Review Journal of Autism and Developmental Disorders

https://doi.org/10.1007/s40489-019-00184-9

REVIEW PAPER

Naturalistic Interventions to Improve Prelinguistic Communication

for Children with Autism Spectrum Disorder: a Systematic Review
Ashley H. Dubin 1,2,3 & Rebecca G. Lieberman-Betz 4

Received: 8 March 2018 / Accepted: 19 August 2019

# Springer Science+Business Media, LLC, part of Springer Nature 2019

Abstract
The current review identified 11 single-case design (SCD) and 14 group-design evaluations of naturalistic behavioral interven-
tions targeting prelinguistic social communication in young children with autism spectrum disorder (ASD). Three SCD studies
and eight group design studies utilized methodologically rigorous designs and demonstrated a functional relation between the
intervention and child prelinguistic social communication. Results of this systematic review provide information about efficacy
of specific naturalistic behavioral interventions, details about the intervention, and characteristics of the implementation agent
and setting. Such information may be useful to support implementation of evidence-based practices and increase the quality of
future research.

Keywords Autism spectrum disorder . Systematic review . Prelinguistic communication . Intervention

Social communication impairments are often considered the pri- The coordinated use of prelinguistic social communication
mary deficits in Autism spectrum disorder (ASD; e.g., Kasari behaviors to intentionally communicate (Warren et al. 2006)
et al. 2001). One of the earliest recognized social communication can serve the pragmatic functions of showing or labeling (i.e.,
deficits unique to ASD is delayed or absent joint attention (JA; proto-declaratives or initiating joint attention (IJA), defined as
e.g., Mundy et al. 1986), which involves actively coordinating a child use of gaze shifts, gestures, and/or vocalization to
attention between an object or event and another person for the direct another’s attention), requesting (i.e., proto-imperatives
purpose of sharing (Bakeman and Adamson 1984). JA impair- or initiating behavior requests (IBR), which involves behav-
ments may be evident within the first year of life when iors similar to IJA used for the purpose of requesting), and
prelinguistic forms of intentional communication emerge (e.g., social interaction (Bruner 1981). Prelinguistic behaviors (e.g.,
pointing, reaching; Lane and Brown 2016), and differentiate gestures, eye gaze, vocalizations) that comprise intentional
children with ASD from typically developing children and communication for the aforementioned pragmatic functions
children with other developmental delays (e.g., Charman et al. are often impaired in children with ASD (e.g., Chawarska
1997). et al. 2007; Shumway and Wetherby 2009). When compared
with typically developing children and children with other
* Ashley H. Dubin developmental delays, children with ASD display fewer and
Ashley.Dubin@nemours.org less complex forms of intentional communication (e.g.,
Mundy et al. 1990).
Rebecca G. Lieberman-Betz
rglb@uga.edu Given deficits in prelinguistic social communication
across pragmatic functions, children with ASD experience
1
Department of Educational Psychology, University of Georgia, fewer communication and learning opportunities within
Athens, GA, USA the context of social interaction. Reduced participation in
2
Division of Behavioral Health, Department of Pediatrics, Nemours/ social interactions with adults has significant ramifications
A. I. DuPont Hospital for Children, Wilmington, DE, USA for learning and development, as many early skills in-
3
Sidney Kimmel Medical College, Thomas Jefferson University, volved in language and socialization are learned within
Philadelphia, Pennsylvania, USA the context of social interactions between a child and care-
4
Department of Communication Sciences and Special Education, giver or other adult (e.g., Adamson et al. 2004).
University of Georgia, Athens, GA, USA Prelinguistic social communication serving the pragmatic
 Rev J Autism Dev Disord

function of JA has been shown to be concurrently and social interactions as reinforcers, and thus provide a motivat-
predictively related to spoken language (e.g., McDuffie ing context for IJA instruction.
et al. 2005). Such evidence has prompted researchers to Naturalistic behavioral interventions share the goal of pro-
suggest that joint attention may establish the foundation moting generalized behavior change across a variety of skills
for using spoken language to communicate (e.g., Toth (including prelinguistic social communication) through simi-
et al. 2006). In addition, JA is also related to development lar instructional strategies (Kaiser and Trent 2007; Yoder and
of more complex social behaviors in children with ASD Warren 1998, 1999; Yoder and Stone 2006). Examples in-
such as play (e.g., Whalen et al. 2006), imitation (Whalen clude prelinguistic milieu teaching (PMT; Yoder and Warren
et al. 2006), and social competency (e.g., Mundy et al. 1998), pivotal response training (PRT; Koegel et al. 1999),
1994). The variety of communication and learning oppor- and responsive interaction (RI; Kaiser and Delaney 1998).
tunities afforded by JA suggest it may be a pivotal skill These interventions share components including embedding
(e.g., Mundy and Crowson 1997). The classification of instruction within natural environments, following the child’s
JA as a pivotal skill highlights the need for early interven- lead, strategic use and fading of prompts, and providing nat-
tion services that prioritize facilitating JA and associated ural consequences (Kaiser and Trent 2007). Although there is
prelinguistic social communication skills for children no single agreed upon description of naturalistic instruction,
showing deficits associated with ASD (Mundy and naturalistic behavioral interventions share several compo-
Crowson 1997). nents, including embedded instruction in the child’s natural
environment, systematic use of prompts (e.g., time delay),
use of naturally occurring contingencies, following the child’s
lead, shaping, fading, and adult responsiveness (Kaiser and
Early Social Communication Interventions Trent 2007; Schreibman et al. 2015). Reviews indicate empir-
ical support for the use of naturalistic behavioral interventions
Naturalistic interventions are considered effective in teaching to improve prelinguistic social communication skills for chil-
core social communication skills to children with ASD (White dren with ASD (e.g., Goldstein 2002; Prelock et al. 2011;
et al. 2011; Wong 2014) and align with the principles of de- Schreibman et al. 2015; White et al. 2011).
velopmentally appropriate practices for young children. A step towards increased implementation of research-
Compared with highly structured, adult-directed interven- supported naturalistic interventions for prelinguistic social
tions, naturalistic interventions incorporating behavioral communication is to determine the state of the current inter-
learning principles may be more appropriate for teaching vention research for children with ASD. This systematic re-
prelinguistic social communication to very young children view of the literature will provide information about the types
who develop skills within the context of social relationships of interventions that have been researched, the quality of the
and meaningful routine and play activities (Schreibman et al. studies, evidence of behaviors affected by the interventions,
2015). Additionally, key components of naturalistic interven- and common components across effective interventions.
tions, including instruction embedded in the child’s natural Despite the emphasis on intervention research targeting core
environment and the provision of natural consequences, are ASD deficits and evidence-based practices, a systematic re-
consistent with strategies used to program for generalization view of naturalistic interventions used specifically for
(Stokes and Osnes 1989). In naturalistic interventions, skills prelingiustic social communication skills across communica-
are taught using child preferred items and activities that may tive functions has not been conducted. The purpose of the
increase child engagement and therefore motivation to com- current review is to synthesize findings from prior single-
municate. Motivation is especially important to teach gener- case and group-design researches on naturalistic behavioral
alized IJA, or intentional communication for the sole purpose interventions targeting prelinguistic social communication
of sharing attention, for which the natural consequence is so- skills to guide clinical applications and future research. The
cial interaction (Koegel and Koegel 1995). Jones and Carr aims are as follows: (1) identify the types of naturalistic be-
(2004) suggest finding social interaction reinforcing is a pre- havioral interventions that have been used to target
requisite to acquisition of IJA for children with ASD. For prelinguistic social communication skills in children with
many children with ASD, social interaction is not a natural ASD, (2) gather and synthesize evidence for the quality of
reinforcer, and thus may need to be facilitated for interven- studies conducted on the interventions identified in aim one,
tions to produce generalized improvement in IJA (Dube et al. (3) investigate evidence for the effects of naturalistic behav-
2004; Jones and Carr 2004). By first establishing fun social ioral social communication interventions for children with
routines between the child and adult, and teaching ASD while incorporating evidence of study quality, and (4)
prelinguistic communication for the purpose of requesting identify common components used across interventions with
prior to targeting communication for the purpose of sharing demonstrated effects on prelinguistic social communication
attention, naturalistic behavioral interventions may condition behaviors.
Rev J Autism Dev Disord

