Short Report
Social–communicative effects of the Picture Exchange Communication
System (PECS) in Autism Spectrum Disorders
Anna Lerna†, Dalila Esposito†, Massimiliano Conson‡, Luigi Russo† and Angelo Massagli†
†Scientific Institute I.R.C.C.S. “Eugenio Medea" Regional Branch of Ostuni, Brindisi Department of Neurorehabilitation 2,
Child Psychiatry, Brindisi, Italy
‡Neuropsychology Laboratory, Department of Psychology, Second University of Naples, Caserta, Italy
Abstract
Background: The Picture Exchange Communication System (PECS) is a common treatment choice for non-
verbal children with autism. However, little empirical evidence is available on the usefulness of PECS in treating
social–communication impairments in autism.
Aims: To test the effects of PECS on social–communicative skills in children with autism, concurrently taking
into account standardized psychometric data, standardized functional assessment of adaptive behaviour, and
information on social–communicative variables coded in an unstructured setting.
Methods & Procedures: Eighteen preschool children (mean age = 38.78 months) were assigned to two intervention
approaches, i.e. PECS and Conventional Language Therapy (CLT). Both PECS (Phases I–IV) and CLT were
delivered three times per week, in 30-min sessions, for 6 months. Outcome measures were the following: Autism
Diagnostic Observation Schedule (ADOS) domain scores for Communication and Reciprocal Social Interaction;
Language and Personal–Social subscales of the Griffiths’ Mental Developmental Scales (GMDS); Communication
and Social Abilities domains of the Vineland Adaptive Behavior Scales (VABS); and several social–communicative
variables coded in an unstructured setting.
Outcomes & Results: Results demonstrated that the two groups did not differ at Time 1 (pre-treatment assessment),
whereas at Time 2 (post-test) the PECS group showed a significant improvement with respect to the CLT group
on the VABS social domain score and on almost all the social–communicative abilities coded in the unstructured
setting (i.e. joint attention, request, initiation, cooperative play, but not eye contact).
Conclusions & Implications: These findings showed that PECS intervention (Phases I–IV) can improve social–
communicative skills in children with autism. This improvement is especially evident in standardized measures of
adaptive behaviour and measures derived from the observation of children in an unstructured setting.
Keywords: Picture Exchange Communication System (PECS), autism, intervention, social and communicative
skills.
Address correspondence to: Angelo Massagli, Department of Neurorehabilitation 2, Child Psychiatry Scientific Institute I.R.C.C.S.
‘Eugenio Medea’, Regional Branch of Ostuni–Brindisi, Di Summa Square, I-72100 Brindisi, Italy; e-mails: angelo.massagli@os.lnf.it and
angelo.massagli@tin.it
International Journal of Language & Communication Disorders
ISSN 1368-2822 print/ISSN 1460-6984 online c 2012 Royal College of Speech and Language Therapists
DOI: 10.1111/j.1460-6984.2012.00172.x
610 Anna Lerna et al.
Introduction impairments of functional speech (Flippin et al. 2010),
Communication and social impairments appearing only a small number of studies suggest positive effects
early in life and persisting into adulthood are core in the social domain (Preston and Carter 2009).
features of Autism Spectrum Disorders (ASD; American In a study on three individuals with autism,
Psychiatric Association (APA) 2000, Howlin et al. Charlop-Christy et al. (2002) demonstrated positive
2004). In the past, rehabilitation interventions aimed treatment-related increases in speech and different
at improving language and communication in children social–communicative behaviours, especially initiations,
with ASD mainly focused on verbal production through requests and joint attention. Accordingly, in a single case
response training, while neglecting that, unlike non- study Kravits et al. (2002) reported some increase in
verbal training, this method did not influence communi- duration of social interaction with peers and provided a
cation skills (Howlin 1998). At present, successful demonstration of the effectiveness of PECS in increasing
rehabilitation strategies of non-verbal autism are centred spontaneous communication skills.
on functional spontaneous communication. This has In a randomized control study, Yoder and Stone
favoured implementation of Alternative Augmentative (2006) compared the effect of PECS with that of
Communication (AAC) systems devoted to the develop- another communicative intervention (i.e. Responsive
ment of non-verbal communication by using signs, Education and Prelinguistic Milieu Teaching—RPMT),
pictures and symbols (Howlin 1998, Howlin et al. 2007, on initiating joint attention, requesting and turn-taking,
Mirenda 2003, Schuler et al. 1997). Various forms and found that RPMT facilitated the frequency of
of AAC have been devised. These include the use of generalized turn-taking and generalized initiating joint
manual signs, voice output communication devices and attention more than the PECS did, whereas the PECS
various picture-based systems (for a recent meta-analysis had a superior effect to generalized request in children
on AAC in ASD, see Ganz et al. 2011). The Picture with initially low rates of initiating joint attention. Carr
Exchange Communication System (PECS), originally and Felce (2007) focused on spontaneous communica-
developed by Bondy and Frost (1994), is a picture- tive initiation following PECS training and found
based AAC system that was specifically designed for significant increases in initiation and dyadic interac-
non-verbal children with ASD (for recent reviews, see tion between children with autism and their teachers.
