Article Review

• • • •

Review Article

Systematic Review of Variables Related to Instruction in Augmentative and Alternative Communication Implementation: Group and Single-Case Design Joe Reichle,a James E. Pustejovsky,b Kimberly J. Vannest,c Margaret Foster,d

Lauren M. Pierson,e Sanikan Wattanawongwan,f Man Chen,b Marcus C. Fuller,g

April N. Haas,h Bethany H. Bhat,i Mary Rose Sallese,j S. D. Smith,k Valeria Yllades,l

Daira Rodriguez,f Amara Yoro,f and J. B. Ganzf

a Department of Speech-Language-Hearing Sciences, University of Minnesota, Minneapolis b Department of Educational Psychology, University of Wisconsin–Madison c Department of Education, University of Vermont, Burlington d Center for Systematic Reviews and Research Syntheses, School of Medicine, Texas A&M University, College Station e Department of Social Work and Communication Disorders, Tarleton State University, Fort Worth, TX f Department of Educational Psychology, Texas A&M University, College Station g Department of Education, University of Maryland Eastern Shore, Princess Anne h Life Skills Autism Academy, Plano, TX i Department of Educational Psychology, University of Texas at Austin j Department of Curriculum and Instruction, University of Alabama-Birmingham k Department of Elementary, Early, & Special Education, Southeast Missouri State University, Cape Girardeau l Galliant, San Antonio, TX

A R T I C L E I N F O

Article History: Received September 26, 2022 Revision received January 30, 2023 Accepted March 9, 2023

Editor-in-Chief: Erinn H. Finke Editor: Billy T. Ogletree

https://doi.org/10.1044/2023_AJSLP-22-00314

Correspondence to Sanikan Wattanawongwan: [email protected]. Disclosure: The authors have declared that no competing financial or nonfinancial interests existed at the time of publication.

A B S T R A C T

Purpose: This article provides a systematic review and analysis of group and single-case studies addressing augmentative and alternative communication (AAC) intervention with school-aged persons having autism spectrum disorder (ASD) and/or intellectual/developmental disabilities resulting in complex commu- nication needs (CCNs). Specifically, we examined participant characteristics in group-design studies reporting AAC intervention outcomes and how these com- pared to those reported in single-case experimental designs (SCEDs). In addition, we compared the status of intervention features reported in group and SCED studies with respect to instructional strategies utilized. Participants: Participants included school-aged individuals with CCNs who also experienced ASD or ASD with an intellectual delay who utilized aided or unaided AAC. Method: A systematic review using descriptive statistics and effect sizes was implemented. Results: Findings revealed that participant features such as race, ethnicity, and home language continue to be underreported in both SCED and group-design studies. Participants in SCED investigations more frequently used multiple com- munication modes when compared to participants in group studies. The status of pivotal skills such as imitation was sparsely reported in both types of studies. With respect to instructional features, group-design studies were more apt to utilize clinical rather than educational or home settings when compared with SCED studies. In addition, SCED studies were more apt to utilize instructional methods that closely adhered to instructional features more typically character- ized as being associated with behavioral approaches. Conclusion: The authors discuss future research needs, practice implications, and a more detailed specification of treatment intensity parameters for future research.

American Journal of Speech-Language Pathology Vol. 32 1734–1757 July 2023 Copyright © 2023 American Speech-Language-Hearing Association1734

In the United States, approximately one in 36 chil- dren has been identified with autism1 spectrum disorder (ASD), according to estimates from the Centers for Dis- ease Control and Prevention’s (n.d.) Autism and Develop- mental Disabilities Monitoring. ASD is reported to occur in all racial, ethnic, and socioeconomic groups, although minoritized groups are often diagnosed later, thus delaying access to services. Furthermore, one in six (17%) children aged 3–17 years are diagnosed with a developmental dis- ability, as reported by parents. This means that between 1.1% and 1.34% of children between 3 and 17 years of age experience intellectual delay and roughly 2%–3% of chil- dren are diagnosed with ASD. The overall percentage of children and youth with developmental disabilities is approximately 5%–7% (Zablotsky et al., 2017).

1 This article is written using people-first language, for the sake of consistency. The authors acknowledge that some prefer identity-first language and that the use of person-first language in this article does not imply preference for one over the other.

Some individuals with developmental disabilities require augmentative and alternative communication (AAC) to serve as an interim tool in acquiring speech, whereas others are likely to utilize AAC as a lifelong tool in supplementing communication (Beukelman & Light, 2020; Hustad & Miles, 2010; Johnston et al., 2012). Recent reports suggest that between 0.5% and 1.5% of the population use AAC applications or would benefit from AAC (Beukelman & Light, 2020; Beukelman & Mirenda, 2005).

There are a number of reasons for considering an aided or unaided AAC system. Some individuals with developmental disabilities (a) experience a cognitive impairment that has resulted in a substantial delay in speech and/or language acquisition, (b) produce speech that is not sufficiently intelligible to be understood by most or by a critical subset of communicative partners, (c) have difficulty acquiring communicative behavior at a rate that allows them to readily participate in a range of com- municative exchanges, or (d) have word-finding or other memory limitations that make it difficult to construct mes- sages commensurate with their ability to understand lan- guage morphology and syntax (Beukelman & Light, 2020). This range of potential communicative challenges and needs warrants consideration of participant, instruc- tional, and design variables in building on the evidence- based advances in AAC.

Social-communicative challenges represent a distin- guishing characteristic of persons with ASD and for many individuals diagnosed with an intellectual or developmen- tal disability (IDD; American Psychiatric Association Division of Research, 2013). The magnitude of the

challenges to interventionists is more clearly understood when one considers that approximately 30% of children with ASD are minimally verbal and many do not develop a substantial repertoire of spoken language (Tager-Flusberg & Kasari, 2013). Consequently, the applicability of AAC systems with this population is significantly greater than that in the general population.

Child Characteristics Associated With AAC Acquisition

Individuals using AAC who have ASD and/or IDD are diverse. As a result, our knowledge regarding match- ing characteristics and precursor skills of individuals to strategies for AAC interventions is limited. For instance, one meta-analysis (Ganz et al., 2014) suggested that, while AAC applications had a universally positive effect on speech outcomes, speech-generating devices (SGDs) were more effective with persons who did not experience an intellectual disability and ASD, and the Picture Exchange Communication System (PECS), a low-tech exchange- based aided AAC system, was more effective for persons with ASD who also had intellectual or other developmen- tal disabilities.

Few meta-analyses of educational or speech and lan- guage interventions report racial/ethnic composition of users. Gaias et al. (2020) reported that 27% of empirical investigations and 94% of meta-analyses did not report race/ethnicity. Their results further indicated that Asian, Native American, and Latino/a populations are underrep- resented in the research literature and that race, ethnicity, or home language of participants is rarely reported. More specifically, with respect to AAC and individuals with ASD/IDD, Ganz et al. (2023) noted a similar lack of reporting of race, ethnicity, and home language in the empirical literature that uses single-case experimental designs (SCEDs). In addition, there is an underrepresenta- tion of minoritized groups who experience ASD/IDD, their parents, and their educators in SCEDs. However, it is unclear whether this underrepresentation extends to group-design research in this specific area. This lack of racial and ethnic information could result in empirical research that is not applicable to the population as a whole and could inhibit the possibility of tailoring AAC interventions to be culturally sensitive to the needs of learners and their families.

With respect to age, ample literature has reported successful outcomes for young children with ASD and/or IDD who learn to use AAC (Ganz et al., 2011, 2014, 2017; Sievers et al., 2018). Older individuals are also responsive to intervention and should be eligible for AAC interventions, yet the literature on older individuals is sparse (Holyfield et al., 2017).

Reichle et al.: Systematic Review of Variable Instruction in AAC 1735

• • •

With respect to communication mode, it is possible that a learner’s experience with a mode(s) being taught could have an impact on acquisition during intervention. To date, few studies have examined this topic with learners experiencing complex communication needs (CCNs). van der Meer, Didden, et al. (2012) and van der Meer, Sutherland, et al. (2012) demonstrated that preference for a particular mode can influence instructional motivation with a particular communi- cation mode. If the mode used by a learner before intervention is a mismatch to the communicative mode being taught, inter- ference may be created in learning new symbols. Alternatively, if a communicative mode used before intervention is a match to that being implemented during intervention, it may facilitate the acquisition of new symbols as a result of mode familiarity. Although communication mode may influence acquisition for some learners, this has rarely been investigated.