Method commenting, initiating behavior regulation, requesting, ges-

tures, and gaze shifts). Demographic information for study
A primary systematic search of empirical studies targeted nat- participants was also collected.
uralistic behavioral social communication interventions for
children with ASD published in peer-reviewed journals be- Evaluation Criteria
tween 2001 and 2015. PsycINFO and the Education
Resources Information Center (ERIC) databases were Identified studies were evaluated according to features re-
searched using any combination of the following keywords: lated to methodological rigor, quality and breath of mea-
child or student, autis* or pervasive developmental disorder, surement, and evidence of intervention effects using the
nonverbal or prelinguistic, joint attention, social, What Works Clearinghouse (WWC) evidence standards
communication, natural*, behavior*, intervention, treatment, (Kratochwill et al. 2013) and the Single-Case Analysis
teaching, therapy, and training. The initial search yielded 726 and Review Framework (SCARF; Ledford et al. 2015).
articles. A secondary search for articles published between The WWC evidence standards were developed for use with
2016 and 2018 using the combination of keywords “autis*,” SCD and group design studies and are well-suited to assess
“intervention,” and “joint attention” in PsycINFO and ERIC internal validity; however, they do not adequately address
yielded 39 and seven peer-reviewed articles respectively; a procedural fidelity and other factors related to external va-
search of PsycINFO and ERIC using the keywords “autis*,” lidity (Wendt and Miller 2012; Wolery 2013). As such,
“intervention,” and “communication” yielded 428 and 92 SCARF, which incorporates information about external va-
peer-reviewed articles published between 2016 and 2018. lidity, was used in conjunction with the WWC evidence
Abstracts and method sections of all articles obtained were standards to evaluate SCD studies. SCARF criteria were
screened according to the following inclusion criteria: (1) in- also adapted to assess external validity features of group
dication the article investigated an intervention aligned with design studies. Whereas, several criteria are consistent
Kaiser and Trent’s (2007) description of naturalistic interven- across group and SCD studies, due to differences between
tions (i.e., intervention occurs in typical activities and rou- design types, several components within each feature dif-
tines, is based around child-interests and child initiations, is fered slightly for SCD and group design studies.
distributed throughout the day, and may involve behavioral Regardless of design type, a 5-point numeric scale was
[e.g., prompting, time delay, shaping] and social interaction used to rate each criterion (i.e., 4 = exemplary, 3 = accept-
strategies [e.g., modeling, expansions, recasts]); (2) at least able, 2 = minimal, and 1 = unacceptable). A “zero” rating
one dependent variable (DV) included an observational mea- was applied if a feature was not described in a study or if a
sure of a prelinguistic social communication skill with the study reported no effects of intervention on the target
pragmatic function of IJA, IBR, initiating social interaction behavior.
(ISI), including proto-imperative, proto-declarative, gestures, Rigor of group and SCD studies was evaluated according
gaze shifts, vocalization; and (3) participants were children to three elements that reflect internal validity (i.e., sufficiency
under the age of 8 (i.e., considered within the early childhood of data, reliability of dependent variable measurement, and
period of development) diagnosed with or at-risk for ASD and procedural fidelity). Information about study design, as well
significant communication delays (i.e., no consistent use of as timing and frequency of data collection, was used to deter-
spoken language or prelinguistic behaviors for communica- mine sufficiency of data for SCD studies. For group design
tive purposes). Studies of interventions for children with com- studies, criteria for sufficiency of data included design type,
munication delays but not ASD were excluded, as were stud- sample size, and establishment of group equivalence.
ies investigating structured behavioral interventions not con- Reliability of dependent variable measurement was informed
taining naturalistic components. A hand search of the refer- by the type, quality, and quantity of inter-rater reliability re-
ence sections of the studies that met inclusion criteria and prior ported for observational measures as well as psychometric
intervention reviews (e.g., Murza et al. 2016) was also properties of standardized measures. Procedural fidelity was
conducted. assessed in terms of the quality of the intervention description
and procedural fidelity measurement.
Descriptive Information Quality and breadth of measurement (i.e., external validity)
was evaluated similarly for SCD and group design studies
Articles were separated by design type (i.e., SCD or group across seven criteria. Criteria included quality of outcome var-
design). Data were extracted on intervention type (e.g., indi- iable measurement (e.g., operational definitions for outcome
vidual or group), implementation setting, intervention agent, measures; use of multiple sources and measures) as well as
total intervention length, key components, and targeted behav- measurement of response generalization (e.g., measurement
iors. Coders listed and defined all observable measures of of behaviors not directly targeted by the intervention), stimu-
child behavior per study (e.g., initiating joint attention, lus generalization (measurement of target behaviors in
 Rev J Autism Dev Disord

different contexts), and maintenance (e.g., length of time be- Results

tween intervention and follow-up; quantity of maintenance
data). Information about participant descriptions (e.g., pre- Research Designs
intervention skills; age; diagnosis), condition or comparison
group descriptions (e.g., operational definitions of baseline Individual studies were determined based on the definition pro-
condition or comparison group treatments; operational defini- vided in the SCARF framework to allow for evaluation.
tion of intervention; implementation settings), and social va- Specifically, a study was defined as “any stand-alone single
lidity (e.g., clinical significance of effects) was also used to case design with a single dependent variable,” (Ledford et al.
inform measurement quality and external validity ratings. 2015). Based on this definition, 26 of the studies obtained dur-
More detail regarding variables factored into overall ratings ing the initial search met inclusion criteria. One study from the
for each study may contact the first author or view the SCARF follow-up search met inclusion criteria (Chang et al. 2016).
protocol available online. Two of the 26 studies from the initial search (Kaale et al.
Group design and SCD studies with adequate methodolog- 2014; Kasari et al. 2008) were follow-up to other included
ical rigor and measurement characteristics (i.e., average scores studies and thus not included in the full review. Additional
for methodological rigor and quality and breadth of measure- information provided by these follow-up studies (e.g., mainte-
ment greater than or equal to 2) were also evaluated for results nance of treatment gains) is summarized in the appropriate
of the intervention. Criteria regarding treatment effects for section below. Of the 25 remaining studies, 11 studies within
primary and ancillary variables as well as evidence of main- eight articles used SCD methodology and 14 studies were con-
tenance and generalization of gains were used with both de- ducted using a group design. (Harjusola-Webb and Robbins
sign types. Study effect scores ranged from zero (i.e., no evi- (2012) included three studies within a single article; Dykstra
dence of intervention effects) to four (i.e., consistent, clear et al. (2012) included 2 studies within a single article.) Nine of
evidence of moderate to large intervention effects). For group the SCD studies used a multiple baseline across participants
design studies, evaluation criteria for results also included the design while one used a multiple probe across behaviors de-
appropriate use of statistics and reporting of effect sizes. The sign. The other SCD study used an alternating treatments de-
WWC framework was used to determine evidence of inter- sign. Thirteen of the group design studies were randomized
vention effects for SCD studies. Specifically, visual analysis control trials (RCT). The other group design study used a pre-
of data presented in each study was used to assess the level, post design and was classified as quasi-experimental.
trend, and variability of data within and across phases as well
as the immediacy of effect, overlap, and consistency of data Participant Characteristics
across similar phases to determine if a functional relation ex-
ists between the intervention and participant behavior change Across all studies, 387 children between 12 months and 8 years
(Kratochwill et al. 2013). Additional data about intervention of age received intervention (see Table 1). The quality and quan-
effects were extracted from those studies demonstrating ade- tity of information provided about inclusion criteria and partic-
quate methodological rigor, with an emphasis on studies that ipant characteristics varied widely. All 25 studies reported inclu-
also received a score of at least 2 for procedural fidelity. sion criteria or pre-intervention participant characteristics related
Detailed descriptions of effects for primary outcomes, second- to specific behavioral deficits targeted by the intervention.
ary outcomes or collateral gains, and maintenance and gener-
alization are presented for these studies. General Intervention Characteristics

Manualized interventions were used in 12 of the reviewed stud-

Interobserver Agreement ies. Interventions evaluated included Advancing Social-
Communication and Play (ASAP); Early Social Interaction
An independent reviewer rated 20 % of the studies randomly Project (ESI); interpersonal synchrony (IS); Joint Attention
selected from each design type to provide an estimate of inter- Symbolic Play Engagement and Regulation (JASPER) treat-
rater reliability. For each study, an agreement was scored when ment, JASPER with modifications to be used with small groups,
both evaluators provided the same numeric rating for a crite- JASPER precursors involving joint attention instruction;
rion or similar descriptions for descriptive information (e.g., prelinguistic milieu teaching (PMT); interventions combining
circled the same keywords for intervention type). The number pivotal response training (PRT) with other strategies; reciprocal
of agreements was divided by the sum of total agreements plus imitation training (RIT); and interventions without formal names
disagreements and multiplied by 100 to determine percentage (see Table 2). Although theoretically similar, interventions dif-
agreement. Reliability was above 80 % for all extracted infor- fered slightly in procedures. All interventions incorporated sev-
mation (study quality ratings = 80 %, study effects ratings = eral components aligned with descriptions of naturalistic inter-
84 %, descriptive information = 100 %). ventions (Kaiser and Trent 2007; Schreibman et al. 2015).
Rev J Autism Dev Disord

