JOURNAL OF APPLIED BEHAVIOR ANALYSIS 2011, 44, 169–173 NUMBER 1 (SPRING 2011)

RESPONSE INTERRUPTION AND REDIRECTION FOR

VOCAL STEREOTYPY IN CHILDREN WITH AUTISM:
A SYSTEMATIC REPLICATION
MEGAN DUFFY CASSELLA AND TINA M. SIDENER
CALDWELL COLLEGE

DAVID W. SIDENER
GARDEN ACADEMY

AND

PATRICK R. PROGAR
CALDWELL COLLEGE

This study systematically replicated and extended previous research on response interruption and
redirection (RIRD) by assessing instructed responses of a different topography than the target
behavior, percentage of session spent in treatment, generalization of behavior reduction, and
social validity of the intervention. Results showed that RIRD produced substantial decreases in
vocal stereotypy. Limitations of this study were that behavior reduction did not generalize to
novel settings or with novel instructors and that appropriate vocalizations did not improve.
Key words: stereotypy, response interruption and redirection, autism
________________________________________

Response interruption has been shown to be intervention in which the participants were
effective in reducing various topographies of interrupted and redirected to appropriate vo-
stereotypy (e.g., Azrin & Wesolowski, 1980; calizations via social questions and vocal
Sprague, Holland, & Thomas, 1997). Several imitation trials.
studies have evaluated a variation of this The purpose of the present study was to
intervention called response interruption and systematically replicate and extend Ahearn et
redirection (RIRD), which consists of inter- al. (2007) by evaluating the effects of RIRD
rupting the target response and redirecting the using directions that required motor responses
individual to engage in a different response on the vocal stereotypy of two children with
(e.g., Ahearn, Clark, MacDonald, & Chung, autism, based on Ahearn et al.’s speculation
2007; Athens, Vollmer, Sloman, & St. Peter that nonvocal tasks may produce similar
Pipkin, 2008). Ahearn et al. evaluated an RIRD treatment effects. In addition, motor responses
may be manually prompted if necessary
(whereas vocal responses may be difficult to
This study is based on a thesis submitted by the first prompt). Although we have noted that clini-
author, under the supervision of the second author, to the
Department of Psychology at Caldwell College for the MA cians use this variation of RIRD with children
in Applied Behavior Analysis. We thank Christina Romeo, with autism, few studies have evaluated it (e.g.,
Kristie Murano, Sidona Anderson, Stacey Bord, Kim Ahrens, Lerman, Kodak, Worsdell, & Keegan,
Sancho, Jill Bernstein, Nathan Cassella, Kristin Paladino,
Allison Buleza, and Genevieve Williams for their assis-
2011). The current study also extended pre-
tance. vious research on RIRD for vocal stereotypy by
Address correspondence to Tina M. Sidener, Depart- including measurement of time spent in
ment of Applied Behavior Analysis, Caldwell College, 120 treatment, procedures to facilitate and assess
Bloomfield Ave., Caldwell, New Jersey 07006 (e-mail:
tsidener@caldwell.edu). the generalization of behavior reduction, and
doi: 10.1901/jaba.2011.44-169 assessment of social validity.

169
170 MEGAN DUFFY CASSELLA et al.