Regardless of communication mode being taught, there may be some skills that facilitate communication acquisition. Pivotal skills can be described as those that, in and of themselves, are not required to acquire AAC, but the presence of which may positively influence the acquisition of AAC skills. In addition, these pivotal skills may play a role in the collateral acquisition of spoken communicative behavior. One of these pivotal skills is imi- tation. Gevarter et al. (2013) reported that learners with strong motor imitation skills also matched pictures well and did well with multiple types of AAC, whereas those with low motor imitation and more modest picture- matching skills did better with picture exchange systems. Flippin et al. (2010) reported a meta-analysis to investigate pre-AAC intervention child characteristics that predict com- munication outcomes for young children with ASD and IDD. Participants who (a) were somewhat able to verbally imitate before intervention, (b) demonstrated joint atten- tion, and (c) had higher levels of object exploration at base- line were more responsive to PECS intervention.

Identifying Instructional Characteristics That Influence Intervention Outcomes

There are a number of instructional characteristics, such as setting or number of settings in which intervention is implemented, that may influence intervention outcomes. For example, Brady et al. (2013) analyzed communication partner input and the amount of instruction at home and school. They reported that, 1 year later, there was a positive relationship between adult input at home and child commu- nication outcomes, but not between teacher input and child outcomes at school. Most research has been implemented in school or clinical settings (Biggs et al., 2019; Gevarter & Zamora, 2018; Holyfield et al., 2017; Logan et al., 2017; Therrien & Light, 2016) and infrequently in homes or other community settings (Shire & Jones, 2015).

Many instructional components, including strategies to evaluate preferences, specific prompting strategies, or sequencing of specific skills to be taught (Schlosser & Koul, 2015; Therrien & Light, 2016), have received lim- ited attention. Only two meta-analyses (Ganz et al., 2017; O’Neill et al., 2018) included statistical aggregation with respect to this topic. Whether the learner could imitate or alternatively was taught to imitate as part of the interven- tion has not often been addressed. A number of prompt- ing and prompt-fading strategies have some evidence of promise (Ganz et al., 2017; Holyfield et al., 2017; Shire & Jones, 2015; Therrien & Light, 2016). However, most existing systematic reviews have not statistically aggre- gated or compared results by instructional strategy. Ganz et al. (2017) conducted a meta-analysis limited to high- tech AAC, reporting that interventions including prompt fading produced stronger effects than those that did not. As was the case with previously discussed variables, most reviews included heterogeneous populations of individuals with CCNs (e.g., co-occurring with physical disabilities and speech impairment in addition to students with ASD and IDD; Allen et al., 2017; Biggs et al., 2019; O’Neill et al., 2018), making it more challenging to determine the relationship between learner characteristics and moderat- ing variables such as prompt fading or certain pivotal skills (e.g., imitation, joint attention).

An important aspect of instructional strategies involves the general approach to implementation. Natural- istic intervention approaches, such as child-led instruction (Logan et al., 2017) and embedding AAC instruction within authentic activities (Gevarter & Zamora, 2018; Shire & Jones, 2015), are increasingly prevalent in AAC intervention. Although researchers and practitioners regu- larly call for implementing AAC in authentic environ- ments, only a small proportion of studies have reported AAC implementation within authentic contexts and/or with natural communication partners (Biggs et al., 2019; Holyfield et al., 2017; Logan et al., 2017; Shire & Jones, 2015). Relatively few experiments have comprehensively examined outcome differences between authentic and con- trived contexts (Ganz et al., 2017).

Selection of the AAC modes to be implemented dur- ing intervention research requires a number of consider- ations that include preference and relative efficacy of modes. Somewhat limited guidance is available regarding selection of communication modes for individuals who would benefit from AAC. Researchers infrequently pro- vide a rationale for the augmentative communication modes selected for implementation (Ganz et al., 2022). Although some reviews, meta-analyses, and primary research studies have investigated the relative effectiveness of varied modes of AAC with individuals with ASD and/ or IDD, in the aggregate, they have produced somewhat

1736 American Journal of Speech-Language Pathology Vol. 32 1734–1757 July 2023

ambiguous outcomes (e.g., Achmadi et al., 2012; Ganz et al., 2014; Gevarter et al., 2013).

Outcomes other than acquisition are important in evaluating the efficacy of an intervention strategy. Some of these include generalization and maintenance outcomes. McLay et al. (2015) reported comparable acquisition regardless of communication mode selected but better gen- eralization and maintenance with options that involved selecting a graphic symbol versus unaided communication such as sign language. This might be the result of a less- ened memory load with graphic symbols (Johnston et al., 2012). Furthermore, low-tech systems involving symbol exchange between communicative partners may facilitate joint attention skills among early communicators (Johnston et al., 2012).

Research Design Considerations

Existing research on AAC interventions has used a range of designs, including both SCEDs (i.e., ABAB designs, multiple baselines across participants, multiple baselines across behaviors, alternating treatment designs) and between-group designs. However, many past system- atic reviews of the AAC literature have focused exclu- sively on studies that use SCEDs. Ideally, research synthe- ses that aim to inform evidence-based practice recommen- dations should take a comprehensive view of available research evidence, synthesizing and distilling findings from both single-case and group-design studies. Paradigmatic differences between types of research designs make this a difficult task—particularly if different types of study designs tend to be implemented with different profiles of partici- pants, different approaches to intervention, or different classes of dependent variables (Kratochwill et al., 2021). A necessary first step is to examine the characteristics of the available evidence base across both types of studies, to take stock of what research has been conducted, and to consider the similarities and differences in features of the SCED and group-design studies that have been conducted.

Purpose and Research Questions

To address the current knowledge gap, the following research questions represent the focus of the current article:

• What is the status of research addressing participant characteristics in group-design studies reporting AAC intervention outcomes for individuals with ASD and/or IDD who have CCN, and how does this compare to the reporting in SCEDs, with respect to (a) diagnosis; (b) chronological age; (c) ethnicity, race, and home language; (d) the number or type of communication modes used before intervention; (e)

the prompts used; and (f) the status of the learner’s imitation use before intervention?

• What is the status of reporting intervention features in group-design studies, and how does this compare to the reporting in SCEDs, with respect to instruc- tional strategies associated with more naturalistic as well as more behavioral intervention strategies (i.e., child or interventionist initiated, dispersed vs. massed teaching opportunities, contrived vs. embedded activ- ity contexts, group vs. one-on-one instructional for- mats, limited vs. varied teaching stimuli, controlled vs. natural instructional environments)?

Method

The purpose of this review was to synthesize the state of the literature involving AAC interventions for individuals with ASD and IDD, including SCED and group experiments. We report the process and findings following Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) 2020 guidelines (Page et al., 2021). See Figure 1 for a PRISMA flowchart that describes the search and screening process.

Eligibility Criteria

Studies were screened during title/abstract and full- text reviews to determine if each document met the fol- lowing criteria: (a) was written in English, (b) contained one or more participants with an intellectual delay or developmental disability with co-occurring CCNs (e.g., minimally verbal or nonverbal), (c) reported the results of an AAC intervention, (d) included a SCED or group design, and (e) included a measure of social-communicative or social-communicative and challenging behavior out- comes. All aspects of article search, inclusion/exclusion, identification, coding, and analysis procedures were con- ducted between 2018 and 2020 and are described in detail in paragraphs that follow, in Ganz et al. (2020). Articles were inclusive of those published in English between 1970 and 2020. Figure 1 displays a flowchart that summarizes the search and screening process of the review.

Literature Search and Screening Process

Literature searches were conducted by a professional reference librarian with 15 years of experience. The databases included Academic Search Complete, ERIC, Psy- cINFO, Conference Proceedings Citation Index – Social Sci- ence & Humanities (Web of Science), and Proquest Disserta- tions & Theses Global. The keywords were based on three focus areas of AAC: social communication and behaviors,

Reichle et al.: Systematic Review of Variable Instruction in AAC 1737

• • •

outcomes, and persons who experience ASD/IDD with CCN. Search terms for each search engine are included in PROS- PERO registration: CRD42018112428. This search yielded 7,327 documents (after deduplication) that were reviewed for potential inclusion. Documents were screened for inclusion/ exclusion at four distinct points in the review process: title/ abstract screening, full-text review, basic design quality stan- dards review (What Works Clearinghouse [WWC], 2019, 2020), and dependent variable screening. See Figure 1 for a PRISMA flowchart that describes the search procedures.

Figure 1. Preferred Reporting Items for Systematic Reviews and Meta-Analyses flowchart. PIs = principal investigators.