Table 1 Participant descriptions

Article N (age) Gender (%, M) Race/ethnicity Diagnosis

Chang et al. (2016) 66 (3–5 years) 89 % 21 % H, 16 % AS, 13 % AA, 31 % W, 19 % O ASD

Dykstra et al. 2012 3 (44–58 mos) 33 % 33.3 % H, AS, W Autism
Franco et al. 2013 6 (5–8 years) 83 % 40% H; 20 % AS; 20% PI; 20 % W Autism
Goods et al. 2013 7 (3–5 years) NR < 50 % W Autism
Harjusola-Webb and Robbins 2012 3 (37–44 mos) 100 % NR Autism
Ingersoll and Schreibman 2006 5 (29–45 mos) 60 % NR Autism
Ingersoll et al. 2007 5 (34–49 mos) 100 % NR Autism
Ingersoll 2012 14 (27–47 mos) 93 % 64 % W Autism
Jones et al. 2006 5 (2–3 years) 100 % NR Autism or PDD-NOS
Kaale et al. 2012 34 (2–5 years) 77 % NR Autism
Kasari et al. 2006 20 (3–4 years) 75 % 81 % W Autism
Kasari et al. 2010 19 (21–30 mos) 79 % 53 % W Autism
Kasari et al. 2014 52 (2–5 years) 83 % 31 % W ASD
Kasari et al. 2015 43 (2 left early; 22–36 mos) 81 % 7 % H; 63 % W; 9 % AS; 21 % O ASD
Kryzack and Jones 2015 3 (2–8 years) 100 % NR Autism
Landa et al. 2011 24 (21–33 mos) 83 % 79 % W Autism
Wetherby and Woods 2006 9 (3–5 years) NR 44 % W Autism
Vismara and Lyons 2007 3 (26–38 mos) 100 % 33 % AS; 67 % W Autism
Warreyn and Roeyers 2014 18 (3–7 years) 78 % NR Autism or PDD-NOS
Wetherby and Woods 2006 17 (12–24 mos) 88 % 24 % H; 18 % AA; 65 % W Autism or PDD-NOS
Wong 2013 14 (3–6 years) 86 % 50 % H; 43 % AA; 7 % W Autism
Yoder and Stone 2006 17 (1.9–3.5 years) 86 % 22 % AA; Autism or PDD-NOS
69 % W;
8%O

M, male; mos, months; NR, not reported; AS, Asian American; AA, African American; H, Hispanic American; MA, multiracial; PI, Pacific Islander; W,
White; O, other

Nineteen studies evaluated child-directed interventions implemented by individuals outside of the research team
(i.e., ASAP, IS, JASPER or JASPER variations, interven- provided information about implementation training. All
tions combining PRT with other elements, and PMT), parent-implemented interventions featured supervision
making following the child’s lead the most frequently in- and coaching throughout the intervention to support pro-
cluded component in studies reviewed (see Table 2). cedural fidelity. All but two teacher-implemented studies
Additionally, a majority of studies incorporated systemat- (Jones et al. 2006; Kryzack and Jones 2015) featured on-
ic use of prompts, natural consequences, instruction em- going intervention specific training. Kryzack and Jones
bedded in routines, and environmental arrangement. Less (2015) trained teachers to 100 % fidelity prior to
than 50 % of studies implemented time delay or linguistic intervention and Jones et al. (2006) included teachers
mapping as part of their described intervention procedures who received ongoing training in applied behavior analy-
(see Table 2). Additional intervention components fea- sis separately from the study. Indicators of intensity varied
tured in fewer than five studies included play narration, widely across studies and interventions evaluated.
language and play expansions, balanced turns, imitating
child behaviors, interspersing mastered and new tasks, Target Behaviors and Outcomes Measured
and incorporation of short discrete trial training sessions
prior to naturalistic instruction. Twenty-two studies evaluated interventions directly
All of the studies investigated interventions delivered targeting prelinguistic communication behaviors and three
in a one to one format. Three of the studies focusing on studies measured prelinguistic communication behaviors
teacher-implemented interventions also included group as collateral gains. Several of the studies also targeted
components, including strategies implemented weekly play skills (Chang et al. 2016; Dykstra et al. 2012;
during group instruction, and providing teachers with the Goods et al. 2013; Kasari et al. 2014; Kasari et al.
option of using intervention strategies in small group or 2015; Wong 2013). All studies included assessed out-
classroom format. (See Table 2 for details regarding im- comes related to prelinguistic communication behaviors
plementation setting, delivery agent, and duration by (see Table 3); however, specific behaviors and how these
study.) The majority of studies investigating interventions behaviors were defined and measured varied across
Table 2 Intervention and study descriptions

Intervention Article Delivery agent Setting Duration Key intervention components

CD SP NC ER EA TD LM

ASAP* Dykstra et al. 2012* Teacher, TA, SLP School 5–7 weeks X X X – – – –
ESI Wetherby 2006 Parent Home, clinic 1 year – X X X X X –
IS Landa et al. 2011 Ix Clinic-based class 6 months X X X X X – X
JA intervention Kasari et al. 2006* Ed psych GS Clinic-based class 5–6 weeks X X – X X – –
JA intervention Kasari et al. 2010* Parent Clinic 8 weeks X – – – X – –
(replication/modification Kaale et al. 2012* Teacher School 8 weeks X X X X – – –
of Kasari et al. 2006)
Wong 2013* Teacher School 8 weeks X X – X X – –
JASPER Lawton and Kasari 2012* Teacher; para School 6 weeks X X – X X X X
Goods et al. 2013* Ed psych GS School 12 weeks X – – X – X –
Kasari et al. 2014* Parent Home 12 weeks – X – – X – –
Kasari et al. 2015* Parent Clinic 10 weeks X X – X X – –
JASPER with small Chang et al. 2016 Teacher School 8 weeks X – – X X – –
group adaptation
PMT Yoder and Stone 2006 Ix, para Clinic 6 months – X X X – – –
Franco et al. 2013* SLP/BCBA Home 14 sessions X X X X X – X
PRT + other elements Jones et al. 2006 Teacher, TA Clinic-based class 26–157 total IJA sessions – X X – – X –
Vismara and Lyons, 2007 Parent Home/clinic 12 weeks X – X – – – –
Harjusola-Webb and Robbins, 2012 Teacher, SLP Clinic-based class 3–9 weeks X – – – X X –
Warreyn and Roeyers, 2014 Child’s usual therapist Clinic 4.5–5 months/24 sessions X X X X X X –
Prompts w/in CIs Kryzack and Jones, 2015 Parent and teacher Home and school 21–67 10-min sessions total X X X – – X –
RIT Ingersoll and Schreibman, 2006 Under grad students Clinic 10 weeks X X – – – – X
Ingersoll et al. 2007 Under grad students Clinic 10 weeks X X – – – – X
Ingersoll 2012 Grad and under grads Clinic 10 weeks – X – – – – –

*, manualized; ASAP, Advancing Social-Communication and Play; ESI, Early Social Interaction Project; IS, interpersonal synchrony; JA, joint attention; JASPER, Joint Attention and Symbolic Play
Engagement and Regulation treatment; PMT, prelinguistic milieu teaching; PRT, pivotal response training; CI, circumscribed interests; RIT, reciprocal imitation training; CD, child-directed; SP, systematic
use of prompts; NC, natural consequence; ER, instruction embedded in routines; EA, environmental arrangement; TD, time delay; LM, linguistic mapping; cont., continuous; TA, teaching assistant; SLP,
speech language pathologist; SC, self-contained; sess, sessions; ed., education; Ix, interventionist; NR, not reported; psych, psychology; GS, graduate student; Exp., experience; EIP, early intervention
program; para, paraprofessional; inc, inclusion; RT, responsive teaching; PF, performance feedback; CI, circumscribed interest
Rev J Autism Dev Disord
Rev J Autism Dev Disord

Table 3 Descriptions of target prelinguistic social communicative behavior by article

Author Discrete behaviors Aggregate IC Communicative function

Gestures Vocalize Gaze shift IJA IBR Other Undefined

Chang et al. 2016 X* X*

C
Dykstra et al. 2012 X XC XC XC
Franco et al. 2013 X X
Goods et al. 2013 X*C X*C X*C
Harjusola-Webb and Robbins 2012 XC XC XC
Ingersoll and Schreibman 2006 X* X*
Ingersoll et al. 2007 X X X
Ingersoll 2012 X* X*
Jones et al. 2006 X X
Kaale et al. 2012 XC* XC*
Kasari et al. 2006 XP* XP*
Kasari et al. 2010 XP XP
Kasari et al. 2014 X* X*
Kasari et al. 2015 X*
Kryzak & Jones 2015 X X
Landa et al. 2011 X* X*
Lawton & Kasari 2012 XC* XC* XC* XC*
Vismara and Lyons 2007 X X
Warreyn and Roeyers 2014 X X X
Wetherby and Woods 2006 X* X* X* X* X* X* X*
Wong 2013 XC* XC* XC*
Yoder and Stone 2006 X* X* X*

