J Autism Dev Disord (2014) 44:1244–1251

DOI 10.1007/s10803-013-1972-y

BRIEF REPORT

A Randomized Trial Comparison of the Effects of Verbal

and Pictorial Naturalistic Communication Strategies on Spoken
Language for Young Children with Autism
Laura Schreibman • Aubyn C. Stahmer

Published online: 23 November 2013

Ó Springer Science+Business Media New York 2013

Abstract Presently there is no consensus on the spe- Keywords Autism
Behavioral intervention

cific behavioral treatment of choice for targeting lan- Functional communication
Vocal language
guage in young nonverbal children with autism. This intervention
Pictorial communication intervention

randomized clinical trial compared the effectiveness of a Augmentative communication
verbally-based intervention, Pivotal Response Training
(PRT) to a pictorially-based behavioral intervention, the
Picture Exchange Communication System (PECS) on the Introduction
acquisition of spoken language by young (2–4 years),
nonverbal or minimally verbal (B9 words) children with Recently the importance of early intervention (i.e., treat-
autism. Thirty-nine children were randomly assigned to ment before the age of 4 years) has been heavily empha-
either the PRT or PECS condition. Participants received sized in the treatment of autism (Dawson 2008). Indeed,
on average 247 h of intervention across 23 weeks. recent literature estimates that from 20 to 50 % of children
Dependent measures included overall communication, with autism fail to ever acquire spoken language (Tager-
expressive vocabulary, pictorial communication and par- Flusberg et al. 2005). Achieving spoken language by age
ent satisfaction. Children in both intervention groups 5–6 years is associated with better long-term outcomes in
demonstrated increases in spoken language skills, with ASD (National Research Council 2001). Because early
no significant difference between the two conditions. intervention is likely to impact spoken language, exami-
Seventy-eight percent of all children exited the program nation of methods to facilitate language development in
with more than 10 functional words. Parents were very young, nonverbal children with autism is extremely
satisfied with both programs but indicated PECS was important. However, there is no consensus on the specific
more difficult to implement. behavioral treatment model of choice for targeting com-
munication in these children (National Research Council
2001). Although different treatment models have been
developed and separately empirically validated as effective
in teaching communication skills, few have been directly
L. Schreibman (&) compared in a controlled study.
Department of Psychology, University of California, San Diego, One widely used and manualized approach to teaching
9500 Gilman Drive MC 0109, La Jolla, CA 92093-0109, USA spoken language is Pivotal Response Training (PRT), a
e-mail: [email protected]
naturalistic behavioral intervention with strong empirical
A. C. Stahmer support (e.g., Koegel et al. 1987). Using PRT for children
Department of Psychiatry, University of California, San Diego, with ASD results in language improvements and concomi-
9500 Gilman Drive MC 0603, La Jolla, CA 92093-0603, USA tant decreases in inappropriate and disruptive behaviors
e-mail: [email protected]
(Koegel et al. 1992). PRT has been shown to be effective for
A. C. Stahmer improving speech imitation (Koegel et al. 1998; Laski et al.
Rady Children’s Hospital, San Diego, CA, USA 1988), labeling (Koegel et al. 1998), spontaneous speech