Title/Abstract Screening

For title/abstract screening, four raters reviewed all titles and abstracts of documents identified applying

inclusion/exclusion criteria via Rayyan, an online citation management system (Ouzzani et al., 2016). A total of 1,758 documents met the criteria after title/abstract review and were screened with full-text inclusion/exclusion criteria by four raters using Rayyan. Subsequently, 547 documents that remained after full-text screening were reviewed by applying basic design quality standards, following WWC criteria. Finally, 257 SCED documents were reviewed by lead authors to identify relevant dependent variables; during this process, the lead authors conducted a final review of each remaining document for inclusion/exclusion criteria.

All reliability reported was implemented by comput- ing agreements divided by agreements plus disagreements

1738 American Journal of Speech-Language Pathology Vol. 32 1734–1757 July 2023

multiplied by 100%. Four special education doctoral stu- dents coded title/abstract. Each had substantial experi- ence and expertise in conducting meta-analyses and sys- tematic reviews around the topic of AAC. All raters were trained to code until they reached a minimum criterion of 80% agreement. We assigned each rater to practice approximately 20–30 articles per week. Subsequently, all raters discussed any coding disagreements and reached consensus on disagreements. After doing this, they began independent coding. Reliability was collected on 100% of all articles during the title/abstract stage with 88% agreement.

Basic Design Quality Standards Coding

For each included study, we examined the first reported figure (provided that data met the inclusion criteria of the basic WWC design standards [WWC, 2019]). The six basic design standards include a syste- matically manipulated independent variable, measured and reported interobserver agreement (IOA), a mini- mum of 20% IOA collected across data in baseline and intervention separately, at least 80% or .60 kappa IOA scores, and at least three data points per baseline and intervention phases and at least four data per interven- tion phase for alternating treatment design. Each docu- ment was given a rating of (a) met design standards, (b) met design standards with reservations, or (c) did not meet standards. The preceding review steps yielded 162 SCED articles, including a total of 522 participants, which were included in the full methodological stan- dards analysis.

Full-Text Screening

Four raters (doctoral students in special education) screened the full texts of 1,758 documents against full-text inclusion/exclusion criteria via Rayyan. Documents meet- ing eligibility criteria included a total of 132 reports on group-design studies and 547 reports of SCED studies. The remaining stages of review were tailored to the study design. Full-text reliability reported was implemented by computing agreements divided by agreements plus dis- agreements multiplied by 100%. Four special education doctoral students who have expertise in conducting meta- analyses and systematic reviews in AAC areas coded full- text review. We trained all raters to code until they reached a minimum criterion of 80% agreement. We assigned each rater to practice approximately 10–20 arti- cles per week. Before the raters began independent coding, all raters discussed all coding disagreements and made consensus on disagreements. Thirty-nine percent (39%) of the articles for the full-text stage were independently coded, yielding 88% agreement.

Basic Design Quality Standards Screening

The 132 group-design documents and 547 SCED documents were reviewed against basic design quality standards. For group designs, standards were modeled after relevant WWC standards and included criteria for use of random assignment, attrition, and baseline equiva- lence of the analytic sample (WWC, 2020). The inclusion/ exclusion stages (title/abstract, full-text inclusion/exclusion criteria, and basic design quality standards) are described in greater detail in a larger study (Ganz et al., 2020). A total of three raters evaluated 132 group-design documents and 547 SCED documents in the basic design standards screening stage. One rater was a professor in research methods and statistics who served as a principal investigator (PI) for the project and who has extensive experience and expertise in meta-analysis. Another two raters were doctoral students in quantitative methods. All raters discussed the screening criteria and were trained to code using the screening form. Reliability calculations for group design resulted in 89.1% accuracy across 101 of the included articles. Reliability calculations resulted in 89.44% reli- ability for SCED across 20% of 537 SCED documents.

Full Methodological Standards Coding

Full methodological standards for this analysis were developed drawing from expert methodological standards (i.e., Cook et al., 2014; Ganz & Ayres, 2018; Horner et al., 2005; Reichow et al., 2008). This rating scheme was developed from the abovementioned references and used previously with acceptable reliability (Hong et al., 2016; Morin et al., 2018; Neely et al., 2016; Wattanawongwan et al., 2022). Each dimension is summarized below and described in more detail in the Appendix (https://osf.io/cfgds/).

Two doctoral students in special education reviewed for methodological standards in all articles. They were trained on each criteria of screening by a PI discussing any discrepancies. Training and practice continued until at least 80% accuracy per each category was established. Any disagreements were reviewed by two raters until a consensus was reached. Using item-by-item agreement resulted in a mean agreement of 89.87% (range: 82%–

96%) across 20% of included articles.

Dependent Variable Coding Procedures

A total of 59 group-design and 257 SCED studies potentially meeting design criteria were reviewed by PIs to determine dependent variables that were relevant for raw data extraction purposes; during this process, the PIs con- ducted a final review of each remaining document for inclusion/exclusion criteria. A total of 24 group designs

Reichle et al.: Systematic Review of Variable Instruction in AAC 1739

• • •

met group-design standards and included relevant depen- dent variables; these studies were included in the descrip- tive review. A total of 114 SCED studies met basic design standards and included relevant dependent variables. These were included in the descriptive review.

Seven special education doctoral student coders were originally trained by a PI to code these variables. Three of these individuals coded participant characteristics that included (a) diagnosis (i.e., ASD, IDD), (b) chronological age, (c) race/ethnicity, (d) number of word/symbols/sign approximations produced before intervention, (e) commu- nication status before intervention, (f) cognitive/IQ status before intervention, (g) ASD diagnostic assessment to establish ASD diagnosis, (h) communication mode(s) used before study, (i) imitation status, and (j) joint attention sta- tus. Two of these three individuals coded (a) the interven- tion orientation (social behavioral, functional behavioral), (b) instructional features (i.e., environmental arrangement, preference/reinforcement assessment, reinforcement, model- ing, verbal prompting, physical prompting, prompt fading, graphic prompt), and (c) intervention setting. With respect to intervention orientation features, across both group- design and single-case studies, we examined whether partici- pants received instruction with more characteristics associ- ated with behavioral intervention strategies versus those associated with more naturalistic intervention approaches. Indicators associated with each of these approaches are described in Table 1 and were coded as mutually exclusive (e.g., participants were coded as receiving primarily interventionist-initiated or child-initiated learning opportunities).

Included articles were also coded by three doctoral students for the communication characteristics that included (a) communicative function (i.e., behavior regulation, social interaction, joint attention), (b) expressive and receptive communication (i.e., communication production, communication comprehension), (c) communication mode (i.e., natural gesture, manual sign, low-tech aided system,

mid- to high-tech SGD, vocal, verbal), and (d) function of the challenging behavior (if disclosed by author[s]). For descrip- tions of each dependent measure, refer to the Appendix.

Table 1. Characterization of two general intervention frameworks for early communication intervention.

Functional behavioral intervention characteristics Social behavioral intervention characteristics

• Majority of teaching opportunities interventionist initiated

• Largely massed trial teaching opportunities during early instructional phases

• Activities used in training may be contrived.

• Primarily one-to-one

• Limited different teaching stimuli used during acquisition training

• Distractions are minimized.

• In early phases, continuous schedule of reinforcement

• Reinforcers often selected based on preference assessment

• At least some of teaching opportunities are child initiated.

• Adults are responsive to child communicative overtures.

• Teaching opportunities are often distributed across the learner’s daily routine.

• Intervention opportunities are embedded in functional activities.

• One-to-one instruction does not represent the only or even the primary method to deliver instructional opportunities.

• A variety of different teaching situations, interventionists, and teaching stimuli used to promote generalization.

• Teaching typically occurs in authentic environments with other learners present.

Dependent Variable Screening Reliability

Each reliability outcome was computed using item- by-item percentage agreement reliability for individual variables and for aggregated variables. This process was repeated until overall reliability reached 85%. We assigned each rater to practice approximately 5–10 articles per week for variable coding. This process required approxi- mately 16–20 weeks for both group and SCED studies. Approximately seven to eight articles for group-design studies and 20–30 articles for SCED studies were coded before achieving reliability. A PI rated and resolved dis- crepancies following a thorough discussion among all raters. Reliability for each phase of the systematic review including design quality, participant, instructional, and outcomes variables are reported in Table 2.

Effect Sizes for Group Designs

For group designs, we calculated standardized mean difference effect sizes and associated sampling variances from reported summary statistics, following standard methods (Borenstein & Hedges, 2019). The Hedges’s g estimator (Hedges, 1981) was used because many of the group-design studies included a relatively small number of participants. To ensure reproducibility, the calculations were performed using computer scripts for R (R Core Team, 2022), which is a programming language and com- puting environment often used to conduct quantitative data analysis. Of the included group designs, only 12 stud- ies provided sufficient quantitative information for effect size calculations (i.e., means and standard deviations of outcomes in each group or other summary statistics from which effect sizes could be computed).