*Measured using communication and symbolic behavior scales, early social communication scales, or structured laboratory observation
C
Measured in classroom and/or during teacher-child interaction
P
Measured during caregiver-child interaction
No label, measured during sessions and/or other unstructured observation. Comm, communication; Bx, behaviors; IC, intentional communication; IJA,
initiating joint attention; IBR, initiating behavioral requests

studies. Prelinguistic social communication outcomes SCD and group studies (Tables 4 and 5). Strengths (i.e., elements
assessed included IJA (n = 19), IBR (n = 6), ISI (n = 2), with ratings > 2) in SCD studies included reliability, participant
IC acts (n = 1), and/or individual component behaviors descriptions, DV descriptions, condition descriptions, and mea-
such as gestures, gaze shifts, and vocalizations (n = 7). surement of response generalization. Weaker elements of SCD
Semi-structured observations (i.e., the Early Social studies included fidelity, sufficiency of data, social validity, stim-
Communication Scale [ESCS, Mundy et al. 2003] or ulus generalization, and measurement of maintenance.
Communication and Symbolic Behavior Scale [CSBS, Areas of strength for group design studies included reliability,
Wetherby and Prizant 1993]) were used to measure IJA, sufficiency of data, participant descriptions, condition descrip-
IBR, or ISI in several of the studies (e.g., Kasari et al. tions, DV descriptions, stimulus generalization, and response
2006; Landa et al. 2011). Studies that did not use the generalization. Weaker elements of group design studies includ-
ESCS or CSBS developed their own definitions of ed fidelity, social validity, and measurement of maintenance.
prelinguistic communicative behaviors.

Sufficiency of Data
Study Quality
Sufficiency of data ratings for group design studies included
Study quality ratings in the areas of rigor, treatment effects, and establishment of group equivalence and use of appropriate
quality and breadth of measurement are presented separately for statistics and sample size. All but one group design study
Table 4 Single-case design study quality

Study Design Rigor Tx effects Quality and breadth of measurement

Reliability Fidelity Sufficiency Main Generalization Maintenance Social Participant DV Condition Stimulus Response Maintenance
of data effects validity description description description generalization generalization

Harjusola-Webb and MB-P 3 3 2 0 0 0 1 4 3 3 0 0 0

Robbins 2012
(study 1: gesture)
Harjusola-Webb and MB-P 3 3 2 0 0 0 1 4 3 3 0 0 0
Robbins 2012
(study 2: voc)
Harjusola-Webb and Robbins MB-P 3 3 2 3 3 0 1 4 3 3 0 4 0
2012 (study 3: combo)
Vismara and Lyons 2007 Alt-Tx 4 1 3 1 2 0 1 4 4 1 4 4 0
Ingersoll et al. 2007 MB-P 3 1 3 na 2 2 1 4 4 4 3 2 4
Ingersoll and Schreibman 2006 MB-P 3 1 2 na 2 2 1 4 3 4 3 4 4
Jones et al. 2006 MP-B 3 1 1 3 4 4 2 4 4 3 2 3 4
Franco et al. 2013 MB-P 3 1 1 4 4 4 2 4 3 4 0 4 4
Kryzack and Jones 2015 MP-P 3 1 2 1 2 2 1 2 4 4 4 0 2
Dykstra et al. 2012 MB-P 1 0 1 0 1 0 1 4 2 3 3 4 0
(study 1: group)
Dykstra et al. 2012 MB-P 1 0 1 1 2 0 1 4 2 3 0 4 0
(study 2: full)
Average criterion rating 2.62 1.38 1.77 1.44 2.00 1.27 1.23 3.69 3.15 3.08 1.85 2.46 1.92

DV, dependent variable; MB-P, multiple baseline across participants; Alt-Tx, alternating treatments; MP-B, multiple probe across behaviors; MP-P, multiple probe across participants
Rev J Autism Dev Disord
 Rev J Autism Dev Disord

Table 5 Group design study quality

Study Design Rigor Tx effects Quality and breadth of measurement

Reliability Fidelity Sufficiency Main Generalization Maintenance Social Participant DV Comparison Stimulus Response Maintenance
of data effects validity description description description generalization generalization

Chang et al. 2016 RCT 4 4 4 0 0 0 4 4 3 3 4 3 3

Goods et al. 2013 RCT 3 2 3 3/0* 2 0 1 4 4 4 4 3 0
Ingersoll 2012 RCT 3 1 4 na 4 4 0 4 3 4 na 2 3
Kaale et al. 2012 RCT 3 4 4 3 2 0** 1 4 4 4 4 2 0
Kasari et al. 2006 RCT 4 1 4 3 4 0** 0 4 4 3 4 3 0
Kasari et al. 2010 RCT 4 1 4 0 4 0 2 4 4 4 0 3 3
Kasari et al. 2014 RCT 4 0 4 2 4 4 2 4 4 4 4 3 3
Kasari et al. 2015 RCT 4 0 4 0 0 0 1 4 4 4 2 4 3
Landa et al. 2011 RCT 4 4 4 1 3 0 0 4 4 3 4 4 3
Lawton and Kasari 2012 RCT 4 1 4 4 3 0 1 4 4 4 4 3 0
Warreyn and Roeyers RCT 4 0 3 0 1 0 1 4 3 2 4 2 0
2014
Wetherby and Woods Quasi 4 0 1 4 4 0 1 4 4 2 4 4 0
2006
Wong 2013 RCT 4 0 4 2 4 0 2 4 4 3 4 3 0
Yoder and Stone 2006 RCT 4 1 4 3 2 0 0 4 4 4 4 2 0
Average criterion rating 3.79 1.36 3.64 1.79 2.64 .57 1.14 4.00 3.79 3.43 3.45 2.93 1.29

*Study received a rating of 3 for IBR and 0 for IJA

**Information from follow-up studies increases maintenance score to 4
Tx, treatment; DV, dependent variable; RCT, randomized control trial; Quasi, quasi-experimental design
 Rev J Autism Dev Disord

received ratings of three or above for sufficiency of data, as Kaale et al. 2012; Landa et al. 2011), with the latter two studies
randomized controlled trials (RCT) comprised 13 of the 14 also reporting information about reliability of fidelity data and
group design studies; these studies also included an adequate use of evaluators blind to study condition or purpose.
number of participants for the statistics utilized. Sufficiency of
data was more variable across SCD studies, which were eval- Descriptive Information
uated based on the number of data points per condition, the
timing of data collection for multiple baseline studies (i.e., All SCD and group design studies provided adequate infor-
concurrent or non-concurrent data collection), and whether mation about the children receiving the intervention (e.g.,
additional information was necessary to demonstrate a func- standardized testing results, inclusion/exclusion criteria, infor-
tional relation. All 11 studies reported at least three data points mation related to the target behaviors) and the outcomes being
per primary comparison condition. Of the 10 multiple baseline measured (e.g., operationally defined behaviors, use of exam-
or multiple probe designs, six began data collection simulta- ples/non-examples, use of psychometrically sound instru-
neously for all participants. All but two studies (Ingersoll et al. ments). The majority of studies included sufficient condition
2007; Vismara and Lyons 2007) required more data points to or comparison group descriptions (e.g., well-described inter-
support detection of a functional relation. vention procedures, information about dosage, information
about treatments received by comparison group, baseline de-
Reliability and Procedural Fidelity scription). Only one study received unsatisfactory ratings for
condition descriptions due to lack of information about base-
Reliability of DV measurement was assessed in all studies, with line conditions, dosage, setting, and/or interventionist
all group design studies and 82 % of SCD studies receiving characteristics.
satisfactory reliability ratings (i.e., point-by-point calculation of
reliability for at least 20 % of observations, reliability of at least Social Validity
80 % or kappa > 0.6, and/or the use of standardized measurement
tools with adequate psychometric properties). Reliability was Social validity was measured via blind raters of acceptability/
assessed for at least 20 % of observations in the majority of group feasibility of intervention implementation, acceptability of
(n = 13) and SCD (n = 10) studies. In all but one group design DVs, and importance of study results and/or normative com-
study, adequate reliability was reported via statistics (e.g., intra- parisons in five SCD studies but no group design studies.
class correlation) or the use of data collection instruments with Several studies (SCD n = 9; group n = 10) were judged to have
satisfactory psychometric properties. Blind coders and/or evalu- some degree of social validity through use of either indige-
ators were also used in 13 out of 14 group design studies. Seven nous implementers (e.g., parents, teachers) or environments
SCD studies reported adequate reliability values and two SCD (e.g., homes, schools); however, implementers were asked
studies used coders blind to condition to collect reliability data. their opinion about feasibility and acceptability of the inter-
Evaluation of procedural fidelity was not conducted at the ventions in only one SCD study and three group design
same level of rigor as reliability measurement across the major- studies. Psychometrically validated measures of feasibility
ity of included studies. Whereas, all SCD studies included some or acceptability were not included in any of the studies
type of information regarding procedural fidelity, only three reviewed.
studies within the same article received satisfactory ratings.
Harjusola-Webb and colleagues (2012) evaluated procedural Generalization and Maintenance
fidelity via direct observation of at least 20 % of sessions in
both primary conditions. Reliability of the procedural fidelity Generalization was assessed in the majority of studies, in
data was also included. One other study reported fidelity data terms of both measurement of DVs outside of the intervention
from both primary conditions (Vismara and Lyons 2007) and context and measurement of behaviors not directly targeted
one study collected fidelity data for at least 20 % of sessions within intervention sessions. All group design studies
(Jones et al. 2006). No other SCD studies reported fidelity data assessed generalization of behaviors to other people, environ-
separately by condition or evaluated reliability of procedural ments, activities, and/or materials. Six SCD studies also eval-
fidelity data. Of the other SCD studies that included procedural uated stimulus generalization, either via probes throughout the
fidelity information, nine reported at least 80 % fidelity or other study (n = 4), or within the context of a SCD (n = 2). In addi-
evidence of differentiation among conditions. Many group de- tion to measurement in a different setting, pre- and post-
sign studies similarly lacked information about procedural fi- intervention assessments were often conducted by individuals
delity. Of the group design studies that included procedural different from those implementing the intervention using ac-
fidelity information, eight provided evidence supporting correct tivities and materials also not present during intervention ses-
intervention implementation. Three of the eight studies collect- sions. Regarding response generalization, eight SCD studies
ed fidelity data in at least 20 % of sessions (Goods et al. 2013; and all group design studies taught behavioral tendencies
Rev J Autism Dev Disord