123
J Autism Dev Disord (2014) 44:1244–1251 1245

(Laski et al. 1988), and rapid acquisition of functional speech by administration of the Autism Diagnostic Interview—
in previously nonverbal children (Sze et al. 2003). However, Revised (ADI-R; Lord et al. 1994), and the Autism Diag-
the failure of many individuals with autism to acquire lan- nostic Observation Schedule—Generic (ADOS-G; Lord
guage using verbally-based treatment methods has led to use et al. 2000), (b) under 48 months old, (c) no more than nine
of alternative augmentative communication systems. The intelligible words (d) absence of evidence for diagnosis of
most widely used of these approaches, the Picture Exchange primary mental retardation, neurological pathology or
Communication System (PECS; Bondy and Frost 2001) also major sensory impairment, (e) absence of prior treatment
enjoys empirical support and teaches individuals to involving either PECS or PRT, and (f) parental willingness
exchange picture icons to communicate. We have known to participate in parent training and to refrain from the non-
that children with autism can learn to use augmentative assigned treatment during the duration of the study.
systems to communicate (Mirenda and Iacono 1988) and that Monthly contact between the research team and outside
use of an augmentative system offers a functional system providers and weekly discussion with parents indicated that
until spoken language is developed. Further it has been no parents used the non-assigned treatment. Parents were
posited that the acquisition of such a system reduces offered training in the alternative condition at the end of
behavioral difficulties and actually facilitates language the study.
acquisition (Bondy and Frost 2001). Several studies have Forty-one families met the eligibility criteria, however
found that the use of PECS increases spoken communication two families (one each per site, one each per condition)
in some children with ASD (e.g., Charlop-Christy et al. 2002; discontinued participation during the first several weeks of
Ganz et al. 2007; Yoder and Stone 2006a). Romski et al. treatment. One family moved out of the area and one
(2010) reported that augmented language interventions family chose to receive the nonassigned condition. Thirty-
facilitated, rather than hindered, speech production abilities nine children (34 male, 5 female) between 20 and
in young children with developmental delays. 45 months (M = 29.21, SD = 5.67) participated in the
The primary difference between these approaches is that study with 20 children in the PRT condition and 19 in the
PRT teaches communication through verbal strategies and PECS condition. The child’s primary caregiver participated
PECS through pictorial methods. However, only two reports in parent education (32 mothers, 7 fathers). Seventeen
(presenting data for the same set of participants) have sys- children participated at University Site 1 and 22 at Uni-
tematically compared differential effects of verbally and versity Site 2. Table 1 presents demographic information
visually-based communication programs for young children by treatment condition. There were no statistically signif-
with autism. Yoder and Stone (2006a, b) conducted a ran- icant differences between treatment conditions in any of
domized comparison of a verbally-based naturalistic inter- the tested variables at intake (see Table 1).
vention, responsive education and prelinguistic milieu
teaching (RPMT) to PECS. Results indicated that PECS, on
average, was superior to RPMT for improving children’s
spoken communication/spoken language, although results Table 1 Child demographic variables at pre-treatment
varied depending upon child characteristics. Additional data Variables PECS PRT Whole sample
are needed to determine whether PECS, on average, tends to (n = 19) (n = 20) (N = 39)
be superior to other naturalistic behavioral interventions for
Age in months 28.9 (4.2)a 29.5 (6.9)a 29.2 (5.67)a
fostering spoken language in this population.
Gender
The present investigation included a direct comparison
Male 16 (84.2 %) 18 (90.0 %) 34 (87.2 %)
of spoken language outcomes for young, minimally verbal
Female 3 (15.8 %) 2 (10.0 %) 5 (12.8 %)
children with autism taught communication using either
Campus
PECS or PRT. Both interventions are empirically sup-
University site 1 9 (47.4 %) 8 (40.0 %) 17 (43.6 %)
ported and both are commonly used in community treat-
ment settings. University site 2 10 (52.6 %) 12 (60.0 %) 22 (56.4 %)
Words use
No words 11 (57.9 %) 10 (50.0 %) 21 (53.8 %)
Method 1–10 words 8 (42.1 %) 10 (50.0 %) 18 (46.2 %)
Cognitive functioning
Participants Low 8 (42.1 %) 12 (60.0 %) 20 (51.3 %)
High 11 (57.9 %) 8 (40.0 %) 19 (48.7 %)
Participants included referrals to two university-based Numbers are frequency (percent of treatment condition sample)
autism research programs who met the following criteria: unless otherwise noted
a
(a) diagnosis of Autistic Disorder (APA 2000) as confirmed Numbers are M (SD)