1740 American Journal of Speech-Language Pathology Vol. 32 1734–1757 July 2023

Table 2. Reliability for each phase of the systematic review including design quality, participant, instructional, and outcomes variables.

Stage of reliability

Group-design studies Single-case study design

Percentage of documents

coded Percent reliability

Percentage of documents

coded Percent reliability

Title/abstract 100% 88% 100% 88%

Full text 39% 88% 39% 88%

Design quality variables 77% 89% 20% 90%

Variable outcomes Participant characteristic

Diagnosis 67% 82% 30% 96%

Chronological age 67% 86% 30% 100%

Race/ethnicity 67% 100% 43% 95%

Communication modes before intervention

67% 91% 30% 93%

Imitation before intervention 67% 93% 30% 96%

Instructional characteristic

Setting 72% 83% 20% 90%–93%

Instructional features 72% 77%–93% 20% 83%–100%

Behavioral intervention strategies versus naturalistic strategies

72% 80%–97% 20% 89%–97%

Limited number compared to varied number of learning opportunities

72% 97% 20% 88%

Communication outcomes

Communication mode 72% 87% 22% 87%

Communicative function 72% 73% 22% 89%

Descriptive Analysis

For each variable that was the focus of the current investigation, the number and percentage of studies and the number and percentage of participants that provided data were tabulated and reported. Studies and participants were tabulated separately for group-design studies and SCED studies. For analysis of effect sizes from group- design studies, we created forest plots to illustrate the dis- tribution of effect size estimates and calculated descriptive summary statistics. We refrained from conducting formal meta-analysis of the effect sizes due to the high degree of heterogeneity in the interventions and outcomes examined in the included studies.

Software and Reproducibility

To ensure the accuracy and reproducibility of the investigation, all analyses were carried out in the R statis- tical computing environment. Datasets and computer code for replicating all reported analyses are made available in the supplemental materials in Ganz et al. (2020).

Results

Tables 3–7 report both frequency counts and per- centages for the number of studies and participants across SCED and group-design studies. For a given characteristic,

percentages may add up to more than 100% because cate- gories were not exclusive. For example, a study might include participants who were 0–3 years old and partici- pants who were 4–5 years old.

Participant Characteristics

Diagnosis About three fifths of participants in group studies

were diagnosed with ASD compared with three quarters of participants in SCED studies (including those with both ASD and IDD). Persons with IDD comprised less than one in six group study participants, compared to about one third of participants in SCED investigations. Diagno- sis was unknown or not reported for nearly one quarter of participants in group studies.

Chronological Age Figure 2 depicts the distribution of participant ages

in studies using group designs. Notably, just under half of the studies involved participants who were 5 years old or younger. Group studies involving older participants often included a very broad range of ages. Among group stud- ies, about one third of participants were between 0 and 3 years of age. In studies utilizing SCED, only one fifth of participants were between 0 and 3 years of age. Four- and 5-year-olds accounted for a smaller fraction of group par- ticipants than of SCED participants. This trend was similar for proportions of 6- to 10-year-olds. Eleven- to 14-year

Reichle et al.: Systematic Review of Variable Instruction in AAC 1741

• • •

olds were sparsely represented in group studies, comprising less than 10% of the participants, compared to 12% of SCED participants. Those who were 14 years and older accounted for 30% of group study participants but less than 10% of SCED participants.

Table 3. Distribution of participant characteristics.

Category

Group designs Single-case designs

Study, n (%)

Participant, n (%)

Study, n (%)

Participant, n (%)

Diagnosis

ASD 17 (71) 580 (57) 85 (74.6) 224 (67.9)

IDD 4 (17) 151 (15) 41 (36.0) 83 (25.2)

Other 2 (8) 30 (3) — —

Unknown 4 (17) 236 (23) — —

ASD and IDD — — 16 (14.0) 23 (7.0)

Age group (years)

0–3 13 (54) 324 (32) 37 (32.5) 62 (18.8)

4–5 16 (67) 162 (16) 53 (46.5) 105 (31.8)

6–10 12 (50) 161 (16) 54 (47.4) 97 (29.4)

11–14 7 (29) 67 (7) 25 (21.9) 39 (11.8)

14 and older 6 (25) 307 (30) 16 (14.0) 27 (8.2)

Race category

Not specified 23 (96) 728 (71) 86 (75.4) 226 (68.5)

White/Caucasian 4 (17) 205 (20) 29 (25.4) 71 (21.5)

Black/African American 3 (12) 63 (6) 14 (12.3) 21 (6.4)

Asian/Asian American 3 (12) 21 (2) 6 (5.3) 9 (2.7)

Native American 1 (4) 3 (0) — —

Indian American — — 1 (0.9) 1 (0.3)

Multiracial — — 2 (1.8) 2 (0.6)

Ethnicity category

Not specified 23 (96) 914 (90) 113 (99.1) 310 (93.9)

Not Latino/a or Hispanic 1 (4) 58 (6) 2 (1.8) 5 (1.5)

Latino/a or Hispanic 2 (8) 49 (5) 11 (9.6) 15 (4.5)

Home language

English 17 (71) 833 (82) 11 (9.6) 24 (7.3)

Not specified: article is published in English, but the first author’s affiliated country had a different dominant language

5 (21) 168 (16) — —

Not specified: article is in English and published in English-dominant setting/country and no language is specified

2 (8) 20 (2) — —

Amharic — — 1 (0.9) 1 (0.3)

Dual language — — 8 (7.0) 11 (3.3)

Dutch — — 1 (0.9) 2 (0.6)

Spanish — — 4 (3.5) 5 (1.5)

Not specified — — 103 (90.4) 287 (87.0)

Note. Em dashes indicate categories for which there were zero studies and zero participants. ASD = autism spectrum disorder; IDD = intel- lectual or developmental disability.

Race/Ethnicity Very few group-design studies and one quarter of

SCED studies provided information on participants’ race. With respect to ethnicity, all but one of the group studies and nearly all of SCED studies did not report participant ethnicity. Breakdowns of race and ethnicity that were

reported are displayed in Table 3. With respect to race, the largest proportion of participants was White in both group and SCED studies, followed by African American/Black and then Asian American/Asian. In studies for which race categories were reported, the distribution of participants was similar across group-design and single-case studies.

Home language was also scrutinized. Most group- design studies did specify home language, and in all cases, the participants spoke English. In contrast, the bulk per- centage of SCED studies did not explicitly specify home language. Only among SCED studies were non-English

1742 American Journal of Speech-Language Pathology Vol. 32 1734–1757 July 2023

home languages ever reported. A breakdown by specific language is displayed in Table 3.

Figure 2. Mean age and age range of study participants.

Communication Modes Before Intervention Over one third of the participants in group-design

studies used two or more communication modes before research participation. However, communication mode before intervention was not reported for approximately one third of participants (see Table 4). The most fre- quently used communication mode before intervention was verbalization, which was utilized by 13% of the participants, whereas no participants were reported to use AAC only. In comparison, SCED studies reported that about half of participants used two or more com- munication modes before study implementation. Pre-

intervention communication mode was not reported in 14% of the participants in these investigations. Similar to the group-design studies, verbalization was the most frequent communication mode used before intervention among those participating in SCED studies. In SCED studies, participants used a range of other modes before intervention, including one fifth who used any aided or unaided AAC mode as a singular mode before interven- tion. Use of mid- or high-tech communication aids was infrequently reported.

Imitation Before Intervention Participants’ imitation skills were rarely reported in

either group (reported in only 6% of participants) or SCED (reported in only 30% of participants) studies. A

Reichle et al.: Systematic Review of Variable Instruction in AAC 1743

• • •

further breakdown of imitation that was observed is enu- merated in Table 4.

Category

Table 4. Distribution of participant communication modes and imitation use.

