The Analysis of Verbal Behavior 2009, 25, 79–86

Teaching Intraverbal Behavior to Children with Autism:

A Comparison of Textual and Echoic Prompts

Joseph Vedora and Laura Meunier, BEACON Services

Harry Mackay, Northeastern University and Praxis Inc.

Although echoic prompts may be effective for teaching intraverbal behavior to children with autism, the
performance of some children may become dependent on such prompts (i.e., the prompts cannot be
eliminated). Recent research suggests that visual rather than echoic prompts may be used to teach
children with autism a variety of skills and may facilitate independent performance. In the present study,
an adapted alternating treatments design was used to compare the effects of using visual (textual) and
echoic prompts on acquisition of intraverbal responses (answering questions) by 2 children with autism.
The results indicated that the textual prompts were more effective than the echoic prompts. Implications
for the use of visual prompts during instruction with children with autism are discussed.
Key words: echoic prompt, textual prompt, intraverbal, autism

Children with autism often rely on teaching children with autism (e.g., McClan-
prompts provided by teachers to answer nahan & Krantz, 1997; Pierce & Schreibman,
questions and may never come to answer 1994; Schmit, Alper, Raschke, & Ryndak,
the questions independently. Echoic (verbal) 2000). Their use in activity schedules
prompts are often used, for example, to teach (MacDuff, Krantz, & McClannahan, 1993),
children with autism to tact, mand, and ask the picture exchange communication system
questions (e.g., Williams, Carnerero, & (Bondy & Frost, 1994), and schedule boards
Perez-Gonzalez, 2006; Williams, Donley, & (Savner & Myles, 2000) are a few examples.
Keller, 2000). The procedures typically In addition, Goldsmith, LeBlanc, and Sautter
involve presentation of a vocal-verbal model (2007) used pictures as a form of tact prompt
of the desired response, differential rein- to teach intraverbals to 3 children with autism.
forcement, and fading of the model. Howev- To date, only Finkel and Williams (2001)
er, echoic prompts may be difficult to elim- have compared the instructional effective-
inate by fading. Even a single brief sound ness of textual and echoic prompts. They
used as a prompt in the final fading step may used these prompts to teach the intraverbal
come to exert lasting control over responding behavior of answering questions (with sen-
(McClannahan & Krantz, 1997). Such failure tences) in an experiment with a multiple
to eliminate instructional prompts significantly baseline design. The participant was a 6-
limits a child’s independent performance. year-old boy with autism who was described
Visual prompts may provide an alternative as a visual learner with slightly above aver-
to echoic prompts. For example, Quill (1997) age sight-reading skills. Finkel and Williams’s
noted that children with autism benefited results indicated that the participant an-
from visually cued instruction and recom- swered no questions correctly during an
mended a shift in emphasis from language- initial baseline condition. Then, introduction
based instruction to the use of more visual and fading of the echoic prompts slightly
supports. Other research has demonstrated increased partial-sentence answers that were
the effective use of pictorial prompts for maintained in follow-up tests. In contrast,
the use of the textual prompt-fading proce-
This report is based on a thesis submitted by dure increased the number of questions
Laura Meunier to the Department of Counseling answered correctly with target full-sentence
Psychology, Rehabilitation, and Special Educa- answers. That performance was maintained
tion, Northeastern University, Boston. at follow-up. In sum, these data suggested
Correspondence concerning this article should
be addressed to Joe Vedora, BEACON Services, that textual prompts were more effective
321 Fortune Blvd, Milford, Massachusetts 01757. than echoic prompts in teaching intraverbals
(e-mail: jvedora@beaconservices.org). to a child with autism.

79
80 JOSEPH VEDORA et al.