METHOD system because it has been shown to result in

Participants and Setting fewer errors than either partial- or whole-
interval recording (Alvero, Struss, & Rappaport,
Participants were two boys who had been
2008). At the time of each measurement (i.e., at
diagnosed with severe autism. Chris (7 years
10 s), the experimenter observed the participant
2 months old) and Adam (4 years 11 months
for 5 s to obtain an adequate sample of the vocal
old) engaged in high levels of vocal stereotypy
stereotypy. A plus was scored if vocal stereotypy
that occurred across the day and interfered with
occurred at any time during the 5-s observation
learning. Vocal stereotypy consisted of a variety
period. Observers used a 5-s observation
of vocalizations and repetition of words and
window rather than a 1-s observation window
phrases. Prior to the study, both participants
so that they would be able to detect instances of
demonstrated that they could follow at least 15
repetitive vocalizations. During treatment ses-
one-step directions (e.g., stand up, turn around)
sions, data were also collected on time spent in
and could imitate similar motor movements treatment. These data were summarized as
with 100% accuracy when modeled by an percentage of session in treatment by dividing
instructor. Sessions were conducted 5 days per total duration of RIRD by total session duration
week in three areas in the participants’ school. and multiplying by 100%.
Three functional assessments were used to Interobserver agreement on the occurrence of
collect information about the possible function vocal stereotypy and appropriate vocalizations
of the vocal stereotypy of the participants: was assessed using the point-by-point method.
the Functional Assessment Interview (FAI; An agreement was scored in an interval if both
O’Neill et al., 1997), the Functional Assessment observers recorded the occurrence or nonoccur-
Screening Tool (FAST; Iwata & DeLeon, 1995) rence of vocal stereotypy or appropriate vocal-
and the Motivation Assessment Scale (MAS; izations. Agreement data were collected by a
Durand & Crimmins, 1992). These assessments second independent observer for a mean of
were conducted so that individuals who engaged 43% of sessions. For vocal stereotypy, mean
in stereotypy with social functions could be agreement was 93% (range, 84% to 100%) for
excluded from the study. For Chris, responses Chris and 95% (range, 89% to 100%) for
to the FAI suggested an automatic reinforce- Adam. Interobserver agreement for appropriate
ment function, and rating scale results suggested vocalizations was 100% for both participants.
automatic reinforcement or attention functions
(MAS sensory M 5 6.25; FAST attention Design and Procedure
proportional score 5 0.75). For Adam, all An ABAB reversal design was used to evaluate
assessments suggested an automatic reinforce- the effects of RIRD on vocal stereotypy. Baseline
ment function (FAI; MAS sensory M 5 5; sessions were 5 min in duration, and treatment
FAST sensory proportional score 5 0.83). sessions ended following 5 min without the
participant engaging in vocal stereotypy or after
Response Measurement and Reliability 30 min, whichever came first. The participant
Vocal stereotypy was defined as noncontextual had noncontingent access to two moderately
vocalizations, as well as contextual vocalizations preferred nonauditory toys, identified via a
repeated within 3 s of a similar vocalization multiple-stimulus without replacement prefer-
(e.g., saying ‘‘ball’’ repetitively when seeing a ence assessment (DeLeon & Iwata, 1996),
ball). Appropriate vocalizations were defined as during all sessions. If the participant directed
contextual vocalizations that were not repeated an appropriate vocalization toward the experi-
within 3 s of a similar vocalization. Time menter (e.g., ‘‘hi’’), the experimenter briefly
sampling was selected as the data-collection acknowledged it (e.g., ‘‘hi’’).
 RIRD AND VOCAL STEREOTYPY 171

During baseline, the experimenter did not was 99.6% (range, 94% to 100%) for Chris and
present demands or provide programmed 99.8% (range, 97% to 100%) for Adam and
consequences for vocal stereotypy. During was calculated for a mean of 60% of procedural
treatment, each instance of vocal stereotypy integrity sessions.
was followed by the experimenter saying the
participant’s name, establishing attention by RESULTS AND DISCUSSION
gaining eye contact, and giving a one-step
direction that did not require a vocal response Figure 1 shows the percentage of intervals
(e.g., ‘‘touch head’’). The experimenter provid- with vocal stereotypy and the percentage of time
ed a model of the behavior if the participant spent in treatment during the treatment analysis
responded incorrectly or did not respond within and generalization probes. Vocal stereotypy for
5 s of the direction. Gentle manual prompting both participants was high and variable during
baseline. RIRD resulted in an immediate and
with graduated guidance was used if the child
substantial decrease in vocal stereotypy that
did not imitate the model following 5 s of the
remained relatively stable throughout the phase.
model being presented. Behavior-specific praise
For both participants, vocal stereotypy was
was delivered following each completed direc-
lower in the initial session during the return
tion, and the procedure was terminated follow-
to baseline and subsequently increased as
ing three consecutive correct directed responses
sessions continued. Effects of treatment were
(with or without prompts) without engagement
replicated in the second RIRD phase. The high
in vocal stereotypy. Ten different one-step level of vocal stereotypy in Session 30 for Adam
directions were delivered during treatment. coincided with illness.
Generalization was facilitated by implement- Chris spent much less time (i.e., an average of
ing the procedure in two different rooms and by less than 50% of sessions) in treatment than
two different therapists. Generalization was Adam, who spent most of the time during his
assessed by evaluating the procedure in a novel sessions in treatment (i.e., an average of 77%
room with an instructor who had not previously of sessions). This may have been due to the
implemented the procedure. Probes of novel reinforcing properties of stereotypy or attention
locations and instructors were conducted with- provided during RIRD (i.e., vocal directions,
out the implementation of RIRD. Implemen- manual prompts).
tation of the treatment with novel demands was Use of multiple exemplars did not facilitate
assessed as well. generalization across instructors or settings
To evaluate social validity, the children’s during probes when RIRD was not in effect
parents and a teacher were asked to view 5-min (Figure 1). Vocal stereotypy remained at baseline
videotaped segments of the participants during levels during most of these sessions. Probes of
treatment and to respond to questions from a RIRD with novel demands resulted in levels of
modified version of the Treatment Evaluation vocal stereotypy similar to those in other
Inventory—Short Form (Van Norman, 2006). treatment sessions. For Chris, appropriate vocal-
Procedural integrity data were collected izations (data not shown) occurred at low levels
during a mean of 74% of sessions on correct during generalization probes in baseline and
implementation of directions, prompts, and increased slightly during the first RIRD phase
praise, as well as the termination of directions at (M 5 2%; range, 3% to 9%). Adam emitted
the correct time. Mean procedural integrity was appropriate vocalizations in only the first RIRD
99.7% (range, 98% to 100%) for Chris and phase (M 5 0.5%; range, 0% to 3%).
99.6% (range, 97% to 100%) for Adam. Mean The social validity results for Chris indicated
interobserver agreement on procedural integrity that his caregivers found RIRD highly accept-
172 MEGAN DUFFY CASSELLA et al.