(e.g., intentional communication, initiating joint attention) as (Ingersoll et al. 2007), with some evidence of maintenance
opposed to discrete behaviors (e.g., eye contact, vocalization). reported 1-month post-intervention.
Behaviors not directly targeted by the interventions (e.g., im- Over half of the 13 group design studies that met design
itation) were evaluated in several studies included in the pres- standards reported results suggesting minimal to strong effects
ent review or follow-ups to included studies (i.e., Kaale et al. of the intervention on target prelinguistic communication out-
2014; Kasari et al. 2008; Whalen et al. 2006). Evidence for comes, with effect sizes ranging from small to large (see
collateral effects is presented in the following section. Table 7). Interventions evaluated in these studies included
Five SCD studies and six group design studies collected JASPER (or an earlier version that only targeted JA), PMT,
information about maintenance of behavioral gains. All six and the IS curriculum. All but two of these studies (Goods
group design studies and four SCD studies evaluated gains et al. 2013; Kasari et al. 2015) reported results suggesting
maintained at least 1 month post-intervention. Follow-up stud- increased IJA following intervention. Evidence was provided
ies to two group design studies (Kaale et al. 2014; Kasari et al. for intervention effects on increased IBR (Yoder and Stone
2008) included long-term maintenance data not provided in 2006) and communicative gestures used for the purpose of
the initial study. Of the SCD studies measuring maintenance, requesting or joint attention (Goods et al. 2013; Kasari et al.
all but one measured maintenance more than once. No group 2006; Lawton and Kasari 2012). Evidence was inconsistent
design studies included repeated measures of maintenance. for effects outside of the intervention context. Several studies
reported increases in IJA following intervention during class-
Intervention Effects room observations and/or caregiver/teacher-child interactions
but not on a semi-structured assessment (i.e., ESCS) conduct-
Results presented in each study were rated on a zero to four ed by a novel adult (Chang et al. 2016; Kaale et al. 2012;
scale quantifying the study’s main, generalized, and main- Lawton and Kasari 2012; Wong 2013). In contrast, results in
tained effects, with scores of zero indicating no evidence of several studies suggested generalized improvements in show-
effects and four indicating strong evidence (see Tables 4 and ing (Lawton and Kasari 2012) and overall IJA (Kasari et al.
5). Four SCD studies did not meet rigor criteria because base- 2006; Kasari et al. 2014; Yoder and Stone 2006) on the ESCS.
line data were non-concurrent (Dykstra et al. 2012; Franco Maintenance of behavioral improvements was reported for
et al. 2013; Jones et al. 2006) and/or reliability of the DVs RIT (Ingersoll 2012) as well as JASPER or related interven-
was not established (Dykstra et al. 2012). Wetherby and tions (Kaale et al. 2014; Kasari et al. 2008; Kasari et al. 2014).
Prizant (2006) did not meet standards because group equiva-
lence was not established prior to intervention.
Intervention effects were inconsistent for the seven SCD Discussion
studies conducted with sufficient rigor to support detection of
a functional relation (see Table 6). Two studies that evaluated A systematic review of the literature was conducted to identify
modifications of PRT reported results suggesting clear, imme- naturalistic behavioral interventions designed to improve
diate improvements in child expressive communication fol- prelinguistic communication in children with ASD. All inter-
lowing teacher-implemented intervention (Harjusola-Webb ventions contained multiple naturalistic and behavioral com-
and Robbins 2012) and clear (albeit less consistently immedi- ponents, many of which were common across several inter-
ate) improvements in child IJA following parent-implemented ventions. Over half of the eight methodologically rigorous
intervention using perseverative interests (Vismara and Lyons group design studies and three SCD studies demonstrating
2007). IJA was also found to increase for two of three partic- functional relations presented evidence for generalization
ipants following implementation of another intervention not and maintenance of prelinguistic communicative behaviors.
explicitly labeled a modification of PRT that used similar Considering interventions targeted core deficits of ASD, evi-
strategies and circumscribed interests (Kryzack and Jones dence of behavior change is promising. Results of the current
2015). Additionally, results of Kryzack and Jones (2015) pro- review are consistent with previous reviews focused on com-
vided evidence for maintenance of effects for one participant munication more broadly (Goldstein 2002; NRC 2001;
as well as inconsistent evidence of generalized effects during Prelock et al. 2011) and JA skills specifically (White et al.
interactions with different people and materials and in differ- 2011), providing further support for recommendations for
ent locations. RIT was evaluated in two SCD studies that the use of naturalistic behavioral interventions to teach a va-
measured prelinguistic social communication behaviors in ad- riety of skills to children with ASD.
dition to main effects related to imitation. Despite not being Despite promising results, there were several methodolog-
directly targeted by the intervention, a functional relation was ical weaknesses across design types that warrant discussion.
demonstrated between RIT and increased coordinated joint Within SCD studies, issues related to reliability of DV mea-
attention (Ingersoll and Schreibman 2006) as well as sponta- surement, systematic introduction of interventions, and ade-
neous use of gestures and gestures paired with verbalizations quacy of data limited interpretation of intervention effects to
 Rev J Autism Dev Disord

Table 6 Intervention outcomes: single-case design studies with high rigor ratings

Tx Study Tx effects Participant outcomes/key findings

Main effects Generalization Maintenance

PRT variations Harjusola-Webb and 0 0 0 1/3 children clearly increased use of gestures upon
Robbins 2012-G introduction of intervention
Harjusola-Webb and 0 0 0 2/3 children demonstrated increased use of
Robbins 2012-V vocalizations upon introduction of intervention
Harjusola-Webb and 3 3 0 All children demonstrated increased expressive
Robbins 2012-C communication upon introduction of intervention
Vismara and Lyons 2007 1 2 0 All 3 children increased IJA upon tx introduction; gains
evident using PI and NP; gains delayed for 1 child
RIT Ingersoll and Schreibman na 2 2 4/5 children increased CJA upon introduction of RIT;
2006 effects generalized to SLO for 3/5 children, 2/5
children demonstrated generalization to caregiver
Ingersoll et al. 2007 na 2 2 All children increased gesture use; 3 children increased
use of gestures with verbalizations, effects delayed
for 2 children; all children generalized gains.
Other Kryzack and Jones 2015 1 2 2 2/3 children clearly increased IJA upon tx; some
evidence of generalization to novel people, settings,
and activities.