Table 1
Questions and Answers for Each Participant

Child Textual Echoic

Sam Set 1 Q: ‘‘What do you do with a stove?’’ Q: ‘‘What do you do with a cup?’’
A: ‘‘cook’’ A: ‘‘drink’’
Set 2 Q: ‘‘What do you do with a book?’’ Q: ‘‘What do you do with a fork?’’
A: ‘‘read’’ A: ‘‘eat’’
Set 3 Q: ‘‘What do you do with a phone?’’ Q: ‘‘What do you do with a pencil?’’
A: ‘‘talk’’ A: ‘‘write’’
Kevin Set 1 Q: ‘‘What do you do with a vacuum?’’ Q: ‘‘What do you do with a crayon?’’
A: ‘‘clean’’ A: ‘‘color’’
Set 2 Q: ‘‘What do you do with a bike?’’ Q: ‘‘What do you do with a toy?’’
A: ‘‘ride’’ A: ‘‘play’’
Set 3 Q: ‘‘What do you do with a chair?’’ Q: ‘‘What do you do with scissors?’’
A: ‘‘sit’’ A: ‘‘cut’’

The present study sought to extend the Table 1 shows the three sets of questions
research of Finkel and Williams (2001). and the corresponding target single-word
Specifically, we aimed to assess the effects answers that were used in each condition
of echoic and textual prompts on acquisition for each participant. The questions and
and generalization of intraverbals (answering responses were considered to be functional,
questions with single words) by 2 young referring to common objects and activities
children with autism. that the participants might encounter, and the
responses were presumed equally difficult to
learn based on an analysis of the answer
METHOD
words (i.e., number of syllables) and baseline
performance. Each set consisted of two
Participants questions; the answer to one was prompted
by a textual prompt and the answer to the
Two 7-year-old boys with autism partici- other by an echoic prompt. Each textual
pated. Both spoke in three- to four-word prompt was the word to be spoken by the
sentences, could follow a few simple one- or participant in answer to a particular question.
two-step instructions, and had learned a The words were printed on individual
vocabulary of sight words (Sam, about 50 laminated pieces of paper (5 cm by 5 cm).
words, Kevin, about 100). Kevin also was Echoic prompts were the same single words
able to answer a few social questions (e.g., spoken by the experimenter.
‘‘What’s your sister’s name?’’; ‘‘Where do
you live?’’). Both attended school and Dependent Variables and Measurement
received one-on-one instruction based on
the principles of applied behavior analysis. The dependent variable was the accuracy
of a participant’s one-word responses (listed
Setting and Materials in Table 1) to questions. Partial responses,
responses that contained multiple words, and
For Sam, each session was conducted in a prompted responses were scored as incorrect
small classroom where he sat with the experi- (e.g., answering ‘‘cl’’ when asked, ‘‘What do
menter at a table facing a wall to minimize you do with a vacuum?’’), as were responses
distractions. Sessions lasting 10 min oc- that contained part of the questions (e.g.,
curred once or twice a day during school answering ‘‘bike ride’’ when asked, ‘‘What
hours. Sessions with Kevin also lasted about do you do with a bike?’’) and answers that
10 min. They occurred three to five times per did not make sense (e.g., answering ‘‘fork’’
week and were conducted at his desk in his when asked, ‘‘What do you do with a
bedroom. fork?’’).
 TEXTUAL AND ECHOIC PROMPTS 81

