes were placed in a plastic box designed to hold hanging files. A brown

paper bag was placed behind this box to hold the small boxes once they had been

13

presented. These same boxes and toys were used for generalization probes throughout

baseline and all training conditions. Novel boxes were used for some of the post-training

generalization probes. Westbend® timers were used to time session length and to cue the

experimenter.

Procedure

During all baseline and teaching sessions, the experimenter sat across from the

participant at his desk. The hanging file box, filled with 20 small boxes arranged in the

order in which they were to be presented, and the brown paper bag were placed to the

right of the experimenter. The data sheet (see Figure 1 for sample) and the timer were

placed on the desk in front of the experimenter. At the beginning of the session the

experimenter made a general statement, �We�re going to look at some boxes,� or �Let�s

look at some cool things.� Sessions lasted for 10 min or until the experimenter had

presented 20 boxes, whichever came first.

The experimenter presented boxes one at a time, in the order written on the data

sheet. As he opened the box he commented on its contents in an enthusiastic tone of

voice with statements such as, �This is a great one,� or �Wow!�

Baseline

During baseline sessions, the box remained open in the experimenter�s hands for

10 s or until the participant asked a question. If no question was asked, the box was

placed in the brown paper bag and the next box presented. If the participant asked a

14

question (e.g., �Can I have it?�) he was told, �Maybe some other time,� and the next trial

began. Data were collected on the number and type of questions asked.

Teaching

The treatment package consisted of: 1) presenting items hidden within a variety of

containers one at a time; 2) modeling target questions about the item within the container;

3) reinforcement of correct imitations or independently asked questions; and 4) prompt

fading.

Teaching the response form, �What�s that?�. During these sessions, the

participant was given 5 s from the time the box was opened to ask an independent

question. If no question was asked, or a question other than �What�s that?� was asked, a

correct response was prompted using the attached prompt sequence (see Table 1).

Functionally equivalent questions that were full sentences (e.g., �What is it?�) were

counted as correct responses. When the participant asked, �What�s that?� prompted or

unprompted, he was given the item for 20 s. At the end of 20 s, the experimenter said,

�My turn,� and held out his hand for the object. When the object was returned, it was

placed in the brown paper bag and the next trial began. This condition was continued

until the participant asked the question, �What�s that?� independently on 90% of box

presentations across two consecutive sessions.

Teaching the response form, �Can I see (object name)?�. In these sessions, if the

participant asked, �What�s that?� he was told the name of the item, but was not handed

the item. If he did not ask, �What�s that?� this question was prompted. The experimenter

15

then waited an additional 5 s, or until the participant asked another question. If there was

no response or an incorrect response, the experimenter prompted the participant to say,

�Can I see (object name)?� using the attached prompt sequence (see Table 2). When the

participant asked, �Can I see (object name)?� prompted or unprompted, he was given the

item for 20 s. This condition was continued until the participant asked both questions

independently on 90% of box presentations across two consecutive sessions.

Teaching the response form, �Can I have (object name)?�. In these sessions,

�What�s that?� was answered with the object�s name. The experimenter responded to the

question, �Can I see (object name)?� by showing the object, but not letting the participant

hold the object. At this point, the experimenter waited an additional 5 s or until the

participant asked another question. If necessary, the correct question was prompted using

the attached prompt sequence (see Table 3). If the participant asked, �Can I have (object

name)?� prompted or unprompted, he was given the item for 20 s. This condition was

continued until the participant asked all three questions independently on 90% of box

presentations across two consecutive sessions.

Errorless teaching sessions. In these sessions, the experimenter presented the

boxes and waited for the participant to ask, �What�s that?� He then provided the name of

the item and immediately prompted, the whole question, �Can I see (object name)?�

When the participant imitated the questions, he was shown the item as the experimenter

immediately prompted the next whole question, �Can I have (object name)?� The full

16

prompt was used until the participant correctly imitated before making any alternative

response. At this point prompt fading was implemented as in the other training sessions.