Figure 1. Percentage of intervals with vocal stereotypy and percentage of session spent in treatment for Chris (top)
and Adam (bottom).

able (M 5 4.5, range, 4 to 5). Results were ible responses. Some results obtained in the
slightly less favorable (M 5 4.1, range, 3 to 5) current study, however, do not replicate the
for Adam. Adam’s parents gave lower scores on findings of Ahearn et al. Appropriate vocaliza-
the questionnaire regarding the acceptability and tions did not increase as vocal stereotypy
effectiveness of this procedure, perhaps because, decreased. In addition, vocal stereotypy did
as reflected in the videos, Adam spent most of the not remain low without RIRD.
time during his sessions in treatment. It should be noted that, although this
These results replicate and extend the procedure is called redirection, the mechanism
findings of Ahearn et al. (2007) and demon- by which RIRD produces its effects is currently
strate that RIRD can be effective in decreasing unknown. As posited by Ahearn et al. (2007), it
vocal stereotypy without the use of incompat- is possible that presenting demands contingent
 RIRD AND VOCAL STEREOTYPY 173

on vocal stereotypy functioned as a positive Ahrens, E. N., Lerman, D. C., Kodak, T., Worsdell, A. S.,
punishment contingency (see also Ahrens et al., & Keegan, C. (2011). Further evaluation of response
interruption and redirection as treatment for stereo-
2011). Although toys were not removed during typic behavior. Journal of Applied Behavior Analysis,
treatment, it is also possible that the interven- 44, 95–108.
tion functioned as negative punishment via Alvero, A. M., Struss, K., & Rappaport, E. (2008).
Measuring safety performance: A comparison of
interruption of toy play contingent on vocal
whole, partial, and momentary time-sampling record-
stereotypy. ing methods. Journal of Organizational and Behavior
One limitation of the current study was that Management, 27, 1–28.
sessions were often lengthy, with the majority of Athens, E. S., Vollmer, T. R., Sloman, K. N., & Pipkin,
C. S. (2008). An analysis of vocal stereotypy and
sessions lasting for 30 min. Ahearn et al. (2007) therapist fading. Journal of Applied Behavior Analysis,
reported that many treatment sessions were longer 41, 291–297.
than 10 min, but that many averaged 6 min. A Azrin, N. H., & Wesolowski, M. D. (1980). A
second limitation was that aggression (i.e., reinforcement plus interruption method of eliminat-
ing behavioral stereotypy of profoundly retarded
pinching) and crying were observed in several persons. Behaviour Research & Therapy, 18, 113–119.
sessions for Adam. These behaviors decreased as DeLeon, I. G., & Iwata, B. A. (1996). Evaluation of a
RIRD continued to be implemented. A final multiple-stimulus presentation format for assessing
reinforcer preferences. Journal of Applied Behavior
limitation is that momentary time sampling may
Analysis, 29, 519–533.
both over- and underestimate behavior. Durand, V. M., & Crimmins, D. (1992). Motivation
Future research might evaluate the implemen- assessment scale. Topeka, KS: Monaco & Associates.
tation of RIRD by caregivers and in more natural Iwata, B. A., & DeLeon, I. G. (1995). The functional
analysis screening tool (FAST). Unpublished manu-
environments. In addition, future research might script, University of Florida.
examine the behavioral mechanisms in effect for O’Neill, R. E., Horner, R. H., Albin, R. W., Sprague,
RIRD by evaluating RIRD with and without J. R., Storey, K., & Newton, J. S. (1997). Functional
toys, the effects of different contingencies for assessment and program development for problem
behavior: A practical handbook. Belmont, CA: Wads-
compliance (e.g., no rewards, praise), and the worth.
effects of different types of prompts (e.g., manual Sprague, J., Holland, K., & Thomas, K. (1997). The effect
prompts, modeling). In future studies, research- of noncontingent sensory reinforcement, contingent
ers could also evaluate methods to promote sensory reinforcement, and response interruption on
stereotypical and self-injurious behavior. Research in
generalization and appropriate vocalizations Developmental Disabilities, 18, 61–77.
(e.g., by increasing the number of locations and Van Norman, R. K. (2006). The effects of functional
instructors used in treatment, reinforcing appro- communication training, choice making, and an
adjusting work schedule on problem behavior main-
priate vocalizations during RIRD). tained by negative reinforcement (Unpublished doctor-
al dissertation). The Ohio State University, Colum-
bus.
REFERENCES
Ahearn, W. H., Clark, K. M., MacDonald, R. P. F., &
Chung, B. I. (2007). Assessing and treating vocal Received November 11, 2009
stereotypy in children with autism. Journal of Applied Final acceptance July 9, 2010
Behavior Analysis, 40, 263–275. Action Editor, Jennifer Zarcone