123
1246 J Autism Dev Disord (2014) 44:1244–1251

Experimental Design and Procedure components over two treatment sessions) on treatment
fidelity of implementation. Fidelity of implementation was
Children were randomly assigned to PRT or PECS using a tracked after every 10 h of intervention provided by each
stratified randomization procedure. Children were matched therapist. Therapy sessions were videotaped and later
on three, two-level factors: word use (no words or 1–9 coded for fidelity of implementation by a coder blind to the
functional words), age (18–32 mos or 33–47 mos) and study hypotheses. Therapists were not aware of which
cognitive functioning (low or high). A child was catego- sessions would be evaluated for fidelity of implementation.
rized as having no words if he or she was reported to have If a therapist fell below the 80 % criterion, they were
used no words communicatively on the Vineland Adaptive removed from treatment and re-trained to criterion (this
Behavior Scales (VABS; (Sparrow and Cicchetti 1989) and occurred only once). Parent education. Parent educators
observed to use no words the Mullen Scales of Early were doctoral students well experienced in autism and the
Learning (MSEL; Mullen 1995), Autism Diagnostic use of PRT and PECS. All parent educators met criterion
Observation Schedule (ADOS; Lord et al. 2000), Expres- for fidelity of implementation for both interventions. Cri-
sive One-Word Picture Vocabulary Test (EOWPVT; terion for fidelity of implementation was a minimum of
Gardner 1990), and a 25-min parent–child observation. A 80 % correct usage of all treatment components. Parent
child was categorized as having some words if he or she education consisted of reading the assigned manual (PECS
was reported or observed to have used any words com- or PRT), direct one-on-one review of therapeutic proce-
municatively on any of the above measures at intake. dures and exercises in the manuals, and multiple practice
Cognitive functioning was assessed using the visual sessions with modeling and feedback.
reception subscale of the Mullen Scales of Early Learning
(MSEL; Mullen 1995). Age-adjusted visual reception Intervention
scores were derived by dividing a child’s visual reception
age equivalent (AE) score by chronological age and mul- Both PRT and PECS use motivation and child initiation as
tiplying by 100. Low cognitive functioning was defined as a basis for facilitating communication and are based on the
an adjusted score of less than or equal to 50; high cognitive principles of applied behavior analysis. Specifically, both
functioning was defined as an adjusted score of [50. interventions provide natural opportunities for communi-
For every two children assessed that were matched on cation (e.g., communication temptations), require a
all variables, one was randomly assigned to PRT or PECS response from the child, and use direct reinforcement
and the other to the alternative condition. Dependent (reinforcement directly related to the child’s response) to
measures were obtained at program entry, post interven- increase responding. Although both treatment packages
tion, and after a 3-month follow up period during which no included similar materials (with the exception of PECS
intervention was provided. specific picture cards) and were delivered at similar
intensity, the specific mode of the two treatment conditions
Treatment Description and Fidelity of Implementation varied. That is, families in the PRT condition were taught
to use motivational techniques to facilitate verbal com-
Child participants were scheduled to receive a total of munication in their children, while families in the PECS
258 h of treatment with either PRT or PECS. During the condition were taught to use similar to techniques to
course of the study children actually received an average of facilitate augmentative communication in their children.
247 h of treatment (range = 181–263). For the first
15 weeks, parents participated in 2 weekly, 2-h parent Pivotal Response Training (PRT)
education sessions with their child in the laboratory and
children received an additional five 2-h sessions per week Parents and therapists were trained to target the develop-
in the home. This was followed by 8 weeks of one 2-h ment and spontaneous use of functional spoken language.
parent education session per week and two 2-h sessions per Training followed the sequence of the PRT training manual
week in the home. Procedures for both conditions were (Koegel et al. 1987). The reader is directed to the manual
based on their respective treatment manuals (Frost and for a more detailed description of treatment guidelines.
Bondy 2002; Koegel et al. 1989).
Picture Exchange Communication System (PECS)
In-Home Treatment
Parents and therapists were trained to teach children to use
Undergraduate student therapists trained in PECS and PRT picture icons to communicate. Training followed the
provided the intervention to the children. Therapists were sequence of the PECS training manual (Frost and Bondy
trained to criterion (80 % correct usage of all treatment 2002). Procedures, in accordance with the manual,