Generalization Probes

Unless otherwise specified, modeling of responses never occurred during

generalization probes. Questions were reinforced as they had been in the current training

sessions. Throughout all training conditions, generalization probes were conducted each

day immediately before the first training session. For Participant 1, these probes were

initially office probes. Because generalization of question-asking did not occur across

multiple office probes for Participant 1, all remaining training condition generalization

probes for Participant 1 and all training condition generalization probes for Participant 2

were only experimenter present probes. As these latter probes more closely approximated

training conditions, it was hypothesized that generalization would be more likely.

Office probes. In these probe trials, the experimenter placed a small box on a desk

in one of seven offices at the center Then the participant was taken to the office, where

one to three adults were present. The experimenter guided the participant to a spot one ft

from the box and waited 10 s or until a question was asked. Questions about the box were

reinforced as they had been in the current training sessions. If a question was not asked

within 10 s, the participant was taken back to his classroom.

Only experimenter present probes. As for office probes, the experimenter placed

one of the boxes in one of the two rooms. One of them contained only tables and chairs

(empty room). The other room contained a table, chairs, two computers, a desk, many

17

books, journals, and toys (busy room). The participant was guided to a spot one ft from

the box (always in a designated location). The experimenter then waited 30 s or until a

question was asked. If the participant asked a question within those 30 s, the question was

reinforced as it would be in training sessions. The experimenter then waited an additional

10 s or until an additional question was asked. If no question was asked in the initial 30 s,

or if the participant failed to ask an additional question within the subsequent 10s, the

participant was taken back to his classroom.

Post-training probes. When the participants exhibited generalization of all three

questions in both rooms described above, conditions were systematically varied with one

element being changed at a time. Participant 1 took part in 1 post-training probe per day

in the order described below. One aspect of the probe condition (e.g., person, location,

box, setting) was changed at a time. If the participant asked questions in a given probe

type, question-asking was reinforced and the next probe involved a change to another

aspect of the probe condition. If the participant did not ask questions in a given probe

type, the conditions in the next probe more closely resembled those in the last probe type

in which question-asking had occurred.

First, an adult other than the experimenter brought the participant to the rooms

(novel person probes). If generalization did not occur with that adult, the experimenter

and the novel person brought the participant to the room (novel and training person

probes) before returning to the novel person probe. When generalization occurred with a

novel adult, the location of the box in each room was varied (novel location probes). If

18

the participant did not ask questions about the boxes in novel locations, the box was

moved to a new location on the same table as the initial probes (novel placement probes).

At this point, novel boxes were placed in the room (novel boxes probes). Then boxes

were placed in different rooms in the building (empty office probes).

Finally, an additional adult was sitting in the original generalization probe rooms

when the participant entered. In one probe condition, the adult held the box (held box

probes), while in another probe condition, the box was placed in plain view on the desk

(box on desk probes). During box on desk probes, Participant 1 consistently greeted the

adult in the office and did not appear to notice the box until after the probe had ended. To

test whether he would be more likely to attend to the box in the absence of direct adult

attention, an additional generalization probe was conducted, in which the adult in the

office was talking on the phone when he entered (phone probe).

Participant 2 took part in one to six post-training probes per day. Participant 2

completed the following probes on the day after he completed training: held box probe,

novel person probe, novel box probe, and box on desk probe. These probes were to assess

the extent to which question-asking had generalized across settings, people, and stimuli,

without systematic reinforcement of gradual changes in probe conditions (as described

above for Participant 1). He then completed probes in which one aspect of the probe

condition was changed at a time.

19

Data Collection

During baseline, teaching, and generalization sessions, the experimenter collected

data on independent responses, prompted responses, and non-responses for each trial. A

second observer collected reliability data, and reliability scores were calculated as: (total

identically recorded questions/ total possible questions) X 100. Data also were collected

on procedural reliability, including 1) timing of container presentation, 2) experimenter

wording, 3) prompting and fading, and 4) timing of participant access to boxes and toys.