Group designs Single-case designs

Study, n (%) Participant, n (%) Study, n (%) Participant, n (%)

Communication modes before intervention

Manual sign language 1 (4) 160 (16) 8 (7.0) 12 (3.6)

Verbalization 3 (12) 131 (13) 27 (23.7) 50 (15.2)

Two or more categories 8 (33) 371 (36) 74 (64.9) 158 (47.9)

Not reported 12 (50) 359 (35) 15 (13.2) 45 (13.6)

Natural gestures — — 18 (15.8) 32 (9.7)

Vocalization — — 8 (7.0) 14 (4.2)

Low-tech aided AAC — — 8 (7.0) 10 (3.0)

Mid- to high-tech aided AAC — — 4 (3.5) 9 (2.7)

Number of communication modes before intervention

1 4 (17) 291 (29) 63 (55.3) 127 (38.5)

2 5 (21) 299 (29) 51 (44.7) 82 (24.8)

3 2 (8) 64 (6) 32 (28.1) 50 (15.2)

4+ 1 (4) 8 (1) 14 (12.3) 26 (7.9)

Not reported 12 (50) 359 (35) 15 (13.2) 45 (13.6)

Imitation use before intervention

Vocal/verbal imitation 1 (4) 60 (6) 28 (24.6) 54 (16.4)

Not reported 23 (96) 961 (94) 92 (80.7) 230 (69.7)

Gestural imitation — — 15 (13.2) 31 (9.4)

Limited imitation — — 11 (9.6) 15 (4.5)

Note. AAC = augmentative and alternative communication.

Instructional Characteristics

Settings Clinics were the most frequent setting for instruction

implemented in group-design studies, encompassing about one third of participants. For group-design studies, class- rooms were a less frequently utilized setting, encompassing only one fifth of participants. However, classrooms were the most frequently used setting for studies employing SCED, followed by clinical settings. Home settings were the venue for about one fifth of participants in group studies but for only one tenth of the participants in SCED investigations. Approximately 30% of learners participated in interventions implemented in multiple settings in group experimental designs, compared to about 20% of partici- pants in SCED studies. A detailed breakdown of setting information is delineated in Table 5.

Instructional Features Features most often addressed in studies utilizing

group designs included modeling and reinforcement (both implemented with almost three quarters of participants). With respect to SCED studies, reinforcement, systematic

arrangement of the environment, and prompt fading were most often described, implemented in about 90% of studies. Somewhat less reported strategies used with participants in group studies included response prompts. Preference assessments (10%) were infrequently imple- mented with group-design participants. Among partici- pants in SCED studies, a similar rate of modeling and verbal prompts was reported, with physical prompts and preference assessments reported a bit less and graphic prompts infrequently used. Table 5 provides complete delineation results addressing instructional strategies used.

Behavioral Intervention Strategies Versus Naturalistic Strategies

Across both group and SCED studies, participants received more instructional characteristics associated with behavioral intervention strategies when compared with those associated with naturalistic intervention approaches. Our instructional indicators were coded as mutually exclu- sive (e.g., participants were coded as receiving primarily interventionist-initiated or child-initiated learning opportu- nities). Intervention strategies could always be coded for SCED studies but were unreported for a fraction of stud- ies using group designs. Detailed results are reported in Table 6.

1744 American Journal of Speech-Language Pathology Vol. 32 1734–1757 July 2023

Table 5. Distribution of instructional characteristics.

Category

Group designs Single-case designs

Study, n (%) Participant, n (%) Study, n (%) Participant, n (%)

Instructional setting

Classroom 4 (17) 220 (22) 40 (35) 115 (35)

Clinic 10 (42) 315 (31) 31 (27) 92 (28)

Home 5 (21) 181 (18) 12 (11) 31 (9)

Multiple 4 (17) 295 (29) 25 (22) 74 (22)

Other 1 (4) 10 (1) 5 (4) 15 (5)

Not reported — — 1 (1) 3 (1)

Instructional features

Graphic prompt 3 (12) 121 (12) 6 (5) 21 (6)

Modeling 15 (62) 742 (73) 75 (66) 216 (65)

Physical prompts 12 (50) 634 (62) 63 (55) 183 (55)

Prompt fading 13 (54) 514 (50) 103 (90) 295 (89)

Systematic arrangement

14 (58) 454 (44) 102 (89) 294 (89)

Verbal prompts 12 (50) 504 (49) 76 (67) 216 (65)

Reinforcement 19 (79) 724 (71) 106 (93) 304 (92)

Preference assessment 4 (17) 104 (10) 57 (50) 168 (51)

Number of instructional features

0 3 (12) 76 (7) — —

1 1 (4) 81 (8) 2 (2) 8 (2)

2 2 (8) 182 (18) 4 (4) 11 (3)

3 5 (21) 161 (16) 6 (5) 14 (4)

4 1 (4) 8 (1) 14 (12) 42 (13)

5 6 (25) 301 (29) 38 (33) 110 (33)

6 4 (17) 152 (15) 36 (32) 108 (33)

7 2 (8) 60 (6) 14 (12) 37 (11)

Educator- Versus Learner-Initiated Learning Opportunities

Interventionists initiated learning opportunities for over half of participants in group studies and over four fifths of participants in SCED studies (see Table 6).

Massed Versus Distributed Learning Opportunities

There was an overwhelming reliance on massed instructional opportunities, rather than distributing trials among other learning objectives, in both group-design and SCED studies. In addition, it was very common to use nonauthentic instructional opportunities, as over 60% of participants in group studies and over 80% of partici- pants in SCED studies received contrived learning oppor- tunities. Additional, more detailed results are contained in Table 6.

One-on-One Instruction In both group and SCED studies, over 70% of par-

ticipants were the recipients of one-on-one instruction compared to small or large group instruction.

Limited Number Compared to Varied Learning Opportunities

Teaching materials were coded as “limited” or “var- ied.” This is the only indicator for which both group and SCED studies implemented naturalistic approaches more frequently than behavioral approaches. During the imple- mentation of procedures in studies employing a group design, 60% of participants were provided with varied opportunities. This trend also held for participants in stud- ies utilizing a SCED, as 62% received instruction utilizing varied stimuli. Similar percentages of participants in group-design studies (54%) and SCED studies (48%) par- ticipated in controlled intervention contexts, rather than more naturalistic contexts. Opportunities embedded in natural instructional contexts occurred slightly more fre- quently (52%) for participants of SCED studies.

Communication Outcomes: Modes and Functions Utilized During Intervention

During intervention implementation, two or more communication modes (e.g., mid- to high-tech AAC plus verbalizations) were most frequently implemented for both

Reichle et al.: Systematic Review of Variable Instruction in AAC 1745

• • •

group-design and SCED research participants, although it was much less frequently implemented with SCED studies. Aided AAC was implemented as the only communication mode for less than 10% of the group participants but for 40% of the SCED participants. Sign language and gestures were taught to participants as the sole communication mode infrequently for both group and SCED study partic- ipants; similarly, vocalizations and verbalizations were infrequently taught for group and SCED study partici- pants. Group studies were coded as “other” communica- tion mode or not reported communication mode as the focus of intervention for about 10% of participants; in contrast, this rarely happened for SCED participants. Table 7 provides a more detailed breakdown of informa- tion regarding communication modes and communicative functions.

Table 6. Distribution of intervention strategy use.

Category

Group designs Single-case designs

Study, n (%) Participant, n (%) Study, n (%) Participant, n (%)

Number of behavioral intervention strategies

0 3 (12) 158 (15) — —

1 3 (12) 118 (12) 7 (6) 19 (6)

2 1 (4) 12 (1) 4 (4) 13 (4)

3 1 (4) 61 (6) 17 (15) 45 (14)

4 7 (29) 345 (34) 27 (24) 84 (25)

5 2 (8) 24 (2) 37 (32) 101 (31)

6 7 (29) 303 (30) 22 (19) 68 (21)

Initiator of teaching opportunity

Child 6 (25) 207 (20) 17 (15) 49 (15)

Interventionist 14 (58) 573 (56) 97 (85) 281 (85)

Not reported 4 (17) 241 (24) — —

Teaching opportunity distribution

Dispersed 2 (8) 148 (14) 12 (11) 36 (11)

Massed 19 (79) 770 (75) 102 (89) 294 (89)

Not reported 3 (12) 103 (10) — —

Activity context

Contrived 15 (62) 610 (60) 93 (82) 270 (82)

Embedded 6 (25) 308 (30) 21 (18) 60 (18)

Not reported 3 (12) 103 (10) — —

Instructional format

Group 4 (17) 185 (18) 10 (9) 30 (9)

Not reported 2 (8) 91 (9) — —

One-on-one 18 (75) 745 (73) 104 (91) 300 (91)

Teaching stimuli

Limited 9 (38) 396 (39) 41 (36) 124 (38)

Not reported 1 (4) 10 (1) — —

Varied 14 (58) 615 (60) 73 (64) 206 (62)

Instructional environment

Controlled 13 (54) 549 (54) 54 (47) 160 (48)

Natural 8 (33) 327 (32) 60 (53) 170 (52)

Not reported 3 (12) 145 (14) — —

Effect Sizes From Group-Design Studies

Of the 24 included studies, only 12 provided suffi- cient information to calculate effect sizes. Many of these studies included results for multiple outcomes (i.e., multi- ple assessments or multiple follow-up times) or multiple relevant treatment groups. We calculated and reported effect size estimates for all relevant outcomes and treat- ment groups. Within this subset, studies reported between 1 and 33 effect size estimates, with a median of 7 effect size estimates per study.