Treatment effects ratings > 2 indicate at least 3 clear demonstrations of effect

Tx, treatment; PRT, pivotal response training; G, gestures; V, vocalizations; C, combined expressive communication; IJA, initiating joint attention; PI,
perseverative interest; NP, nonperseverative interest; RIT, reciprocal imitation training; CJA, coordinated joint attention; SLO, structured laboratory
observation

only six studies, three of which demonstrated a functional challenges with conducting research in applied settings are
relation. Intervention effects on child outcomes were also often unavoidable). Additionally, systems in place to measure
demonstrated in eight of the 13 group design studies that used fidelity across all study conditions will strengthen conclusions
methodology that permitted the interpretation of intervention that can be drawn regarding effects of an intervention.
effects. Across the majority of SCD and group design studies,
information about procedural fidelity necessary to attribute Implications
participant outcomes to the intervention being studied was
insufficient or completely missing. When limiting interpreta- Results from several studies suggest that children with ASD
tion of effects to studies providing adequate fidelity data, only can be taught prelinguistic communication skills using natu-
one SCD study and three group design studies were found to ralistic behavioral strategies. Of the interventions examined in
demonstrate intervention effects. Additional areas of weak- the present review, JASPER or earlier iterations received the
ness identified in some studies include descriptions of inter- most research support. Evidence from eight RCT studies sug-
ventions lacking detail required for replication, limited de- gests that participation in JASPER and related JA interven-
scriptions of child characteristics relevant to behaviors tions may lead to increased use of gestures for the purposes of
targeted by the intervention, and limited measurement of gen- IJA and requesting, coordinated gaze shifts, overall IJA, and/
eralization and maintenance. Although not necessary for in- or time spent jointly engaged for children ages 21 months to
terpretation of effects, such information is important to under- 6 years with ASD. It should be noted that despite large effects
stand to whom, what behaviors, and in what environments for improved joint engagement in two of these studies, gains
effects may generalize. Finally, explicit measurement of social in IJA were minimal or nonexistent, which authors report may
validity was absent from most studies. Future research should be attributable to core deficits such as IJA requiring larger
directly address methodological limitations by improving up- dosage of intervention. RIT, PMT, and two variations of
on these study design elements, thereby strengthening the ev- PRT also appear to have similar positive effects on
idence supporting the effectiveness of naturalistic behavioral prelinguistic social communication targets. Evidence of im-
interventions in improving prelinguistic social communica- proved IJA and gesture use was also provided in studies of
tion behaviors. For example, attention to current design stan- RIT despite these prelinguisitic skills not being directly
dards for group- and single-case design studies prior to imple- targeted. Interventions, dependent variables, and measure-
mentation can increase the likelihood primary and reliability ment contexts varied across studies, making it difficult to de-
data will be adequate for making causal inferences (although termine which specific strategies impacted which prelinguistic
Rev J Autism Dev Disord

Table 7 Intervention outcomes: group design studies with high rigor ratings

Tx Study Comparison groups Effects ratings Participant prelinguistic social

communication outcomes
Main effects Generalization Maintenance

JASPER Goods et al. 2013 TAU (30 h ABA/week) 3/0* 2 0 JASPER > control post-tx classroom
IBR (d = 1.51); NS differences in
classroom IJA and ESCS IJA/IBR
Chang et al. 2016 VB-MAPP 0 0 0 NS group differences in ESCS IJA/IBR
Kasari et al. 2014 Group CEM 2 4 4 JASPER > CEM IJA (f = 0.14)
Kasari et al. 2015 PEI 0 0 0 NS group differences in IJA
Lawton and Kasari 2012 Typical education 4 3 0 JASPER > control post-tx classroom
IJA (d = 1.85); NS differences on
ESCS
JA Intervention Kaale et al. 2012 Typical preschool 3 2 0** Tx > control post-tx IJA during
teacher-child play (d = 0.44); NS
differences for ESCS IJA
Kasari et al. 2006 Symbolic play tx; control 3 4 0** Tx > control post-tx IJA (EF = 1.50)
Kasari et al. 2010 Waitlist control 0 4 0 NS group differences in IJA
Wong 2013 Symbolic play tx; control 2 4 0 Tx > control IJA growth rate
(β = − 0.09, SE = 0.04, p = .03)
IS Landa et al. 2011 EIP 1 3 0 IS group significantly increased IJA
pre- to post-tx (d = 1.59); NS group
differences in post-tx IJA
PMT Yoder and Stone 2006 PECS 3 2 0 Children in RPMT group with > 7
pre-tx IJA exhibited greater IJA
gains post-tx. (ΔR2 = .36, 95 %
CI = .12–.94).
RIT Ingersoll 2012 TAU in the community na 4 4 RIT > control post-tx ESCS IJA
(η2 = 0.16)
PRT Warreyn and Roeyers 2014 TAU 0 1 0 NS group differences post-tx IJA

A rating of “3” for “Main Effects” indicates small to moderate effects for all primary outcome variables; a rating of “4” for “Main Effects” indicates
moderate to large effects for all primary outcome variables
Tx, treatment; JASPER, Joint Attention Symbolic Play Engagement and Regulation intervention; TAU, treatment as usual; ABA, applied behavior
analysis; IBR, initiating behavior requests; NS, non-significant; IJA, initiating joint attention; ESCS, Early Social Communication Scales; VB-MAPP,
Verbal Behavior Milestones Assessment and Placement Program (Sundberg 2008); CEM, caregiver education module; PEI, parent-only
psychoeducational intervention; JA, joint attention; IS, interpersonal synchrony; EIP, early intervention program; PMT, prelinguistic milieu teaching;
PECS, picture exchange communication system; RPMT, responsivity education prelinguistic milieu teaching; RIT, reciprocal imitation training; PRT,
pivotal response training
*
Study received a rating of 3 for IBR and 0 for IJA
**
Information from follow-up studies increases maintenance score to 4

social communication behaviors. Future reviews should con- support was child-directed and featured systematic prompting
sider methodologies such as meta-analysis across a single in- or time delay, instruction embedded in routines, environmen-
tervention (or similarly described independent variables) to tal arrangement, and naturally occurring consequences.
quantify efficacy through a summary statistic (e.g., Murza Additional similarities across interventions with positive out-
et al. 2016). Such a review was beyond the scope of this paper. comes were related to dosage, with implementation occurring
Overall, these studies demonstrated that parents, teachers, at relatively high dosages in many of the studies that demon-
para-professionals, graduate, and undergraduate students can strated effects. It is to be expected that interventions must be
be trained to implement naturalistic behavioral interventions conducted at high dosages to lead to generalized behavior
in clinics, schools, and the home. Various factors were shared change given that persistent deficits in social communication
across many interventions with evidence of treatment effects behaviors are hallmarks of ASD. As the resources required for
from at least one methodologically rigorous study, including such intensive interventions are barriers for many families,
the emphasis on training and weekly supervision of individ- evidence supporting the use of indigenous implementers such
uals implementing the intervention, use of a treatment manual, as caregivers and teachers provided by many of the studies
and incorporation of several key naturalistic behavioral com- reviewed is promising and should allow for high dosage im-
ponents. The majority of the interventions with research plementation that is feasible and cost-effective.
 Rev J Autism Dev Disord