Examples of procedural deviations include using a full verbal prompt when the procedure

dictates use of a partial prompt or allowing the participant access to the item for shorter

or longer than 20 s. Procedural reliability was calculated as: (total procedural

deviations)/(total trials) X 100.

20

RESULTS

Reliability

The second observer collected data on questions asked during 63% of baseline

sessions, 43% of �What�s that?� training sessions, 36% of �Can I see?� training sessions,

and 38% of �Can I have?� training sessions. A second observer also collected data during

42% of generalization probes. The average reliability score was 100% during baseline,

97% during �What�s that?� training sessions (range 93% to 100%), 100% during �Can I

see?� training sessions, 100% during �Can I have?� training sessions, and 99% during

generalization probes (range 67% to 100%).

Procedural reliability data were collected during 39% of sessions. The average

procedural reliability score was 97% (range 75% to 100%).

Training

Both participants learned to ask all three questions with at least 90% accuracy

(see Figures 2 and 3). Participant 1 learned to ask, �What�s that?� after 18 training

sessions, �Can I see (object name)?� after eight additional training sessions, and �Can I

have (object name)?� after three further training sessions. Participant 2 learned to ask,

�What�s that?� after three training sessions, and �Can I see (object name)?� after three

additional training sessions. When training the question, �Can I have (object name)?� was

introduced, Participant 2�s accuracy in asking the, �Can I see (object name)?� decreased.

At that point the errorless training procedure was introduced. Following an additional six

21

training sessions, Participant 2 relearned the question, �Can I see (object name)?� as well

as the question, �Can I have (object name)?�

Generalization During Training Condition

Asking the questions as they were being trained generalized for both participants

(see Figures 2 and 3) to probe conditions. For Participant 1, asking �What�s that?� and,

�Can I see�?� generalized to the empty room at approximately the same time these

questions were acquired during training sessions (sessions 17 and 28). Asking, �What�s

that?� generalized to the busy room only after an additional question (�Can I see�?�)

was acquired. Thereafter, both questions were asked in the busy room probe (session 30).

Asking, �Can I have?� generalized quickly to the busy room, and was asked in the empty

room during the first post-training probe. Participant 2 asked all questions in both empty

room and busy room, but question-asking in both of these rooms was variable throughout

the training period.

Post Training Generalization

Table 4 presents trial-by-trial data for Participant 1 of generalization trials

conducted after training was complete. This participant asked all three questions in 8 out

of the 12 post-training probe trials. More specifically, Participant 1 asked all questions

during the novel person probe (trial 1), and the novel location probe in the empty room

(trial 3). Initially he did not ask any questions during the novel location probe in the busy

room (trial 2). However, following a novel placement probe in that room (trial 4), during

which he asked all three questions, he did ask questions during a novel location probe

22

(trial 5). Further, he asked all questions during the novel box probes in both busy room

(trial 6) and empty room (trial 7), empty office probe (trial 8), and held box probe (trial

9). However, this participant did not ask any questions during two box on desk probes

(trials 10 and 11), or in the phone probe (trial 12). Anecdotally, while participating in an

unrelated experiment, this student asked, �What�s that?� when presented with a puzzle

piece for which he did not know the label.

Table 5 presents trial-by-trial data for Participant 2 of generalization trials

conducted after training was complete. Participant 2 asked all three questions unprompted

in 10 of the 21 post-training trials, including the final seven trials. Initially, Participant

2�s unprompted question-asking generalized only during the box on desk probe (trial 4).

During the held box probe (trial 1), he grabbed the box from the adult who was holding it

and saw the item. Therefore, he was able to ask, �Can I see (item)?� and �Can I have

item?� without asking, �What�s that?� Following prompting to ask questions during the

novel and training person probes (trial 11), the novel person probe (trial 12), the novel

location probe (trial 13), and the novel box probe (trial 14), this participant�s question-

asking generalized during all office probes (trials 15-17), and during the following probe

conditions in empty room: novel and training person probe (trial 18), novel person probe

(trial 19), novel location probe (trial 20), and novel box probe (trial 21).