Flippin et al. 2010, and Ganz et al. 2011). PECS Analogously, in another randomized control study
uses basic behavioural principles, particularly reinforce- Howlin et al. (2007) investigated potential effects
ment techniques, to teach children to use functional of PECS training on spontaneous communication of
communication in a social interaction context (Bondy children with ASD in a school setting and found that
and Frost 1994, 1998). rates of communicative initiations and PECS usage were
The PECS protocol includes six instructional phases significantly increased immediately following interven-
(Bondy and Frost 1998): Phases I and II include teaching tion (also Gordon et al. 2011). The authors also
the student to exchange pictures for preferred items examined ADOS-G (Lord et al. 2000) domain scores
and activities (Phase II expands on Phase I by teaching for communication and reciprocal social interaction,
generalization across distance, communicative partners and only observed a specific decrease in the severity
and a variety of items and pictures); in Phase III score for reciprocal social interaction at the 10-month
students learn to discriminate between preferred and follow-up.
non-preferred pictures and eventually between a variety This review shows that only three group studies
of preferred items; in Phase IV students are taught are available on social–communicative effects of PECS
to make requests using complete sentences; Phase V training in ASD. In two studies a specific analysis
involves instruction in answering questions; and in the was performed on behavioural measures derived
final phase, Phase VI, students learn to make a variety of from observation of participants in an unstructured
comments. Thus, the first four phases focus on teaching setting, without providing a systematic assessment with
the child to use a symbol to make a request, while the standardized tests (Carr and Felce 2007, Yoder and
final two phases expand the range of communicative Stone 2006). In the third study, a less deep behavioural
functions used by the child. observation has been combined with standardized tests
Despite the fact that PECS is a popular specifically assessing language and only one measure of
communication-training programme for non-verbal social–communicative abilities (i.e. ADOS Communi-
children with autism (Flippin et al. 2010, Howlin et al. cation and Reciprocal Social Interaction domains;
2007), only a few empirical studies are available on its Howlin et al. 2007). Recently, Preston and Carter (2009)
efficacy in treating social–communication impairments claimed that investigations on social–communicative
in autism. More precisely, while several investiga- effects of PECS training should concurrently take
tions indicated that PECS can provide a viable means into account standardized psychometric data, standard-
to facilitate communication and to improve severe ized functional assessment of adaptive behaviour, and
Social–communicative effects of PECS in ASD 611
information on social–communicative variables coded Table 1. Means and standard deviations (SD) of demographic
in an unstructured setting. On this basis, in the present variables, GMDS Non-verbal IQ and ADOS Total score,
separately for the two groups
study we conducted an experimentally controlled group
investigation to test the effects of the first four PECS PECS CLT
phases on social–communicative behaviours of children Variable Mean SD Mean SD
with ASD. To this aim, we collected psychometric
data and functional measures of adaptive behaviour Chronological age (months) 38.78 7.43 41.11 8.67
Mother’s education (years) 13.56 3.00 11.89 2.21
by means of standardized tasks, and evaluated social– Father’s education (years) 11.67 3.84 10.44 3.17
communicative skills in an unstructured setting through GMDS Non-verbal IQ 73.56 18.15 68.56 16.97
observation of children’s free-play interactions with an ADOS Total score 16.78 4.06 16.11 3.86
adult.
receptive and expressive language, and personal– (version XT.7; Noldus Information Technology 2007)
social subscale, assessing proficiency in the software package to transform a standard computer
activities of daily living, level of independence and keyboard into a point and state events recorder. The
interaction with other children, were employed as point events were behaviours that only take an instant
outcome measures. in time, or whose duration was not important, i.e. joint
• Autism Diagnostic Observation Scale (ADOS; Lord attention, requests and initiation; these variables were
et al. 1999). Communication and Reciprocal coded in terms of frequency. The state events were
Social Interaction domain scores were used to behaviours that take a period of time and therefore
assess change over time. have a duration, i.e. cooperative play, eye contact; these
• Vineland Adaptive Behavior Scales, Second Edition variables were coded in seconds. The duration of the
(VABS; Sparrow et al. 2005). VABS were used to session was exactly 15 min for all children; therefore
gather parent report of child communication and no correction for the length of the observation was
social abilities at home and in the community. necessary.