123
J Autism Dev Disord (2014) 44:1244–1251 1247

included the use of a cloze procedure (e.g., ‘‘I want Inventory (CDI; Fenson et al. 2006), a standardized parent
_____’’) as a prompt for the child to engage in spoken report instrument of early language competence that mea-
language during later phases of the intervention. The reader sures both receptive and expressive communication. Raw
is directed to the manual for a more detailed description of scores for words produced on this vocabulary checklist
treatment guidelines. were utilized, as standard scores are not available for this
assessment.
Outside Interventions
Adaptive Communication
To further characterize the children’s intervention and to
establish that the two groups did not differ systematically Parents completed the Vineland Adaptive Behavior Scales,
from one another, we kept track of the amount and type of 2nd Ed, (VABS; Sparrow and Cicchetti 1989), a stan-
outside treatments the participants received. Number of dardized measure used to assess the child’s competence
hours weekly of outside speech therapy, occupational and independence in his/her daily living environment. Our
therapy, preschool/daycare, and in-home early intervention analyses included standardized scores with a mean of 100
were monitored via parental report. and standard deviation of 15 in only the communication
subdomain.
Setting and Materials
Augmentative Communication
Parent education was conducted in small playrooms that
included a variety of toys specific to the child’s preferences The phase of PECS being taught at the end of treatment
and developmental level. A generalization setting at each was used as a measure of augmentative communication for
site contained a sofa, chairs and a coffee table. No inter- children assigned to the PECS condition. Children received
vention occurred in the generalization setting. Child treat- a score of 1 through 6 depending upon which phase they
ment was conducted in the child’s home. were currently learning (but had not yet mastered).

Dependent Measures Parent Satisfaction

Spoken Language At post intervention parents completed a satisfaction sur-

vey consisting of questions regarding their overall
The Mullen Scales of Early Learning (MSEL; Mullen impressions of the program including program effective-
1995) measures cognitive ability in a variety of domains ness, intervention techniques, child improvements, and the
for children ages birth to 68 months. For the purposes of parent education format. Each area had several questions
this study we examined data from the expressive language and parents were asked to rate each question on a scale of
scale, with a mean of 50 and standard deviation of 10, and 1–7.
the early learning composite with a mean of 100 and a All intake assessments and dependent measures were
standard deviation of 15. administered by trained staff. Dependent measures were
completed at pre-treatment, post-treatment and at a
Spoken Vocabulary 3-month follow-up. Staff conducting the assessments were
not involved in provision of intervention for the child they
Children were assessed using a standardized measure of assessed. Three-quarters of ADOS assessments were
spoken vocabulary, the Expressive One-Word Picture scored by blind coders from an outside research laboratory
Vocabulary Test-Revised (EOWPVT; Gardner 1990) (i.e., naı̈ve as to assigned condition) across sites, time
which provides a measure of a child’s expressive vocabu- points and condition. These coders had specialized exper-
lary with respect to population norms. Standardized scores tise in the administration and scoring of this instrument. All
are available with a mean of 100 and standard deviation of assessors were blind to condition for pre-treatment
15. A large majority of children in our sample (i.e., 37) assessments as random assignment was conducted after
were unable to establish a basal score. To facilitate intake assessments were complete. Half of all intake and
appropriate analyses of the data, children were categorized dependent assessments conducted at Site 1 were conducted
based on scores falling into one of five ordinal categories by assessors blind to condition at post and follow-up.
that correspond with the assessments’ normal distributions While blind assessments were not conducted at Site 2,
(No score, 55–70, 71–85, 86–100, [100). Parents com- there were no across-site differences in scores by group or
pleted the MacArthur Communicative Developmental time period across the sites.