In every session, 10 questions were pre- Baseline. Each question in a set was
sented. The mastery criterion, at least nine presented 10 times (20 trials total). No
correct answers to the 10 questions asked in prompts were delivered, and no programmed
each of two consecutive sessions, was the consequences followed correct or incorrect
same for both participants in both prompt responses.
conditions. After both questions in a set were Textual prompt. The textual prompts listed
mastered, baseline and training began for the in Table 1 were used to teach three intra-
following set. verbal responses to Sam and Kevin. The
training used a progressive prompt-
Experimental Design delay procedure (Touchette, 1971). Initially,
prompts immediately (0-s delay) followed
An adapted alternating treatments design question presentations. Then the delay be-
(Sindelar, Rosenberg, & Wilson, 1985) was tween the question presentation and the
used to compare the effects of textual and prompt was increased gradually to allow
echoic prompts on the acquisition and gen- the participant time to respond independently
eralization of the intraverbal responses. The before the prompt was presented. Correct
order in which the different prompts were responses canceled the scheduled presenta-
used in training was varied unsystematically. tion of the prompt, and after two consecutive
This design permitted assessment of the correct trials the delay was increased by 1 s
effects of the training conditions within and up to a maximum of 5 s. If the participant did
across participants. When criterion was not not respond or made an error, the prompt was
met after four sessions of training with one given, the trial was scored as incorrect, and
prompt type, it was replaced by the other the delay was reduced by 1 s on the follow-
type of prompt in a final best treatment ing trial. Tokens and praise were delivered
phase. contingent on correct responses and prompt-
ed responses during the 0-s delay condition.
Echoic prompt. The echoic prompts listed
Interobserver Agreement in Table 1 were used. The training used the
same progressive prompt-delay procedure
Interobserver agreement data were collect- used in the textual prompt condition.
ed in 33% of the sessions by having an Generalization tests. After the trained
independent observer score the answer given performances were mastered, two additional
to each question. Agreement was defined as teachers (Sam) and two family members
both observers scoring a response correct, or (Kevin) implemented trials with the trained
both observers scoring a response incorrect. stimuli to examine generalization across
Interobserver agreement was determined by people. Sam’s tests were conducted during
dividing the number of agreements by the his school day in the separate room where
number of agreements plus disagreements training occurred or in his regular classroom.
and multiplying by 100%. Agreement was Kevin’s tests occurred after school at his
100% for each participant. home. Ten trials were implemented for each
question.
Procedure
RESULTS
Preliminary teaching. This training was
given to ensure correct oral naming of the Figure 1 displays Sam’s unprompted an-
words to be used in the textual prompt con- swers to questions during baseline, training
dition before experimental training began. with both textual and echoic prompts, and
Sam learned to name these printed words generalization tests. During the baseline
in a total of three sessions. Kevin named conditions with all three stimulus sets, he
the necessary words prior to the study. The answered no questions correctly. Sam then
responses to be used in the echoic condi- learned to answer all six questions, but the
tion were also evaluated. On these trials, effectiveness of the prompting conditions
the experimenter asked the students to, differed. He learned the answer presented by
‘‘Say —.’’ Both students did so on all trials. textual prompt in Set 1 (top) in five sessions.
82 JOSEPH VEDORA et al.

Figure 1. Number of independent correct responses per session for Sam. The upper, middle, and bottom
panels show data for Sets 1, 2, and 3, respectively. The open data points represent generalization tests.
 TEXTUAL AND ECHOIC PROMPTS 83

On all the trials on which answers were on all trials in the textual prompt conditions,
not given independently (e.g., all textually thus demonstrating errorless learning of the
prompted trials in Session 5), delivery of the answers to the questions presented during
prompt produced the correct answer. In these conditions. In contrast, although both
contrast, training with the echoic prompt children demonstrated generalized echoic
took 10 sessions, and errors (e.g., partial repertoires prior to this study, they made
answers) as well as prompted correct answers errors during training with echoic prompts
occurred. On generalization tests, Sam inde- and thus required more instructional trials to
pendently answered almost all questions. The reach criterion. During his first exposure to
same advantages for textual prompts oc- the echoic prompt condition, Kevin did not
curred for Sets 2 (middle) and 3 (bottom). respond independently on any trial in four
These performances generalized across per- sessions. However, after the echoic prompts
sons regardless of training method. were replaced by textual prompts, he learned
Figure 2 displays the results for Kevin. to answer the same question appropriately
During baseline, no questions were answered in three sessions. These results suggest that
correctly for any stimulus set. In the first the participants’ possession of generalized
session with Set 1 (top), he responded echoic repertoires did not suffice to enable
correctly in four textually prompted trials. effective use of echoic prompts as critical
Criterion then was met in Sessions 2 and 3. In training cues.
the echoic prompt condition, he failed to The present study involved several meth-
reach criterion for the response to the ques- odological and procedural differences from
tion, ‘‘What do you do with a crayon?’’ In Finkel and Williams (2001), including a
the first four sessions, he responded incor- different experimental design, teaching sin-
rectly by either repeating the last word of the gle word rather than short sentence answers
question or giving an approximation of the to questions, and the use of prompt delay
word. The textual prompts were then applied, instead of fading. Of particular interest is the
and criterion was met after three sessions. use of the prompt delay to transfer control
On generalization tests, he independently from the prompt to the relevant antecedent
answered all questions. His results for Set 2 verbal stimulus. Despite this procedural dif-
(middle) and Set 3 (bottom) show faster ference, the textual prompts were more
acquisition in the textual prompt condition. effective, suggesting that the type of prompt
Performance on generalization tests was rather than the transfer-of-control procedure
highly accurate. was responsible for the differences in re-
Table 2 displays the number of trials to sponding.
criterion for each participant. Both required Accounting for the greater effectiveness of
fewer trials to reach criterion for each stim- the textual prompts may involve both sub-
ulus set that was taught using the textual ject and procedural variables. First, some
prompt. researchers suggest that children with autism
are visual learners (e.g., Quill, 1997; Tissot
DISCUSSION & Evans, 2003). Previous research has
demonstrated that visual prompts can be
The present results extend the findings of used effectively with children with autism
Finkel and Williams (2001) with 2 partici- to establish a variety of skills, including
pants with autism and different tasks and communication, self-help, and leisure skills
training procedures. The data suggest that the (Bondy & Frost, 1994; McClannahan &
use of textual prompts and a progressive Krantz, 1997; Pierce & Schreibman, 1994;
prompt-delay procedure established intraver- Quill; Schmit et al., 2000). The present
bal skills, like those studied by Finkel and research extends these findings and, together
Williams, more rapidly than echoic prompts with that of Finkel and Williams (2001),
with the same delay procedure. At the start suggests that visual (here textual) prompts
of the experiment, both participants failed also are effective for teaching intraverbal
to answer the questions. Both then quickly behavior. It is not clear, however, to what
reached criterion performance in the textual extent such results may reflect a greater
prompt conditions. They responded correctly sensitivity of individuals with autism to
84 JOSEPH VEDORA et al.