There were no consistent differences in child character- more efficient training. For example, the strategies re-
istics between rigorous studies that did and did not dem- quired to teach gaze shifting to a child with minimal use
onstrate an intervention effect. Studies that reported pos- of eye contact or other prelinguistic communication skills
itive outcomes generally included children who were at may be different from those required to teach IJA to a
least 3 years old, whereas the average participant age student who exhibits prelinguistic communication skills
was younger than 3 years in the majority of studies with but only uses them to request. Also, studies should in-
minimal or no effects. Several studies reporting positive clude measures of social validity to gather information
outcomes included participants who engaged in at least about the acceptability and feasibility of training and im-
five instances of pre-intervention IJA; however, other plementation in the classroom. Finally, to promote under-
studies that reported positive outcomes included partici- standing of mechanisms for change in prelinguistic com-
pants with similarly low frequencies of IJA to those stud- munication skills, examination of the strategies consistent
ies that did not demonstrate intervention effects. with applied behavior analysis within the context of nat-
Interestingly, Yoder and Stone (2006) reported that partic- uralistic behavioral interventions should be considered in
ipants who received RPMT demonstrated greater im- future research.
provements in IBR and IJA when compared with partici-
pants who received PECS training, but only if they en-
gaged in relatively high rates of IJA prior to intervention. Limitations
Additional research similar to that conducted by Yoder
and Stone, in which participant outcomes are analyzed Limitations of the current review are worth considering
in the context of pre-intervention characteristics, is re- when interpreting the findings. First, inclusion criteria
quired to clarify participant characteristics that may be were relatively broad, resulting in studies varying in
predictive of response to a particular intervention. the degree of behavioral and naturalistic components
Considering these findings in the context of teacher included in the interventions. Additionally, IOA data
training, the existing evidence supports training teachers were not collected on initial decisions to include or
in JASPER or PRT intervention packages (Goods et al. exclude a study. Also, when coding quality of studies
2013; Harjusola-Webb and Robbins 2012; Kaale et al. using SCARF, the same rating was occasionally given
2012; Lawton and Kasari 2012; Wong 2013). Although when an element was completely left out of a study and
effects were also demonstrated for PMT and RIT when when an element was included but not addressed ade-
implemented in clinics or home settings (e.g., Ingersoll quately. These similar ratings provided accurate infor-
2012; Yoder and Stone 2006), further research is required mation in terms of evaluating study methodology.
to evaluate their implementation in schools. Given that However, in making recommendations for future re-
teachers often lack time provided for training (Closs and search, some studies receiving the same scores for dif-
Lewin 1998), large intervention packages may not always ferent elements may require different recommendations.
be feasible. Although it is not possible to confidently de- Next, comparing effects across study design types was
termine which components were “active ingredients” in not possible; effect sizes were not calculated or com-
the studies reviewed, it may be helpful to train teachers pared across studies. A direct comparison of quality
to use the individual naturalistic behavioral strategies that cannot truly be made between group and SCD studies,
were most frequently utilized in effective interventions as criteria differed slightly across design types to ac-
(e.g., following the child’s lead, systematic use of count for different ways in which studies were conduct-
prompting and time delay) as an alternative when time ed. These criteria may have differed in stringency and
and resources do not permit utilization of larger packages. resulted in inflated scores for group design studies.
Future research should include investigations of PMT Another limitation is that the small number of studies
and RIT in the classroom as well as component analyses demonstrating sufficient methodological rigor precludes
to determine the naturalistic behavioral strategies or com- an ability to determine which treatment is optimal for
bination of strategies that are most likely to facilitate be- treating social communication deficits and for whom.
havior change. These studies should report specific infor- Further research is needed in this area. Finally, the cur-
mation about participant characteristics, with emphasis on rent review focused specifically on naturalistic behavior-
strengths and weaknesses related to the target behavior al interventions for very young children with ASD. This
(e.g., whether or not the student uses individual focus may have led to the exclusion of studies
prelinguistic behaviors to communicate). Additionally, re- employing adult-directed, highly structured approaches,
search should include measures of discrete as well as providing a narrow focus for our understanding of ef-
broad behaviors, as information about the specific strate- fective interventions targeting prelinguistic communica-
gies that lead to change in specific behaviors may lead to tion in young children with ASD.
Rev J Autism Dev Disord

Conclusion Dube, W. V., MacDonald, R. P., Mansfield, R. C., Holcomb, W. L., &
Ahearn, W. H. (2004). Toward a behavioral analysis of joint atten-
tion. The Behavior Analyst, 27, 197–207.
Despite methodological flaws apparent in several studies, re- *Dykstra, B., Watson, Crais, & Baranek (2012). The impact of the ad-
sults of the present review suggest that naturalistic behavioral vancing social-communication and play (ASAP) intervention on
interventions can be used to teach JA and associated preschoolers with autism spectrum disorder. Autism, 16, 27–44.
prelinguistic social communication skills to young children *Franco, J. H., Davis, B. L., & Davis, J. L. (2013). Increasing social
interaction using prelinguistic milieu teaching with nonverbal
with ASD. Furthermore, evidence exists for the generalization school-aged children with autism. American Journal of Speech-
and maintenance of these skills following intervention. Given Language Pathology, 22, 489–502.
the use of multi-component interventions and variation in def- Goldstein, H. (2002). Communication intervention for children with au-
initions and measurement of target outcomes, it is difficult to tism: a review of treatment efficacy. Journal of Autism and
Developmental Disorders, 32, 373–396.
determine what components produced effects for whom and
*Goods, K., Ishijima, E., Chang, Y., & Kasari, C. (2013). Preschool based
for what behaviors. Overall, it appears that interventions im- JASPER intervention in minimally verbal children with autism: pilot
plemented by individuals trained to criterion and exposed to RCT. Journal of Autism and Developmental Disorders, 43, 1050–
ongoing coaching or supervision featuring some combination 1056.
of naturalistic and behavioral components may lead to im- *Harjusola-Webb, S. M., & Robbins, S. H. (2012). The effects of teacher-
implemented naturalistic intervention on the communication of pre-
provements in prelinguistic social communication behaviors schoolers with autism. Topics in Early Childhood Special
for young children with ASD when implemented in relatively Education, 32(2), 99–110.
high dosages. Future research using rigorous methodology *Ingersoll, B. (2012). Brief report: effect of a focused imitation interven-
should focus on evaluating specific components that may tion on social functioning in children with autism. Journal of Autism
and Developmental Disorders, 42, 1768–1773.
serve as “active ingredients” in promoting behavior change *Ingersoll, B., & Schreibman, L. (2006). Teaching reciprocal imitation
as well as specific child characteristics that may lead to differ- skills to young children with autism using a naturalistic behavioral
ential responsivity to interventions. approach: effects on language, pretend play, and joint attention.
Journal of Autism and Developmental Disorders, 36, 487–505.
*Ingersoll, B., Lewis, E., & Kroman, E. (2007). Teaching the imitation
Compliance with Ethical Standards and spontaneous use of descriptive gestures in young children with
autism using a naturalistic behavioral intervention. Journal of
Conflict of Interest The authors declare that they have no conflict of Autism and Developmental Disorders, 37, 1446–1456.
interest. Jones, E. A., & Carr, E. G. (2004). Joint attention in children with autism
theory and intervention. Focus on autism and other developmental
disabilities, 19, 13–26.
*Jones, E. A., Carr, E. G., & Feeley, K. M. (2006). Multiple effects of
References joint attention intervention for children with autism. Behavior
Modification, 30, 782–834.
References marked with an asterisk were included in the full *Kaale, A., Smith, L., & Sponheim, E. (2012). A randomized controlled
trial of preschool-based joint attention intervention for children with
review autism. Journal of Child Psychology and Psychiatry, 53, 97–105.
Kaale, A., Fagerland, M. W., Martinsen, E. W., & Smith, L. (2014).
Adamson, L. B., Bakeman, R., & Deckner, D. F. (2004). The develop- Preschool-based social communication treatment for children with
ment of symbol-infused joint engagement. Child Development, 75, autism: 12-month follow-up of a randomized trial. Journal of the
1171–1187. American Academy of Child & Adolescent Psychiatry, 53, 188–198.
Bakeman, R., & Adamson, L. B. (1984). Coordinating attention to people Kaiser, A. P., & Delaney, E. M. (1998). Responsive conversation: creating
and objects in mother–infant and peer–infant interaction. Child opportunities for naturalistic language teaching. Young Exceptional
Development, 55, 1278–1279. Children Monograph Series, 3, 13–23.
Bruner, J. (1981). Intention in the structure of action and interaction. Kaiser, A. P., & Trent, J. A. (2007). Communication intervention for
Advances in Infancy Research, 1, 41–56. young children with disabilities: naturalistic approaches to promot-
*Chang, Y., Shire, S. Y., Shih, W., Gelfand, C., & Kasari, C. (2016). ing development. Handbook of developmental disabilities, 224–
Preschool deployment of evidence-based social-communication in- 246.
tervention: JASPER in the classroom. Journal of Autism and Kasari, C., Freeman, S., & Paparella, T. (2001). Early intervention in
Developmental Disorders, 46, 2211–2223. autism: joint attention and symbolic play. International Review of
Charman, T., Swettenham, J., Baron-Cohen, S., Cox, A., Baird, G., & Research in Mental Retardation, 23, 207–237.
Drew, A. (1997). Infants with autism: an investigation of empathy, *Kasari, C., Freeman, S., & Paparella, T. (2006). Joint attention and
pretend play, joint attention, and imitation. Developmental symbolic play in young children with autism: a randomized con-
Psychology, 33, 781. trolled intervention study. Journal of Child Psychology Psychiatry,
Chawarska, K., Paul, R., Klin, A., Hannigen, S., Dichtel, L. E., & 47, 611–620.
Volkmar, F. (2007). Parental recognition of developmental problems Kasari, C., Paparella, T., Freeman, S., & Jahromi, L. B. (2008). Language
in toddlers with autism spectrum disorders. Journal of Autism and outcome in autism: randomized comparison of joint attention and
Developmental Disorders, 37, 62–72. play interventions. Journal of Consulting and Clinical Psychology,
Closs, S., & Lewin, B. (1998). Perceived barriers to research utilization: a 76, 125.
survey of four therapies. International Journal of Therapy and *Kasari, C., Gulsrud, A. C., Wong, C., Kwon, S., & Locke, J. (2010).
Rehabilitation, 5, 151–155. Randomized controlled caregiver mediated joint engagement
 Rev J Autism Dev Disord