123
1248 J Autism Dev Disord (2014) 44:1244–1251

Table 2 Children’s mean scores on standardized assessments at program entry and exit
Pre treatment mean (SD) Post treatment mean (SD) Follow-up mean (SD)
PECS PRT PECS PRT PECS PRT

Mullen Scales of Early Learning (n = 38)

Expressive communication subtest (SS) M = 50 SD = 10
Expressive comm. 20.3 (3.2) 18.5 (2.8) 26.7 (12.7) 22.5 (7.0) 28.7 (16.5) 23.7 (11.2)

MacArthur CDI (n = 35) raw number of words

Words produced 5.3 (9.4) 11.9 (20.5) 88.7 (105.5) 83.2 (88.3) 129.8 (117.9) 113.3 (108.3)

Vineland Adaptive Behavior Scales (n = 35) (SS) M = 100 SD = 15

Communication 62.2 (4.7) 60.2 (7.5) 67.1 (14.9) 59.7 (9.6) 68.4 (14.5) 62.6 (12.7)

Results no Group 9 Time interaction), we collapsed results for

both conditions across time period. F. and p-values, and
Main and Interaction Effects of Time and Treatment effect sizes presented in Table 3. Chi square and p-values
for the EOWPVT are also presented in Table 3.
Results are summarized in Table 2. A per protocol analysis For each dependent measure, a main effect of time
was followed, resulting in the exclusion of the two par- indicated improvement in child behavior in areas of spoken
ticipants who were initially enrolled in the study but dis- language, adaptive communication, and spoken vocabu-
continued. No data beyond pre-treatment assessments were lary. When collapsed across conditions, gains on each of
available for either participant. Standardized assessment these assessments from pre-treatment to post-treatment to
gains were analyzed with a 3 9 2 (Time 9 Treatment) follow up were statistically significant. In many cases,
mixed model repeated measures ANOVA with Green- effect sizes were quite large, especially for vocabulary, the
house-Geisser correction for all measures except the EO- main target of the intervention. There was no main effect of
WPVT. Type I error probability was maintained at .05 treatment type for any assessments. No Time 9 Treatment
(two-tailed) for all analyses. Main and interaction effects of interaction was found. Standard deviations were quite large
time and treatment condition were analyzed. Means by for a majority of measures indicating a heterogeneous set
condition and time period are listed in Table 2. Gains on of individual response patterns in both conditions.
the EOWPVT were analyzed with a 3 9 2 (Time 9
Treatment) ordinal mixed factor model. At intake, 95 % of Augmentative Communication—PECS Phase
children in both groups received no basal score on the
EOWPVT. At exit, approximately 50 % of children Of the 19 children in the PECS condition, 12 reached Phase
(range = 45–53 %) obtained a basal and completed the 6, that is, they had mastered requesting and were learning
assessment. The frequencies for each category by condition to comment using pictures. Two children reached Phase 5
are available from the authors. (responding and attributes), two reached Phase 4
Since results for all of the assessments in the ANOVA (requesting items using a sentence strip), two reached
followed the same pattern when analyzed separately (sig- Phase 3 (learning to discriminate pictures), and one reached
nificant effects of Time within each treatment group with Phase 2. The three children in Phases 2 and 3 were not

Table 3 Statistical analyses of

Time F, p, Effect sizea Treatment F, p Time 9 Treatment F, p
change over time and across
conditions Mullen Scales of Early Learning (n = 38)
Expressive comm. 9.95, .000, .216 1.775, .191 .551, .510
MacArthur CDI (n = 35)
Words produced 31.26, .000, .486 .045, .833 .313, .645
Vineland Adaptive Behavior Scales (n = 35)
Communication 4.09, .037, .110 2.263,.142, 1.765, .190
Expressive One Word Picture Vocabulary Test (n = 39 pre/post and 37 at follow-up; Chi-sq; df; p)
a
Effect sizes are reported for 26.637, 2, .001 .490, 1, .484 .027, 2, .987
significant results only