Figure 2. Number of independent correct responses per session for Kevin. The upper, middle, and
bottom panels show data for Sets 1, 2, and 3, respectively. The open data points represent generalization
tests.
 TEXTUAL AND ECHOIC PROMPTS 85

Table 2 may improve performance on tests designed

Total Number of Trials to Criterion for to measure academic performance (Gold-
Each Condition smith et al., 2007). Thus, Sam’s and Kevin’s
teachers could continue to use textual
Child Textual Echoic prompts to expand the intraverbal repertoires
Sam Set 1 50 100 already established and improve conversa-
Set 2 30 50 tional skills and performance on academic
Set 3 40 60 assessments. Lastly, the implications of the
Kevin Set 1 30 —a present study are practical because textual
Set 2 30 50 prompts can be applied in a variety of
Set 3 30 60 programs. Although the materials used were
printed and laminated, they may be made
a
Did not meet criterion. more accessible and even less expensive by
writing the cues on a piece of paper.
visual cues or, as Finkel and Williams de- The present findings are limited by the
scribed, a prior history that did not establish failure to demonstrate formally the equiva-
or maintain attending to echoic prompts. lence of the instructional sets (Sindelar et al.,
Such a history would render the echoic 1985). Thus, it is possible that the results
prompts used in this study less effective at reflect differences in the difficulty of ques-
evoking the responses to be trained and, more tions and responses rather than effectiveness
generally, would help to account for the of instructional procedures. However, this
behavior that underlies the notion that concern is lessened by the successful repli-
children with autism may attend better to cations with each participant. Another po-
objects than to people (Charlop-Christy, Le, tential limitation in the use of textual
& Freeman, 2000). In addition, aspects of the prompts to teach intraverbal skills is the
prompt procedures used in this study may need to teach the reading responses that
contribute to the difference in effectiveness ensure the effectiveness of the printed
of the textual and echoic prompts. The prompts. This was not an issue in Kevin’s
textual prompts remained available until the case because he was able to read the textual
child answered the question. In contrast, the prompts prior to beginning the study. How-
spoken echoic prompts were transitory. They ever, Sam required training to do so, al-
were presented only once and were not though that was not time consuming. Also,
repeated, thus yielding a delayed prompting although generalization tests were conducted
condition that may have failed to support the following training, generalization was not
desired behavior of answering the question formally assessed, and further research is
by imitating the prompts. Further research is needed to clarify whether different instruc-
needed to clarify these possibilities. tional prompts affect generalized responding.
The results of this study are socially signi- Finally, the present study included only 2
ficant. Both participants demonstrated rapid children with autism. Future research should
acquisition of responses that also occurred in examine generality by including more par-
novel situations in which people who were ticipants of different ages and disabilities.
not involved in the training asked the ques-
tions. Several empirically supported curricula REFERENCES
for children with autism recommend that
similar intraverbal performances be included Bondy, A., & Frost, L. (1994). The picture
as objectives in instructional programs (Leaf exchange communication system. Focus
& McEachin, 1999; Partington & Sundberg, on Autistic Behavior, 9, 1–19.
1998). Such development of an intraverbal Charlop-Christy, M. H., Le, L., & Freeman,
repertoire may enhance a child’s conversa- K. A. (2000). A comparison of video
tional skills and enable him or her to respond modeling with in vivo modeling for
more effectively and appropriately to other teaching children with autism. Journal of
people. In addition, a student’s ability to Autism and Developmental Disorders, 30,
respond to statements and answer questions 537–552.
86 JOSEPH VEDORA et al.