23

DISCUSSION

These results indicate that the teaching package, including presentation of hidden

objects, prompting, and reinforcement, used by Williams, Donley, and Keller (2000)

could be successfully replicated with older participants in a center-based program for

children with autism. Both participants learned to ask three questions about objects in

their environment and this skill generalized across settings, people, and stimulus items.

This treatment package is an efficient, relatively easy way, to help children with

autism gain important skills. Asking, �What�s that?� may enable children with autism to

gain information about items in the environment as typical children do, by questioning

adults and other children around them. Taylor and Harris� (1995) results indicate that

labels can be acquired in this way. Asking, �Can I see/ have (item name)?� is a socially

appropriate way of gaining access to desired items in the environment and may replace

other, less desirable, means of obtaining items.

These results extend the findings of Williams, Donley, and Keller (2000) in that

they provide a closer examination of generalization of skills learned in this teaching

format. Generalization probes here occurred before, rather than immediately after,

teaching sessions each day providing a more stringent test of generalization. The present

study also separately examined generalization across people, settings, and stimulus items.

Finally, generalization results from this study suggest two possible methods for training

generalization when it does not occur.

24

Participant 1�s asking questions generalized to novel boxes placed in both empty

room and busy room, and also to a novel person and a novel location when boxes were

placed in a room with minimal distractions (empty room). In a room with greater

potential distractions (busy room), generalization to a new location in the room did not

occur immediately. However, when a box was placed in a novel location closer to the

initial location, Participant 1�s asking questions did generalize. Following reinforcement

of those questions, generalization also occurred when the box was again placed further

away. This suggests one procedure for enhancing generalization. If each generalization

probe differs minimally from the one before, generalization should be more likely. If

question-asking occurs and is reinforced in a situation, success may be more likely in

another situation if the change to another setting, person, or stimulus item is minimal.

However, the effectiveness of this technique would need to be more systematically

explored to make any definitive statement. As there were not repeated probes within each

room, it is possible that variations in activity level in the busy room affected

generalization rather than reinforcement of responses in different locations.

Participant 1�s question-asking never generalized in the box on desk probes. It

appeared that he was so focused on the person in the room that he did not even notice the

box until after the 30 s had elapsed. Previously this participant previously had received

extensive training in responding to people, rather than items, when entering a room. The

phone probe was intended to shift focus away from the person in the office, but

generalization was not exhibited in this probe condition either.

25

For Participant 2, the procedure of minimally changing each generalization probe

condition did not appear to be effective overall. He required prompting to ask questions

during some generalization conditions. Prior to prompting, generalization of all three

questions occurred only in the box on desk probe. There was also generalization of the

questions, �Can I see (object)?� and �Can I have (object)?� in the held box probe. At the

beginning of this probe, the participant grabbed the box out of the adult�s hand and was

able to see what the item was. Therefore, it was not necessary for him to ask the question,

�What�s that?� to obtain the name of the item. Perhaps the fact that the experimenter held

the box during training sessions was the most salient feature for this participant. If that

were the case, the held box probe would most closely approximate training conditions.

The box on desk probe, conducted later that day was a minor variation of the held box

probe, so reinforcement of questions asking in held box probe may have made question

asking in the box on desk probe more likely.

Failure to generalize to a novel person, prior to prompting, may have been a result

of the fact that the participant had a longer learning history with the experimenter. His

presence may have exerted unintended stimulus control over his question asking.

For the remaining generalization probes, there was no clear pattern of

generalization. It appears that during some generalization probes momentary

contingencies may have exerted a more powerful influence than did the programmed

contingencies. For example, on the day that probes indicated with ** (see table 5) were

conducted, the participant asked to run at the start of the probe session. Running was

26

generally a powerful reinforcer for this child. He was told that he could run after he took

a walk with the experimenter. Therefore, it is likely that escape from the probe sessions

on this day was a powerful reinforcer. In this scenario, initial responding may not have

occurred because access to the hidden items was not an effective reinforcer at that time.