intervention for toddlers with autism. Journal of Autism and National Research Council. (2001). Educating children with autism.
Developmental Disorders, 40, 1045–1056. (Committee on educational interventions for children with autism,
*Kasari, C., Lawton, K., Shih, W., Barker, T. V., Landa, R., Lord, C., et al. division of behavioral and social sciences and education).
(2014). Caregiver-mediated intervention for low-resourced pre- Washington: National Academy Press.
schoolers with autism: an RCT. Pediatrics, 134, e72-e79. Prelock, P. A., Paul, R., & Allen, E. M. (2011). Evidence-based treat-
*Kasari, C., Gulsrud, A., Paparella, T., Hellemann, G., & Berry, K. ments in communication for children with autism spectrum disor-
(2015). Randomized comparative efficacy study of parent- ders. In B. Reichow, P. Doehring, D. Cicchetti, & F. Volkmar (Eds.),
mediated interventions for toddlers with autism. Journal of Evidence-based practices and treatments for children with autism
Counseling and Clinical Psychology, 83, 354–563. (pp. 93–169). NY: Guilford Press.
Koegel, R. L., & Koegel, L. K. E. (1995). Teaching children with autism: Schreibman, L., Dawson, G., Stahmer, A., Landa, R., Rogers, S., McGee,
strategies for initiating positive interactions and improving learning G. G., et al. (2015). Naturalistic developmental behavioral interven-
opportunities. Baltimore: Paul H Brookes Publishing. tions: empirically validated treatments for autism spectrum disorder.
Koegel, L. K., Koegel, R. L., Harrower, J. K., & Carter, C. M. (1999). Journal of Autism and Developmental Disorders, 45, 2411–2428.
Pivotal response intervention I: overview of approach. Research and Shumway, S., & Wetherby, A. M. (2009). Communicative acts of chil-
Practice for Persons with Severe Disabilities, 24, 174–185. dren with autism spectrum disorders in the second year of life.
Kratochwill, T., Hitchcock, J., Horner, R., Levin, J., Odom, S., Journal of Speech, Language, and Hearing Research, 52, 1139–
Rindskopf, D., & Shadish, W. (2013). Single-case intervention re- 1156.
search design standards. Remedial and Special Education, 34, 26– Stokes, T. F., & Osnes, P. G. (1989). An operant pursuit of generalization.
38. Behavior Therapy, 20, 337–355.
*Kryzack, L. A., & Jones, E. A. (2015). The effect of prompts within Sundberg, M. L. (2008). Verbal behavior milestones assessment and
embedded circumscribed interests to teach initiating joint attention placement program: the VB-MAPP. Concord: AVB Press.
in children with autism spectrum disorders. Journal of Toth, K., Munson, J., Meltzoff, A. N., & Dawson, G. (2006). Early pre-
Developmental and Physical Disabilities, 27, 265–284. dictors of communication development in young children with au-
*Landa, R. J., Holman, K. C., O’Neill, A. H., & Stuart, E. A. (2011). tism spectrum disorder: joint attention, imitation, and toy play.
Intervention targeting development of socially synchronous engage- Journal of Autism and Developmental Disorders, 36, 993–1005.
ment in toddlers with autism spectrum disorder: a randomized con- *Vismara, L. A., & Lyons, G. L. (2007). Using perseverative interests to
trolled trial. Journal of Child Psychology and Psychiatry, 52, 13–21. elicit joint attention behaviors in young children with autism: theo-
retical and clinical implications for understanding motivation.
Lane, J. D., & Brown, J. A. (2016). Promoting communication develop-
Journal of Positive Behavior Interventions, 9, 214–228.
ment in young children with or at-risk for disabilities. In B. Reichow,
Warren, S., Bredin-Oja, S., Fairchild, M., Finestack, L., Fey, M., & Brady,
B. A. Boyd, E. E. Barton, & S. L. Odom (Eds.), Handbook of early
N. (2006). Responsivity education/prelinguistic milieu teaching. In
childhood special education. New York: Springer.
R. J. McCauley & M. E. Fey (Eds.), Treatment of language disor-
*Lawton, K., & Kasari, C. (2012). Teacher-implemented joint attention
ders in children (pp. 47–75). Baltimore: Paul H. Brooks Publishing
intervention: pilot randomized controlled study for preschoolers
Co..
with autism. Journal of Consulting and Clinical Psychology, 80,
*Warreyn, P., & Roeyers, H. (2014). See what I see, do as I do: promoting
687–693.
joint attention and imitation in preschoolers with autism spectrum
Ledford, J. R., Lane, J. D., Ayres, K. A., & Sandback, M. (2015). Single- disorder. Autism, 18, 658–671.
case analysis and review framework (SCARF). Available at: https:// Wendt, O., & Miller, B. (2012). Quality appraisal of single-subject exper-
docs.google.com/forms/d/1_uoEKNXChXs_xOsT_ imental designs: an overview and comparison of different appraisal
tcPoiYuyil5Y11g14k2_QnWXes/viewform tools. Education and Treatment of Children, 35, 109–142.
McDuffie, A., Yoder, P., & Stone, W. (2005). Prelinguistic predictors of Wetherby, A. M., & Prizant, B. M. (1993). Communication and symbolic
vocabulary in young children with autism spectrum disorders. behavior scales manual (Normed ed.). Chicago: Riverside.
Journal of Speech, Language, and Hearing Research, 48, 1080– *Wetherby, A. M., & Woods, J. J. (2006). Early social interaction project
1097. for children with autism spectrum disorders beginning in the second
Mundy, P., & Crowson, M. (1997). Joint attention and early social com- year of life: a preliminary study. Topics in Early Childhood Special
munication: implications for research on intervention with autism. Education, 26, 67–82.
Journal of Autism and Developmental Disorders, 27, 653–676. Whalen, C., Schreibman, L., & Ingersoll, B. (2006). The collateral effects
Mundy, P., Sigman, M., Ungerer, J., & Sherman, T. (1986). Defining the of joint attention training on social initiations, positive affect, imita-
social deficits of autism: the contribution of non-verbal communi- tion, and spontaneous speech for young children with autism.
cation measures. Journal of Child Psychology and Psychiatry, 27, Journal of Autism and Developmental Disorders, 36(5), 655–664.
657–669. White, P. J., O’Reilly, M., Streusand, W., Levine, A., Sigafoos, J.,
Mundy, P., Sigman, M., & Kasari, C. (1990). A longitudinal study of joint Lancioni, G., et al. (2011). Best practices for teaching joint attention:
attention and language development in autistic children. Journal of a systematic review of the intervention literature. Research in Autism
Autism and Developmental Disorders, 20, 115–128. Spectrum Disorders, 5(4), 1283–1295.
Mundy, P., Sigman, M., & Kasari, C. (1994). Joint attention, develop- Wolery, M. (2013). Single-case design technical document of the What
mental level, and symptom presentation in autism. Development and Works Clearinghouse: a commentary. Remedial and Special
Psychopathology, 6, 389–401. Education, 34, 39–43.
Mundy, P., Delgado, C., Block, J., Venezia, M., Hogan, A., & Seibert, J. *Wong, C. S. (2013). A play and joint attention intervention for teachers
(2003). The early social communication scales. University of of young children with autism: a randomized controlled pilot study.
Miami. Available from: http://www.ucdmc.Ucdavis.edu/ Autism, 17, 340–357.
mindinstitute/ourteam/faculty_staff/esces.pdf [last accessed 11 Wong, C., Odom, S.L., Hume, K., Cox, A.W., Fettig, A., Kucharczyk,
June 2014]. S., & Schultz, T.R. (2014). Evidence-based practices for children,
Murza, K. A., Schwartz, J. B., Hahs-Vaughn, D. L., & Nye, C. (2016). youth, and young adultswith Autism Spectrum Disorder. Chapel
Joint attention interventions for children with autism spectrum dis- Hill: The University of North Carolina, Frank Porter Graham
order: a systematic review and meta-analysis. International Journal Child Development Institute, Autism Evidence-Based
of Language and Communication Disorders, 51, 236–251. PracticeReview Group.
Rev J Autism Dev Disord

*Yoder, P., & Stone, W. L. (2006). Randomized comparison of two com- Yoder, P. J., & Warren, S. F. (1999). Facilitating self-initiated proto-de-
munication interventions for preschoolers with autism spectrum dis- claratives and proto-imperatives in prelinguistic children with devel-
orders. Journal of Consulting and Clinical Psychology, 74, 426–435. opmental disabilities. Journal of Early Intervention, 22(4), 337–354.
Yoder, P. J., & Warren, S. F. (1998). Maternal responsivity predicts the
prelinguistic communication intervention that facilitates generalized Publisher’s Note Springer Nature remains neutral with regard to juris-
intentional communication. Journal of Speech Language and dictional claims in published maps and institutional affiliations.
Hearing Research, 41, 1207–1219.