Finkel, A. S., & Williams, R. L. (2001). A Quill, K. (1997). Instructional considerations

comparison of textual and echoic prompts for young children with autism: The
on the acquisition or intraverbal behavior rationale for visually-cued instruction.
in a six-year-old boy with autism. The Journal of Autism and Developmental
Analysis of Verbal Behavior, 18, 61–70. Disabilities, 27, 697–714.
Goldsmith, T. R., LeBlanc, L. A., & Sautter, Savner, J. L., & Myles, B. S. (2000). Making
R. A. (2007). Teaching intraverbal behav- visual supports work in the home and
ior to children with autism. Research in community: Strategies for individuals with
Autism Spectrum Disorders, 1, 1–13. autism and Asperger syndrome. Shawnee
Leaf, R., & McEachin, J. (Eds.). (1999). Mission, KS: Autism Asperger Publishing.
A work in progress: Behavioral manage- Schmit, J., Alper, S., Raschke, D., & Ryndak,
ment strategies and a curriculum for D. (2000). Effects of using a photographic
intensive behavioral treatment of autism. cueing package during routine school
New York: DRL Books. transitions with a child who has autism.
MacDuff, G. S., Krantz, P. J., & McClanna- Mental Retardation, 38, 131–137.
han, L. E. (1993). Teaching children Sindelar, P. T., Rosenberg, M. S., & Wilson,
with autism to use photographic activity R. J. (1985). An adapted alternating treat-
schedules: Maintenance and generaliza- ments design for instructional research.
tion of complex response chains. Journal Education and Treatment of Children, 8,
of Applied Behavior Analysis, 26, 89–97. 67–76.
McClannahan, L. E., & Krantz, P. J. (1997). Tissot, C., & Evans, R. (2003). Visual teach-
In search of solutions to prompt depen- ing strategies for children with autism.
dence: Teaching children with autism to Early Child Development and Care, 173,
use photographic activity schedules. In D. 425–433.
M. Baer & E. M. Pinkston (Eds.), Touchette, P. E. (1971). Transfer of stimulus
Environment and behavior (pp. 271– control: Measuring the moment of trans-
278). Boulder, CO: Westview. fer. Journal of the Experimental Analysis
Partington, J. W., & Sundberg, M. L. (1998). of Behavior, 15, 347–354.
Teaching language to children with au- Williams, G., Carnerero, J. J., & Perez-
tism or other developmental disorders. Gonzalez, L. A. (2006). Generalization of
Pleasant Hill, CA: Behavior Analysts, Inc. tacting actions in children with autism.
Pierce, K. L., & Schreibman, L. (1994). Journal of Applied Behavior Analysis, 39,
Teaching daily living skills to children 233–237.
with autism in unsupervised settings Williams, G., Donley, C. R., & Keller, J. W.
through pictorial self-management. Jour- (2000). Teaching children with autism to
nal of Applied Behavior Analysis, 27, ask questions about hidden objects. Journal
471–481. of Applied Behavior Analysis, 33, 627–630.