Figure 3 displays a forest plot of effect size estimates reported by each study, with studies ordered from smallest to largest average effect size estimate. Effect size estimates varied widely, both across studies and across outcomes

1746 American Journal of Speech-Language Pathology Vol. 32 1734–1757 July 2023

within a given study. Across the 144 effect size estimates from 12 studies, effect size estimates ranged from −2.17 to 2.17, with a median of 0.22. Averaging the effect size esti- mates within each study, study-level aggregated effects ranged from −2.17 to 0.69, with a median of 0.20. Due to the range of interventions and outcomes reported in these studies, we refrained from reporting a formal meta- analysis.

Table 7. Distribution of communication outcomes.

Category

Group designs Single-case designs

Study, n (%) Participant, n (%) Study, n (%) Participant, n (%)

Communication modes during intervention

Two or more categories 20 (83) 913 (89) 47 (41) 139 (42)

Low-tech aided AAC 1 (4) 61 (6) 18 (16) 57 (17)

Mid- to high-tech aided AAC 1 (4) 10 (1) 26 (23) 75 (23)

Manual sign/natural gestures 1 (4) 15 (1) 19 (17) 50 (15)

Verbalization/vocalization 1 (4) 39 (4) 14 (12) 36 (11)

Other 1 (4) 17 (2) — —

Not reported 3 (12) 98 (10) 1 (1) 3 (1)

Aided AAC vs. unaided AAC

Exclusively aided AAC 17 (71) 881 (86) 62 (54) 187 (57)

Exclusively unaided AAC 6 (25) 118 (12) 26 (23) 71 (22)

Exclusively verbalization/vocalization 1 (4) 39 (4) 14 (12) 36 (11)

Not reported 3 (12) 98 (10) 1 (1) 3 (1)

Both aided and unaided AAC — — 20 (18) 59 (18)

Vocalization/verbalization

Vocal/verbal 1 (4) 39 (4) 43 (38) 118 (36)

Not vocal/verbal 23 (96) 999 (98) 79 (69) 234 (71)

Not reported 3 (12) 98 (10) 1 (1) 3 (1)

Number of communication modes

1 5 (21) 142 (14) 68 (60) 193 (58)

2 19 (79) 905 (89) 29 (25) 86 (26)

3+ 2 (8) 70 (7) 22 (19) 64 (19)

Not reported 3 (12) 98 (10) 1 (1) 3 (1)

Communication systems

Comprehension 3 (12) 67 (7) 5 (4) 16 (5)

Production 22 (92) 978 (96) 107 (94) 310 (94)

Both 5 (21) 94 (9) 6 (5) 13 (4)

Not reported 5 (21) 129 (13) 2 (2) 6 (2)

Communication functions

Behavioral regulation 9 (38) 242 (24) 61 (54) 169 (51)

Joint attention 4 (17) 72 (7) 21 (18) 62 (19)

Social interaction 9 (38) 440 (43) 18 (16) 55 (17)

Multiple 3 (12) 119 (12) 12 (11) 34 (10)

Not reported 13 (54) 629 (62) 4 (4) 14 (4)

Note. AAC = augmentative and alternative communication.

Discussion

This review uncovered numerous differences—and some similarities—between group-design studies and SCED

studies in terms of the participant characteristics, interven- tion implementation factors, and communication modes measured and reported. Overall findings suggest that partic- ipant features such as race, ethnicity, and home language continue to be underreported in both SCED and group- design studies. The infrequent reporting of participants’ race, ethnicity, and home language represents a significant challenge. Many aspects of intervention are dependent on understanding the culture of the recipients of intervention and their families. Without basic information about the demographic diversity of participants, it is impossible to consider this. Furthermore, without addressing these vari- ables, it is unclear the extent to which a given intervention is suitable for a particular learner.

Reichle et al.: Systematic Review of Variable Instruction in AAC 1747

• • •

Figure 3. Forest plot of effect size estimates from group-design studies.

SCED studies differed from group-design studies in focusing on participants at younger ages, with compara- tively few participants over the age of 14 years. A reason- able hypothesis is that younger children with less commu- nication delay might fall into a learner-initiated interven- tion, whereas older children with a more significant delay might have their needs best met via an educator-initiated treatment. An important variable may involve a learner’s rate of spontaneous communicative acts. With few sponta- neous acts, there is little for an interventionist to build on.

Participants in SCED investigations more frequently used multiple communication modes when compared to participants in group studies. In particular, group stud- ies were far less likely to report the status of the com- municators before implementation than were SCED

studies. Furthermore, nearly half of the SCED partici- pants were reported to use two or more communication modes compared to approximately one third of the par- ticipants in the group studies. Not reporting communi- cation mode used before intervention may be problem- atic in that prior exposure and use may prove to enhance performance during the implementation of the independent variable in studies focused on AAC inter- vention. We do not yet have a good understanding why these differences occurred, but the findings suggest that it is somewhat difficult to compare results across these two types of research methodologies. Further scrutiny is needed to determine if there are certain population or methodological biases inherent in the two types of study design or whether our findings represent a fluke of sampling.

1748 American Journal of Speech-Language Pathology Vol. 32 1734–1757 July 2023

The status of pivotal skills such as verbal and ges- tural imitation was seldom reported in either SCED or group studies. By reporting the status of imitation skills, researchers facilitate greater potential flexibility in response prompts that can be used to teach new skills. Learners who have mastered imitation often acquire new skills more quickly than those who have not and may be able to or are on the cusp of being able to learn using more vicarious learning strategies. For example, it is likely that individuals with ASD/IDD and CCNs with prior verbal imitation skills may require a lower dosage of AAC intervention (Flippin et al., 2010). Those with less developed motor imitation skills may respond better to picture exchange systems rather than high-tech AAC (Gevarter et al., 2013). Conse- quently, the disclosure of information regarding skills such as imitation can better inform both researchers and practitioners.

With regard to instructional practices, implementa- tion settings varied across group and SCED studies. Group studies relied more on clinical settings (including university- connected labs) and infrequently in more authentic class- room and home settings. That said, although SCED studies more often involved implementation in classroom and home settings, over one quarter of studies were conducted in clinical settings. For both types of designs, researchers implemented intervention across multiple settings in fewer than 25% of studies (less than 30% of participants). This could have negative implications for generalization and maintenance. Reliance on inauthentic settings is problem- atic given the greater maintenance of communication out- comes when AAC interventions are implemented in both school and home settings (Brady et al., 2013).

In this investigation, we have organized some com- ponents of intervention into more developmental versus more behavioral in orientation. Given this characteriza- tion, behaviorally oriented instructional features were used more frequently in SCED studies but were also imple- mented frequently within group-design studies. This is not surprising given the high proportion of participants with ASD. Participants overwhelmingly received interventions in contexts that were interventionist initiated, involved massed teaching opportunities, were incorporated within contrived activity contexts, and provided one-on-one instruction. Use of naturalistic approaches exceeded use of more structured behavioral approaches in only two areas. That is, teaching stimuli were more often varied, rather than limited, in both group and SCED studies, and imple- mentation took place in natural instructional environ- ments slightly more than half the time for SCED partici- pants. Naturalistic approaches are increasingly recom- mended and have been demonstrated to be effective (Bruinsma et al., 2020), particularly in promoting generali- zation and maintenance of interventions and learning.

However, the research is yet to catch up to these recom- mendations, as noted by other authors reporting on the literature on AAC interventions (Ganz et al., 2017; Holyfield et al., 2017; Shire & Jones, 2015; Therrien et al., 2016).

This model of understanding environment–behavior relations can be applied across a broad variety of training and intervention strategies. More structured approaches such as discrete trial training can be analyzed and under- stood within a framework that includes motivating opera- tions. This same framework (e.g., Leaf & McEachin, 1999) can be applied to what have sometimes been referred to as naturalistic developmental behavioral inter- ventions (Schreibman et al., 2015). Often included in this category are approaches such as incidental teaching (McGee, 2005), pivotal response training (Koegel & Koegel, 2019), and the Early Start Denver Model (Dawson et al., 2010). It is important to note that proponents of such approaches have consistently emphasized that they are fully consistent with applied behavior analytic principles and procedures (e.g., Schreibman et al., 2015). Thus, although minimizing distractions, teaching one-to-one, and so forth are often attributed to more behavioral approaches, these features are increasingly being embraced as part of a continuum that can be very helpful during the early phases of intervention.