Following prompting, however, question asking was at strength. If escape from the task

was the only reinforcer at this point, question asking, which required minimal effort

following prompting, may have occurred because it decreased the length of the task.

Alternately, Participant 2�s consistent use of questions in the latter probe

conditions may be a result of the fact that prompting in the earlier probes effectively

taught the skill across a variety of conditions. This training across conditions may have

enhanced the likelihood of generalization to novel conditions.

During training, Participant 2 had some difficulty learning the distinction

between, �Can I see (item)?� and �Can I have (item)?� Perhaps this related to the fact that

he acquired the response, �Can I see (item)?� in only three trials. As he had very little

time to practice asking this question before, �Can I have (item)?� was introduced both

acquisition and later generalization may have been impacted.

It is likely that this difficulty was related to a more global issue pertaining to how

the questions were taught. In William, Donley, and Keller�s (2000) study, frequency of

asking, �Can I see it?� decreased for both participants following the introduction of

training, �Can I have it?� For one of the participants, frequency of asking, �What�s that?�

also decreased following the introduction of the other two questions. This highlights the

27

fact that their procedure did not clearly differentiate the function of each question. In

other words, once �Can I see it?� was acquired, the participant had no independent

motivation to ask, �What�s that?� At this point, the participant could obtain the item by

only asking, �Can I see it?� Similarly, once �Can I have it?� was acquired, the participant

had no independent motivation to ask, �What�s that?� or �Can I see it?� as the item could

be obtained by simply asking, �Can I have it?� Assuming that obtaining the item was the

reinforcer for question-asking (rather than obtaining the label of the item or seeing the

item), is understandable that the participants would expend the least amount of effort (ask

the least number of questions) that would lead to the reinforcer (access to the item).

The present study partially addresses this limitation, but the lack of distinction

remains for two of the questions. The item name was required in order to ask, �Can I see

(item name)?� or �Can I have (item name)?� Therefore, asking, �What�s that?�

maintained an independent function even after the other questions were acquired. On the

other hand, �Can I see (item name)?� lost independence of function once the final

question was learned. During training sessions, the participant was required to ask all

questions in order. However, seeing the item without gaining access to it neither appeared

to be a reinforcer, nor provided the learner with usable information.

Future studies might vary the value, for each individual participant, of the hidden

items. For instance, if a participant found touching certain textures reinforcing and

touching other textures unpleasant, items might vary along this dimension. With this

modification, seeing the item would provide information to the participant. Given that

28

information, s/he could ask (or not) to have the item based on whether or not obtaining

the item would be desirable. This modification might also be beneficial in that it would

be less likely to create chained responding.

Future research might also address generalization of question-asking to items in

the absence of a box or other container. Anecdotal data indicated this type of

generalization for the question, �What�s that?� for Participant 1, but this needs to be more

systematically explored.

In summary, the teaching package was effective in teaching two students to ask

three different questions. Minimal additional training was required to produce

generalization of asking these questions across people, location, and stimulus items.

Further research is needed to separate the functions of the three questions learned by the

participants in this study.

29

Figure 1

Sample Data Sheet: Question-asking

Date: _______ Time: _______

Observer (circle one): Primary/ Secondary

Current training target (circle one): Baseline/ �What�s that?�/ �Can I see it?�/ �Can I have it?�

Box presentations

Box # Item Name

�What�s that?� �Can I see (object)?�

�Can I have (object)?�

39 Flute 59 Rocks 47 Snakes 48 Kaleidoscope 14 Teeth 38 Smile balls 33 Yo-yo 34 Round toy 51 Horse 24 Football

cards

45 Frogs 8 Bird

42 Cat mask 29 Balloons 6 Shape puzzle

60 Smiley toy 11 Puzzle 52 Whistle 20 Calculator 54 Phone

30

Figure 2

Training Data: Participant 1

"What's that?"