123
J Autism Dev Disord (2014) 44:1244–1251 1249

using spoken language to communicate. Of the children variability in the verbal progress of enrolled children. At
who reached Phase 4 or higher, all but four of them (one in the final measurement period, 78 % of the children across
Phase 4 and three in Phase 6) were reported to have more both treatment groups, were reported to use at least 10
than 10 words on the CDI. Of the children who reached spoken words.
Phase 6, five of them had expressive language age equiv- In the PECS condition 79 % of children learned to use
alent scores on the MSEL of less than 12 months, indi- the system functionally and reached Phase 6 (commenting).
cating the possibility that their complexity of Although each of the PECS phases is not specifically
communication was higher when using PECS. associated with a typical age equivalent, the commenting
phase requires skills similar to those seen in typically
Parent Satisfaction developing children’s spoken language at approximately
15–20 months. Approximately 42 % of the children in the
Parents were satisfied with the intervention overall, with PECS condition may have been using their picture system
mean ratings of 5.7 (1 being very dissatisfied and 7 being in a more complex manner than their spoken language.
very satisfied) for PRT and 6.0 for PECS. Overall However, this needs to be explored further in future
improvement across all areas of communication, self help research to examine generalization of skills and use of
skills, and behavior was moderate with mean ratings of 4.4 specific types of communicative functions. Similar to other
for PRT and 4.5 for PECS (1 being no improvement and 7 studies, (e.g., Charlop-Christy et al. 2002) children in the
extreme improvement). Parents rated the general teaching PECS condition often began to use spoken language once
format and usefulness of the program highly at 6.3 for PRT they reached Phase 4, which includes the use of cloze
and 6.3 for PECS (1 being not useful and 7 being very procedures and expectant waiting for speech production.
useful). Parents rated the specific intervention strategy Parents in both groups were satisfied with the inter-
difficulty at 5.6 for PRT and 4.6 for PECS (1 being very vention and reported progress in their children. However,
difficult and 7 being not difficult). Difficulty of the inter- parents found PECS to be more difficult to implement in
vention strategies was the only statistically significant the home. Perhaps the more effort required to prepare
difference between PRT and PECS (F (1,28) = 9.413, PECS icons, prepare PECS books and having to ensure the
p = .005). child had his PECS book with him, etc., contributed to this
reported increased difficulty. This is in contrast to PRT
Outside Interventions where materials in the current, natural environment were
all that was required.
We conducted analyses of weekly number of hours the Reportedly, some parents and practitioners have been
participants received other treatment while participating in reluctant to recommend augmentative communication
this investigation. T test analyses of speech therapy and systems for children with autism, fearful that these systems
occupational therapy (PRT: .94 h/week, PECS: .94 h/ may interfere with the development of spoken language.
week), preschool/daycare (PRT: .3, PECS: 1.5) and in- Given the randomized comparison design, these findings
home early intervention (PRT: 2.4, PECS: 3.4) indicated no suggest that PECS may be as effective as naturalistic verbal
significant difference in the amount of these treatments language training programs such as PRT for facilitating
received across the two conditions. language. It is noteworthy that PECS did not inhibit growth
in spoken language. However, due to the variability in
child progress these data also raise questions regarding
Discussion when to use which methodology. It is hoped future research
in this area will assist our efforts to individualize treatment
This study provided a systematic, randomly controlled protocols in this area.
comparison of two empirically-validated behavioral treat- As was expected, average change on standardized
ments, PRT and PECS. Due to the lack of a usual care assessments misrepresented a wide variability of treatment
control group we cannot draw conclusions regarding the response. Children receiving PECS and PRT shared a
relationship between gains in spoken language and either strikingly similar pattern of responsivity for spoken lan-
of the treatment conditions. One intervention modality was guage outcomes. A unique contribution of the present
not superior to the other. Results indicate that based on study, as compared to most previous treatment studies, is
mean scores, the 2- to 3-year-old, nonverbal and minimally that many of the participants were under 3 years of age and
verbal children in both the PRT and PECS conditions made minimally verbal. Given that even at this early age,
similar gains in spoken communication. On average, chil- approximately 50 % of children had good outcomes while
dren gained approximately 80 spoken words across the others made slower progress, it may be that alternative
6-month study period, although we found extreme strategies are needed for these children to jump-start their