The use of physical prompts, verbal prompts, and prompt fading is aligned with prior findings that prompt- ing and prompt fading are frequently implemented during AAC intervention (Ganz et al., 2017; Holyfield et al., 2017; Shire & Jones, 2015; Therrien & Light, 2016). The high prevalence in the use of tangible reinforcers is likely because implementers were focusing more on behavioral regulation (e.g., object requesting) than on more socially oriented communication functions.

Future Areas of Research Need

Further research in a number of areas would better inform intervention practices in AAC with the population that is the focus of this investigation. We will briefly address some of these areas that have recently received increasing attention in the mainstream of the language intervention literature.

Race/Ethnicity/Home Language The research community continues to do an under-

whelming job of specifying race, ethnicity, and home lan- guage of participants. This information is important to determine if other variations of interest differ across popu- lations. In addition, a critical feature of selecting partici- pants is balancing the distribution of participants’ race, ethnicity, and home language. The current literature base

Reichle et al.: Systematic Review of Variable Instruction in AAC 1749

• • •

excludes major components of the U.S. population, which is a reflection of generations of racist and exclusionary policies and practices. Ethical researchers must address needs of minoritized populations (Maenner et al., 2020; Magaña et al., 2013). Researchers can increase trust through community-based rapport building (Steinbrenner et al., 2022) and by including diverse and disabled educa- tors and parents across all phases of research (Kerkhoff et al., 2022; McNulty et al., 2019). Researchers must elim- inate the exclusionary tactics, such as research sites based in majority-population settings, unavailability of appropri- ate language and translation services, and historically and currently racist practices and policies (Steinbrenner et al., 2022).

Cognitive/Communication Assessment Features In the set of studies reviewed, it was very difficult to

compare the language and cognitive status of participants across studies. Few standard assessment protocols were used. Going forward, identifying a core set of recom- mended assessment tools for researchers to use would be very helpful and would facilitate comparison and generali- zation across investigations.

Pivotal Skills Imitation, as a mediator in establishing collateral

gains resulting from AAC intervention, has received lim- ited attention, despite being frequently cited as associated with an increase in vocal/verbal output among those who are recipients of aided and unaided communication inter- vention. Various elements of the extant literature have suggested that verbal imitation may be associated with the acquisition of collateral speech as a function of both unaided and aided communication systems (Johnston et al., 2012; Millar et al., 2006; Yoder & Layton, 1988).

In the current investigation, we did not directly examine receptive language skills. However, this is an important area for future scrutiny. Learners acquiring pro- ductive AAC skills have been reported to make corre- sponding gains in their receptive communication skills (e.g., Harris & Reichle, 2004). The degree of bidirectional- ity is also a topic that has been sparingly explored. To what extent is more prolific comprehension of spoken lan- guage associated with better performance during AAC production intervention? With the population that is the focus of this article, this topic requires further scrutiny. In many of the studies that were reviewed, there was very limited information describing participants’ language com- prehension skills.

Treatment Intensity Currently, there is limited experimental literature

addressing treatment intensity with persons with ASD and

IDD who experience CCNs (Reichle et al., 2021). We know even less about how aspects of treatment intensity interact with learner characteristics (e.g., specific cogni- tive status, language comprehension, and/or production proficiency). Further research in this area would allow interventionists to better individualize interven- tion protocols.

Maintenance, Generalization, and Conditional Use of Treatment Outcomes

There is little known about long-term maintenance of newly taught communicative behavior. It is rare that communication maintenance is scrutinized beyond 1 month after intervention. When maintenance data are presented, it is often for a period of a week to a month. However, we also know that time post acquisition is a threat to maintenance. This represents a particularly important topic among those who received functional communica- tion training to address a problem behavior that served an important communicative function. After mastery, an identified set of competing socially unacceptable behaviors identified before training could reemerge if their reinforce- ment history becomes more attractive than that for the newly established communicative alternative. In part, this is why it is so important that, at the conclusion of acquisi- tion, care be taken to examine natural maintaining contin- gencies (Stokes & Baer, 1977) that may be available in the environments in which the individual will operate. None of the articles reviewed examined this issue.

The possible interaction of communication mode with maintenance and generalization is a topic that has been rarely explored. For example, if a learner generalizes, was generalization supported by natural maintaining con- tingencies? If so, learners who generalize may derive a maintenance advantage in that they are locating contexts where use of a newly established behavior is apt to be rein- forced, which, in turn, will serve to maintain the behavior.

Put simply, conditional use means using a particular communication symbol when appropriate and refraining from using it when it is inappropriate to do so. Horner et al. (1986) described an early practical application of an instructional format “general case instruction” designed to concurrently establish conditional use. Since then, several investigators have described applications in communica- tion. In addition to what we have discussed thus far, gen- eral case instruction also utilizes multiple teaching exam- ples that sample critical stimulus features comprising the concept being taught. For example, in teaching “apple,” red, yellow, green, big, and little examples would be used. However, this paradigm has been used sparingly in the lit- erature pertaining to the population that is the focus of this article and was not fully executed in any of the inves- tigations in the current analysis.

1750 American Journal of Speech-Language Pathology Vol. 32 1734–1757 July 2023

Navigation. In studies using SGDs, very few have addressed navigational skills that include scrolling, page linking, and managing pop-ups, among others. Few researchers have addressed the relative advantage of teaching navigation skills independently of early vocabulary skills during the early phases of aided AAC intervention. With the increasing sophistication of SGDs, this is a topic that merits additional attention.

Summary

Participant characteristics, including cognitive level, receptive and expressive communication skills before inter- vention, race/ethnicity, and prior interventions, are all rel- evant variables in interpreting the potential efficacy (including contextual fit) of an intervention protocol for a given learner. Currently, many of these parameters are not described or are poorly described in a large proportion of the research literature.

An important initial step in better understanding AAC intervention protocols is to be able to replicate the procedure described in a research study. Unfortunately, this would be challenging in many of the investigations that were the focus of this systematic review. Often, ante- cedent conditions before an intervention opportunity were not well described. In addition, details of prompting and prompt-fading strategies were often not clearly articulated. We also found that treatment intensity parameters sum- marized by Warren et al. (2007) were not available. Pro- cedures to address longer term maintenance and generali- zation were often not addressed. When maintenance was mentioned, it was for a relatively brief period. Generali- zation analyses often focused on a single aspect (e.g., across settings). Generalization probes often contained a common stimulus (e.g., the interventionist who was also present during intervention). Consequently, it was diffi- cult to differentiate good stimulus control from general- ized performance.

It is important to acknowledge that the intervention landscape is changing due, in part, to the COVID-19 pan- demic and advances in telehealth technology. These changes are apt to have significant implications for main- tenance and generalization going forward. It is also likely to spawn substantial interest in studying reliability and fidelity with respect to parents and other stakeholders as prospective interventionists. Although it is generally acknowledged that applied researchers report reliability with respect to scrutinizing learner behavior, they are often far less likely to address fidelity. Fidelity is critical in that, during intervention, it ensures implementation consistency of intervention procedures. An intervention with a higher proportion of teaching opportunities stands to better ensure a less confusing task for the learner. It is

beyond the scope of this article to provide a detailed dis- cussion of fidelity. However, we believe that it is an important parameter of treatment intensity. Furthermore, it can be challenging to evaluate in settings with multiple interventionists.

Finally, this study has several limitations. We did not perform a meta-analysis as we have previously done so with SCED studies with respect to participant charac- teristics, instructional strategies, and design quality. There were relatively few group-design studies, and the group designs identified included highly heterogeneous participant- and intervention-related characteristics. Con- sequently, we chose to descriptively summarize and make comparisons between group and SCED studies, where possible. All included studies met WWC criteria for design quality, which eliminated a number of investiga- tions that could have had an impact on participant or instructional methodology outcomes.

Is There a Need for a Gold Standard Protocol Describing Key Study Information?

At present, there is no widely used standard for describing information that we believe should be included in an empirical investigation. WWC (2019) has generated stan- dards addressing both group and single-participant experi- mental design characteristics as well as some recommenda- tions in other areas. Standards have been proposed for treat- ment intensity/treatment dosage (e.g., Warren et al., 2007) but sparingly reported in our exploration of the studies in this investigation. Still, other groups have addressed some aspects of participant characteristics (https://apastyle.apa. org/products/publication-manual-7th-edition).