• Unstructured Free-Play with Examiner (UFPE). Reliability of coded variables was analysed on
Following previous studies on social– independently coded and randomly selected samples.
communicative effects of PECS (Charlop-Christy Two independent observers were trained to use the
et al. 2002, Yoder and Stone 2006) a 15-min coding system developed for the purposes of this study.
session of UFPE was planned in order to collect Each of the observers had a Master’s degree in clinical
data on specific variables, i.e. cooperative play, eye or developmental psychology and received training by
contact, joint attention, requests and initiation. an experienced clinician using video samples of children
The operational definition of these variables was not involved in the study. The observers (coders) were
based on Charlop-Christy et al.’s (2002) criteria blind to the purposes of the study and the treatment
(table 2). UFPE was conducted in a room with assignment. Inter-rater reliability was determined by
a variety of age appropriate toys or objects (i.e. double coding of 20% of the observations and was very
bubbles, balloons, stuffed and rubber animals, good (Kappa = 0.87).
female baby doll, building blocks, wind up car,
electronic toys, a small table and two chairs)
accessible to the children. The PECS book was Treatments
present in pre- and post-treatment and both PECS and CLT were implemented within a psycho-
groups had the possibility to use it. The examiner education rehabilitation programme based on the
played with the same or similar toy as the child TEACCH methodology (Mesibov et al. 2004).
by imitating the child’s play. If the child did not All children followed a structured teaching for a
attend to any toy for 10 s, then the examiner total 12 h a week; children were taught skills
selected an interesting object and used it in a way including attention, basic discrimination, language and
that was suitable to the cognitive level of the child. spontaneous communication, daily living, socialization,
Examiners verbally commented on the child’s or play, fine and gross motor control and pre-academics
their own actions, but they were instructed not to (Watson et al. 1989). A semi-structured setting was
use any type of communication prompt, such as also implemented, using incidental teaching techniques,
time delays, questions or gestural prompts. to enhance generalization, increase motivation, develop
social skills and to decrease problem behaviours.
Social–communicative behaviours coded during the Treatments sessions took place in a therapeutic
UFPE session were derived from videotaped records room equipped with small desks, chairs, familiar and
of the free-play session using the Observer Video Pro unfamiliar objects and toys; children were offered 30-
Social–communicative effects of PECS in ASD 613
Table 3. Brief description of the PECS phases implemented in the study
Phase Description
I: Picture exchange The child is taught to approach a communicative partner, give him or her a picture card, and receive a
preferred item
II: Increasing distance The child is taught to retrieve a picture from his PECS board, which had been moved away from him,
walk to a communicative partner and persist in handing the picture to the communicative partner,
who had moved across the room
III: Picture discrimination The child is taught to discriminate among multiple pictures on the PECS board. At the end of this
phase the child can choose from six or more preferred and non-preferred items
IV: Sentence structure The child creates a ‘sentence’ on the sentence strip by combining the ‘I want card’ and the card of a
desired item, seeks out the communicative partner, and gives him or her the sentence strip
min individual therapy sessions (PECS or CLT) three reinforcements. In Italy it represents one of the
times a week for 6 months (i.e. 72 sessions). most widely employed methods to treat language and
Experts leading both PECS and CLT were special- communicative disorders in developmental disabili-
ists in speech–language pathology with extensive ties in accordance with the guidelines of the Italian
expertise with ASD. As specifically regards PECS, National Health System. CLT employs a variety of
the trainer attended two workshops about PECS largely operant approaches to language intervention
basic and advanced training provided by an expert implemented in accordance with discrete trial training
consultant of Pyramid Educational Consultants UK, formats (Goldstein 2002, Lovaas 1981, 2002): the child
following the highly prescribed format of the training is taught to attend to adults and respond to simple
manual (Frost and Bondy 2002), and received a instructions (receptive language training); she/he has to
certificate of attendance at the conclusion of each first imitate manual, oral-motor and vocal behaviour,
workshop. and then to imitate speech. Association learning is
Phases I–IV of PECS were implemented according used for teaching increasingly sophisticated expressive
to the standard procedures devised by Bondy and Frost language skills, and motivation is favoured through use
(1998); table 3 provides a brief description of each the of various external rewards. CLT uses didactic, adult-
four PECS phases. Before beginning of the training, directed instructions in a one-to-one interaction: each
parents were required to record their children’s favourite session begins by presenting the child with his own
food and toys on a card in order to make picture cards preferred items. If the child shows motivation to obtain
to be included in the communication book, according the item (e.g. reaching for it), the therapist waits for
to the standard procedure. a few seconds to provide the opportunity for the child
The percentage of correct requests per session and to say the word independently. If the child does not
the number of sessions per criterion were measured produce the word, the therapist repeats the name of
for each of the four PECS phases. Observations were the object three times with time delay. When the child
carried out independently by two autism specialists, spontaneously names or imitates the name, or produces
speech-and-language therapists, one of which was the a word approximation, she/he receives the desired
PECS trainer. There was 100% agreement regarding the item.