0%

20%

40%

60%

80%

100%

% C

orre

ct

% Correct Questions

Office

Empty Room

Busy Room

Baseline Intervention

"Can I see (item)?"

0%

20%

40%

60%

80%

100%

% C

orre

ct

"Can I have (item)?"

0%

20%

40%

60%

80%

100%

1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 33 35

Sessions

% C

orre

ct

31

Figure 3

Training Data: Participant 2

"What's that?"

0%

20%

40%

60%

80%

100%

% C

orre

ct

% Correct Questions

Empty Room

Busy Room

Baseline Intervention

"Can I see (item)?"

0%

20%

40%

60%

80%

100%

% C

orre

ct Errorless

"Can I have (item)?"

0%

20%

40%

60%

80%

100%

1 4 7 10 13 16 19 22 25

Sessions

% C

orre

ct

32

Table 1

Prompt Sequence: Teaching the Response Form, �What�s that?�

Response Prompt

No response for 5 s or incorrect response Model full question: �What�s that?�

Correct Imitation on 2 consecutive trials Fade prompt: �What�s th�?�

Correct Imitation on 2 consecutive trials Fade prompt: �What�s�?�

Correct Imitation on 2 consecutive trials Fade prompt: �Wh�?�

Correct Imitation on 2 consecutive trials Fade prompt: Form �w� sound with mouth,

but no sound

Correct Imitation on 2 consecutive trials Fade prompt: expectant look

Correct independent response No prompt

Incorrect response on 3 consecutive trials Return to the previous prompt level

33

Table 2 Prompt Sequence: Teaching the Response Form, �Can I see (object name)?�

Response Prompt

No response for 5 s or incorrect response Model full question: �Can I see (item

name)?�

Correct Imitation on 2 consecutive trials Fade prompt: �Can I see (first letter of

object name)?�

Correct Imitation on 2 consecutive trials Fade prompt: �Can I see�?�

Correct Imitation on 2 consecutive trials Fade prompt: �Can I s�?�

Correct Imitation on 2 consecutive trials Fade prompt: �Can I�?�

Correct Imitation on 2 consecutive trials Fade prompt: �Can�?�

Correct Imitation on 2 consecutive trials Fade prompt: �C�?�

Correct Imitation on 2 consecutive trials Fade prompt: Form �c� sound with mouth,

but no sound

Correct Imitation on 2 consecutive trials Fade prompt: expectant look

Correct Independent Response No prompt

Incorrect response on 3 consecutive trials Return to the previous prompt level

34

Table 3

Prompt Sequence: Teaching the Response Form, �Can I have (object name)?�

Response Prompt

No response for 5 s or incorrect response Model full question: �Can I have (item

name)?�

Correct Imitation on 2 consecutive trials Fade prompt: �Can I have (first letter of

object name)?�

Correct Imitation on 2 consecutive trials Fade prompt: �Can I have�?�

Correct Imitation on 2 consecutive trials Fade prompt: �Can I h�?�

Correct Imitation on 2 consecutive trials Fade prompt: �Can I�?�

Correct Imitation on 2 consecutive trials Fade prompt: �Can�?�

Correct Imitation on 2 consecutive trials Fade prompt: �C�?�

Correct Imitation on 2 consecutive trials Fade prompt: Form �c� sound with mouth,

but no sound

Correct Imitation on 2 consecutive trials Fade prompt: expectant look

Correct Independent Response No prompt

Incorrect response on 3 consecutive trials Return to the previous prompt level

35

Table 4 Post-Training Generalization Probes: Participant 1

Trial # Probe Type Location What�s that? Can I see? Can I have?