123
1250 J Autism Dev Disord (2014) 44:1244–1251

progress. Further studies and further analyses of the current treatment when change to the other system should be
data set will examine the specific predictors of positive considered. Further, an important variable in this decision
outcomes in an effort to determine whether child charac- will undoubtedly be child characteristics associated with
teristics may indicate differential responding to either success with either system. For example, Yoder and Stone
intervention. (2006a) found that PECS was superior for the development
In the only other direct comparison of PECS and a of nonimitative words for children who entered treatment
naturalistic behavioral verbally-based approach (RPMT), with higher levels of object exploration while RPMT was
Yoder and Stone (2006a, b) found a main effect in favor of superior for children who entered with relatively low object
PECS for spoken vocabulary measures in a similar sample exploration. Thus research focusing on child characteristics
of participants as the present study. This contrasts our associated with response to these treatment modalities will
findings of no difference between the conditions. It is serve to inform early intervention targeting communication
interesting to speculate as to why we failed to replicate in young minimally verbal children with autism and allow
Yoder and Stone’s (2006a, b) findings. One possibility is interventionists to provide more tailored interventions.
that the participants in the studies differed on some key
characteristics (object exploration, social initiation, avoid- Acknowledgments This research was supported in part by
U.S.P.H.S. Research Grants MH 39434 and MH 28210 from the
ance) that were not assessed in the current work. It is also National Institute of Mental Health. The authors acknowledge the
possible that the different measures utilized in the current ongoing support and important contributions of Drs. Andrew Bondy
study (i.e., natural language samples versus standardized and Gail McGee for providing consultation and supervision to ensure
measures and parent report) may have contributed to the appropriate program procedures. We also wish to thank Dr. Allison
Jobin for her valuable contributions to the preparation of this manu-
failure to find a main effect for condition. Given that to script and to Dr. Mark Appelbaum for statistical consultation.
date there have been so few randomized clinical trials of
these communication interventions it is likely that further
research directed at these important questions will yield References
clarification of these, as well as other, issues.
There are some limitations to this study. As noted above American Psychiatric Association. (2000). Diagnostic and statistical
the present study did not include a no-treatment control manual of mental disorders (4th ed.). Washington, DC: Author.
Bondy, A. S., & Frost, L. A. (2001). The picture exchange
group. Therefore treatment gains may have been due to communication system. Behavior Modification, 25, 725–744.
development rather than a function of either intervention. Charlop-Christy, M., Carpenter, M., Le, L., LeBlanc, L., & Kellet, K.
Further research will be necessary in order to address (2002). Using the picture exchange communication system
whether gains maintained across time and environments. (PECS) with children with autism: Assessment of PECS
acquisition, speech, social- communicative behavior, and prob-
This study was limited to only nonverbal and minimally lem behavior. Journal of Applied Behavior Analysis, 35,
verbal young children with autism and therefore may not 213–231.
generalize to children who are verbal at a young age. A Dawson, G. (2008). Research: Early behavioral intervention, brain
subset of assessments was conducted by coders who were plasticity, and the prevention of autism spectrum disorder.
Development and Psychopathology, 20, 775–803.
aware of the treatment condition, which represents a threat Fenson, L., Marchman, V., Thal, D., Dale, P., Reznick, S., & Bates, E.
to the internal validity of the findings. (2006). MacArthur-Bates Communicative Development Inven-
tories (CDIs) (2 ed.): Brookes Publishing.
Frost, L., & Bondy, A. (2002). The picture exchange communication
system training manual (2nd ed.). Newark, Delaware: Pyramid
Conclusions Educational Products.
Ganz, J. B., Simpson, R. L., & Corbin-Newsome, J. (2007). The
The results of this study suggest that a majority of young, impact of the picture exchange communication system on
nonverbal or minimally verbal children with autism will requesting and speech development in preschoolers with autism
spectrum disorders and similar characteristics. Research in
learn to use spoken language at a young age. This is a very Autism Spectrum Disorders, 2, 167–169.
promising finding that is consistent with current early Gardner, M. F. (Ed.). (1990). Expressive one-word picture vocabulary
intervention literature. These findings indicate that PECS test. Novato, CS: Academy Therapy Publications.
and PRT may both be useful in improving children’s Koegel, R., Camarata, S., Koegel, L., Ben-Tall, A., & Smith, A.
(1998). Increasing speech intelligibility in children with autism.
communication, however further study is needed due to the Journal of Autism and Developmental Disorders, 28, 241–251.
lack of differences between the groups. The PECS system Koegel, R. L., Koegel, L. K., & Surratt, A. (1992). Language
did not appear to inhibit or facilitate the development of intervention and disruptive behavior in preschool children with
spoken communication. Decisions regarding whether to autism. Journal of Autism and Developmental Disorders, 22,
141–153.
recommend one system over another will likely ultimately Koegel, R. L., O’Dell, M. C., & Koegel, L. K. (1987). A natural
be determined by research focusing on generalization and language teaching paradigm for nonverbal autistic children.
maintenance of communication gains, and the point in Journal of Autism and Developmental Disorders, 17, 187–200.