PECS level between the two assessors. The criterion for
advancement within a phase was set at 80% independent
on symbol exchanges made during PECS instruction. Results
Training within each phase of PECS was continued until Standardized test assessment
all children were able to demonstrate independently a
correct response score of 80% or higher for a minimum We investigated group differences before treatment. To
of three consecutive sessions, in accordance with the this aim, a MANOVA was performed with demographic
criteria outlined in the PECS training manual (Bondy characteristics and mean standard scores on formalized
and Frost 1994). The mean number of PECS sessions tests at Time 1 (tables 1 and 4) as dependent variables
for each phase was 6.3 (SD = 2.3); within these sessions, and group as independent variable. Results showed
children were provided with a mean number of 91 that there were no significant group differences in
(SD = 38.9) individual trials. demographic data and Time 1 formalized tests scores (all
Conventional Language Therapy (CLT) is a p > 0.05), thus demonstrating that the two groups were
language training based on a systematic, step-by- well matched with respect to pre-treatment measures
step teaching technique using prompts and useful and that no covariate was needed to analyse group
differences on Time 2 measures.
614 Anna Lerna et al.
Table 4. Means and standard deviations (SD) of standard scores on formalized outcome measures at Times 1 and 2 separately for the
two groups
Time 1 Time 2
PECS CLT PECS CLT
Variable Mean SD Mean SD Mean SD Mean SD
GMDS
Language 35.00 7.98 36.00 13.18 38.44 14.70 36.67 17.43
Social 61.78 8.00 53.56 12.03 63.00 8.41 55.22 11.17
ADOS
Communication 6.44 1.74 6.22 2.39 5.89 3.44 6.11 2.09
Reciprocal Social Interaction 10.11 3.02 10.22 1.99 9.56 2.40 9.89 1.76
VABS
Communication 56.33 13.20 55.67 10.79 64.89 8.64 60.56 17.08
Social 64.56 6.95 60.00 3.46 82.67 19.91 60.78 3.67
Table 5. Means and standard deviations (SD) of behavioural outcome measures coded during free-play at Times 1 and 2 separately
for the two groups
Time 1 Time 2
PECS CLT PECS CLT
Variable Measure Mean SD Mean SD Mean SD Mean SD
Cooperative play Duration 40.56 30.06 32.56 36.41 209.78 131.72 74.89 63.73
Eye contact Duration 4.00 4.18 9.11 16.21 19.22 21.22 26.00 39.46
Joint attention Frequency 1.00 1.12 1.11 2.03 4.11 3.86 0.44 0.53
Requests Frequency 1.78 2.22 3.44 2.35 6.44 4.72 2.67 2.12
Initiation Frequency 0.22 0.44 0.67 1.41 3.67 4.00 0 0
A MANOVA was performed with mean standard (table 5) showed significant group differences in
scores on formalized outcome measures at Time 2 cooperative play, F (1, 16) = 7.648, p = 0.014,
(table 4) as dependent variables and group as η2 p = 0.323, joint attention, F (1, 16) = 7.993,
independent variable. Results showed significant group p = 0.012, η2 p = 0.333, requests, F (1, 16) = 4.797,
differences on VABS—Social, F (1, 16) = 10.519, p = 0.044, η2 p = 0.2312, and initiation, F (1,
p = 0.005, η2 p = 0.397, but not on all the other Time 16) = 7.562, p = 0.014, η2 p = 0.321, but not in eye
2 measures (all p > 0.05). contact (p > 0.05).
A within-group comparison between Times 1 and A within group comparison between Times 1 and
2 scores was performed separately in PECS and CLT 2 scores was performed separately in PECS and CLT
group. As regards the PECS group, paired t-tests showed group. As regards the PECS group, paired t-tests showed
a significant improvement of scores on both VABS— a general improvement in all the five coded variables
Communication (t(8) = –2.484, p = 0.038) and (cooperative play, t(8) = –3.941, p = 0.004, eye contact,
VABS—Social (t(8) = –3.331, p = 0.010) domains, but t(8) = –2.369, p = 0.045, joint attention, t(8) = –
not on the other measures (all p > 0.050). The same 2.865, p = 0.021, requests, t(8) = –2.682, p = 0.028,
analysis performed on the CLT group did not reveal and initiation, t(8) = –2.486, p = 0.038). The same
significant difference on any of the considered measures analysis performed on the CLT group did not reveal
(all p > 0.050). significant difference on any of the considered measures
(all p > 0.05).