1 Novel person Busy + + +

2 Novel location Busy - - -

3 Empty + + +

4 Novel placement Busy + + +

5 Novel location Busy + + +

6 Novel box Busy + + +

7 Empty + + +

8 Empty office Office + + +

9 Held box Office + + +

10 Box on desk Office - - -

11 Box on desk Office - - -

12 Phone Office - - -

36

Table 5 Post-Training Generalization Probes: Participant 2

______________________________________________________________________________________ Trial # Probe Type Location What�s that? Can I see? Can I have? 1 *Held box Office - + +

2 *Novel person Busy - - -

3 *Novel box Empty - - -

4 *Box on desk Office + + +

5 Novel person Empty - - -

6 Empty - - -

7 Busy - - -

8 Only experimenter Busy + + +

present

9 Empty - - -

10 Empty + + +

11 **Novel and training Busy P P P

person

12 **Novel person Busy P P +

13 **Novel location Busy P + P

37

______________________________________________________________________________________ Trial # Probe Type Location What�s that? Can I see? Can I have?

14 **Novel box Busy P + P

15 **Empty Office + + +

office

16 **Held box Office + + +

17 **Box on desk Office + + +

18 ***Novel and Empty + + +

training person

19 ***Novel person Empty + + +

20 ***Novel location Empty + + +

21 ***Novel box Empty + + +

________________________________________________________________________

*; **; *** Probes with the same number of asterisks were conducted on the same day.

P indicates a prompted trial.

38

REFERENCES

American Psychiatric Association. (1994). Diagnostic and statistical manual of mental

disorders (4th ed.).Washington, DC: Author.

Charlop, M. H., & Milstein, J. P. (1989). Teaching autistic children conversational speech

using video modeling. Journal of Applied Behavior Analysis, 22, 275-285.

Esbenshade, P. H., & Rosales-Ruiz, J. (2001). Programming common stimuli to promote

generalized question-asking: A case demonstration in a child with autism. Journal

of Positive Behavior Interventions, 3, 199-210.

Hung, D. W. (1977). Generalization of �curiosity� questioning behavior in autistic

children. Journal of Behavior Therapy and Experimental Psychiatry, 8, 237-245.

Koegel, L. K., Camarata, S. M., Valdez-Menchaca, M., & Koegel, R. L. (1998). Setting

generalization of question-asking by children with autism. American Journal on

Mental Retardation, 102, 346-357.

Koegel, L. K. (1995). Communication and language intervention. In R. L. Koegel & L.

K. Koegel (eds.), Teaching children with autism (pp. 17-32). Baltimore, MD:

Brookes.

Le, L. (2002). Increasing vocabulary through question-asking for children with autism.

Dissertation Abstracts International: Section B: The Sciences and Engineering,

63(5-B), 2569.

39

Minkin, N., Braukman, C. J., Minkin, B. L., Timbers, G. D., Timbers, B. J., Fixsen, D.

L., Phillips, E. L., & Wolf, M. M (1976). The social validation and training of

conversational skills. Journal of Applied Behavior Analysis, 9, 127-139

Paul, L. (1985). Programming peer support for functional language. In Warren & Rogers-

Warren (Eds.) Teaching functional language: Generalization and maintenance of

language skills (pp. 289-307). Baltimore, MD: University Park Press.

Postman, L. (1968). Association and performance in the analysis of verbal learning. In T.

R. Dixon & D. L. Horton (Eds.) Verbal behavior and general behavior theory.

Englewood Cliffs, N. J.: Prentice-Hall.

Rimland, B. (1964). Infantile autism: The syndrome and its implications for a neural

theory of behavior. Appleton-Century Crofts, New York.

Taylor, B. A., & Harris, S. L. (1995). Teaching children with autism to seek information:

acquisition of novel information and generalization of responding. Journal of

Applied Behavior Analysis, 28, 3-14.

Williams, G., Donley, C. R., & Keller, J. W. (2000). Teaching children with autism to ask

questions about hidden objects. Journal of Applied Behavior Analysis, 33, 627-

630.

Yamamoto, J. (1987). On the formation of functional �mand for instruction� in autistic

students. The Japanese Journal of Educational Psychology, 35, 97-106.