123
J Autism Dev Disord (2014) 44:1244–1251 1251

Koegel, R. L., Schreibman, L., Good, A., Cerniglia, L., Murphy, C., & augmented and nonaugmented language interventions for tod-
Koegel, L. K. (Eds.). (1989). How to teach pivotal behaviors to dlers with developmental delays and their parents. Journal of
children with autism: A training manual. CA: Santa Barbara. Speech, Language, and Heraring Research, 53, 350–364.
Laski, K. E., Charlop, M. H., & Schreibman, L. (1988). Training Sparrow, S. S., & Cicchetti, D. V. (1989). The Vineland adaptive
parents to use the natural language paradigm to increase their behavior scales. In C. S. Newmark (Ed.), Major psychological
autistic children’s speech. Journal of Applied Behavior Analysis, assessment instruments (Vol. 2, pp. 199–231). Needham
21, 391–400. Heights, MA: Allyn and Bacon.
Lord, C., Risi, S., Lambrecht, L., Cook, E. H. J., Leventhal, B. L., Sze, K., Koegel, R. L., Brookman, L., & Koegel, L. K. (2003). Rapid
DiLavore, P. C., et al. (2000). The autism diagnostic observation acquisition of speech in nonverbal children with autism:
schedule-generic: A standard measure of social and communi- Developing typical, social, and communicative interactions in
cation deficits associated with the spectrum of autism. Journal of children with autism using pivotal response training and self-
Autism and Developmental Disorders, 30, 205–223. management. Paper presented at the Association for Behavior
Lord, C., Rutter, M., & Le Couteur, A. (1994). Autism diagnostic Analysis.
interview-revised: A revised version of a diagnostic interview for Tager-Flusberg, H., Paul, R., & Lord, C. (2005). Language and
caregivers of individuals with possible pervasive developmental communication in autism. In D. Cohen & F. Volkmar (Eds.),
disorders. Journal of Autism and Developmental Disorders, 24, Handbook of autism and pervasive developmental disorders (3rd
659–685. ed., Vol. 1, pp. 335–364). New Jersey: John Wiley and Sons Inc.
Mirenda, P., & Iacono, T. (1988). Strategies for promoting augmen- Yoder, P., & Stone, W. L. (2006a). A randomized comparison of the
tative and alternative communication in natural contexts with effect of two prelinguistic communication intervention on the
students with autism. Focus on Autism and Other Developmental acquisition of spoken communication in preschoolers with ASD.
Disabilities, 3, 1–16. Journal of Speech, Language, and Hearing Research, 49,
Mullen, E. (Ed.). (1995). Mullen scales of early learning (AGS ed.). 698–711.
Circle Pines, MN: AGS Publishing. Yoder, P., & Stone, L. (2006b). Randomized comparison of two
National Research Council. (2001). Educating children with autism. communication interventions for preschools with autism spec-
Washington DC: National Academy Press. trum disorders. Journal of Consulting and Clinical Psychology,
Romski, M., Sevcik, R. A., Adamson, L. B., Cheslock, M., Smith, A., 74, 426–435.
Barker, R. M., et al. (2010). Randomized comparison of

123
Reproduced with permission of the copyright owner. Further reproduction prohibited without
permission.