To some degree, investigators seem to have difficulty understanding the importance of specifying participant, interventionist, and instructional characteristics addressed in this article. Given that research on intervention with persons experiencing complex communication predomi- nantly makes use of SCED, replication is essential in strengthening the evidence base. Although the community of medical researchers has placed a premium on replica- tion, it is a much less emphasized aspect of communica- tion intervention research, to the detriment of establishing a sound evidence base. It is possible that scholarly jour- nals may have to adopt more stringent guidance and stan- dards for reporting required information.

The purpose of this article was to describe the cur- rent status of the extent to which evidence-based literature addresses variables that are relevant to a comprehensive database that will influence descriptions of exemplary intervention. Generally, our findings call for more rigor- ous requirements for the description of variables in com- munication intervention research. To accomplish this, it

Reichle et al.: Systematic Review of Variable Instruction in AAC 1751

• • •

may be necessary for funding agencies and journals to institute policies requiring more rigorous reporting stan- dards with respect to participant, interventionist, and instructional characteristics.

Data Availability Statement

The data sets generated during and/or analyzed dur- ing this study are available in the Open Science Frame- work repository, https://edarxiv.org/nzq4h/.

References

Achmadi, D., Kagohara, D. M., van der Meer, L., O’Reilly, M. F., Lancioni, G. E., Sutherland, D., Lang, R., Marschik, P. B., Green, V. A., & Sigafoos, J. (2012). Teaching advanced operation of an iPod-based speech-generating device to two students with autism spectrum disorders. Research in Autism Spectrum Disorders, 6(4), 1258–1264. https://doi.org/10.1016/j. rasd.2012.05.005

Allen, K. A., Bredero, B., Van Damme, T., Ulrich, D. A., & Simons, J. (2017). Test of Gross Motor Development-3 (TGMD-3) with the use of visual supports for children with autism spectrum disorder: Validity and reliability. Journal of Autism and Developmental Disorders, 47(3), 813–833. https:// doi.org/10.1007/s10803-016-3005-0

American Psychiatric Association Division of Research. (2013). Highlights of changes from DSM-IV to DSM-5. Focus, 11(4), 525–527. https://doi.org/10.1176/appi.focus.11.4.525

Beukelman, D. R., & Light, J. (2020). Augmentative and alterna- tive communication for children and adults. Brookes.

Beukelman, D. R., & Mirenda, P. (2005). Augmentative and alter- native communication. Brookes.

Biggs, E. E., Carter, E. W., & Gilson, C. B. (2019). A scoping review of the involvement of children’s communication partners in aided augmentative and alternative communica- tion modeling interventions. American Journal of Speech- Language Pathology, 28(2), 743–758. https://doi.org/10. 1044/2018_AJSLP-18-0024

Borenstein, M., & Hedges, L. V. (2019). Effect sizes for meta- analysis. In H. Cooper, L. V. Hedges, & J. C. Valentine (Eds.), The handbook of research synthesis and meta-analysis (3rd ed., pp. 207–242). Russell Sage Foundation. https://doi. org/10.7758/9781610448864.14

Brady, N. C., Thiemann-Bourque, K., Fleming, K., & Matthews, K. (2013). Predicting language outcomes for children learning augmentative and alternative communication: Child and environmental factors. Journal of Speech, Language, and Hearing Research, 56(5), 1595–1612. https://doi.org/10.1044/ 1092-4388(2013/12-0102)

Bruinsma, Y. E., Minjarez, M. B., Schreibman, L., & Stahmer, A. C. (2020). Naturalistic developmental behavioral interven- tions for autism spectrum disorder. Brookes.

Centers for Disease Control and Prevention. (n.d.). Data and sta- tistics on autism spectrum disorder. Retrieved April 27, 2023, from https://www.cdc.gov/ncbddd/autism/data.html

Cook, B., Buysse, V., Klingner, J., Landrum, T., McWilliam, R., Tankersley, M., & Test, D. (2014). Council for Exceptional Children: Standards for evidence-based practices in special

education. Teaching Exceptional Children, 46(6), 206–212. https://doi.org/10.1177/0040059914531389

Dawson, G., Rogers, S., Munson, J., Smith, M., Winter, J., Greenson, J., Donaldson, A., & Varley, J. (2010). Random- ized, controlled trial of an intervention for toddlers with autism: The Early Start Denver Model. Pediatrics, 125(1), e17–e23. https://doi.org/10.1542/peds.2009-0958

Flippin, M., Reszka, S., & Watson, L. R. (2010). Effectiveness of the Picture Exchange Communication System (PECS) on communication and speech for children with autism spectrum disorders: A meta-analysis. American Journal of Speech-Language Pathology, 19(2), 178–195. https://doi.org/ 10.1044/1058-0360(2010/09-0022)

Gaias, L. M., Duong, M. T., Pullmann, M. D., Brewer, S. K., Smilansky, M., Halbert, M., Carey, C. M., & Jones, J. (2020). Race and ethnicity in educational intervention research: A systematic review and recommendations for sam- pling, reporting, and analysis. Educational Research Review, 31, 100356. https://doi.org/10.1016/j.edurev.2020.100356

Ganz, J. B., & Ayres, K. M. (2018). Methodological standards in single-case experimental design: Raising the bar. Research in Developmental Disabilities, 79, 3–9. https://doi.org/10.1016/j. ridd.2018.03.003

Ganz, J. B., Earles-Vollrath, T. L., Mason, R. A., Rispoli, M. J., Heath, A. K., & Parker, R. I. (2011). An aggregate study of single-case research involving aided AAC: Participant charac- teristics of individuals with autism spectrum disorders. Research in Autism Spectrum Disorders, 5(4), 1500–1509. https://doi.org/10.1016/j.rasd.2011.02.011

Ganz, J. B., Mason, R. A., Goodwyn, F. D., Boles, M. B., Heath, A. K., & Davis, J. L. (2014). Interaction of participant charac- teristics and type of AAC with individuals with ASD: A meta-analysis. American Journal on Intellectual and Develop- mental Disabilities, 119(6), 516–535. https://doi.org/10.1352/ 1944-7558-119.6.516

Ganz, J. B., Morin, K. L., Foster, M. J., Vannest, K. J., Genç Tosun, D., Gregori, E. V., & Gerow, S. L. (2017). High-tech- nology augmentative and alternative communication for indi- viduals with intellectual and developmental disabilities and complex communication needs: A meta-analysis. Augmenta- tive and Alternative Communication, 33(4), 224–238. https:// doi.org/10.1080/07434618.2017.1373855

Ganz, J. B., Pustejovsky, J., Reichle, J., Vannest, K., Pierson, L. M., Wattanawongwan, S., Chen, M., Foster, M., Fuller, M., Haas, A., Hamilton, B., Sallese, M., Smith, S. D., & Yllades, V. (2020). Methodology: Systematic review and meta- analysis of the AAC literature for people with autism spectrum disorder or intellectual disabilities who have complex commu- nication needs through 2020. https://hdl.handle.net/1969.1/ 189532

Ganz, J. B., Pustejovsky, J. E., Reichle, J., Vannest, K., Foster, M., Pierson, L. M., Wattanawongwan, S., Bernal, A., Chen, M., Haas, A., Sallese, M. R., Skov, R., & Smith, S. D. (2022). Participant characteristics predicting communication out- comes in AAC implementation for individuals with ASD and IDD: A systematic review and meta-analysis. Augmenta- tive and Alternative Communication. Advance online publi- cation. https://doi.org/10.1080/07434618.2022.2116355

Ganz, J. B., Pustejovsky, J. E., Reichle, J., Vannest, K., Foster, M., Pierson, L. M., Wattanawongwan, S., Fuller, M. C., Haas, A., Sallese, M. R., Smith, S. D., & Yllades, V. (2023). A case for increased rigor in AAC research: A methodological quality review. Education and Training in Autism and Develop- mental Disabilities, 58(1), 3–21.

1752 American Journal of Speech-Language Pathology Vol. 32 1734–1757 July 2023

Gevarter, C., O’Reilly, M. F., Rojeski, L., Sammarco, N., Lang, R., Lancioni, G. E., & Sigafoos, J. (2013). Comparisons of inter- vention components within augmentative and alternative commu- nication systems for individuals with developmental disabilities: A review of the literature. Research in Developmental Disabilities, 34(12), 4404–4414. https://doi.org/10.1016/j.ridd.2013.09.018

Gevarter, C., & Zamora, C. (2018). Naturalistic speech- generating device interventions for children with complex communication needs: A systematic review of single-subject studies. American Journal of Speech-Language Pathology, 27(3), 1073–1090. https://doi.org/10.1044/2018_AJSLP-17-0128

Harris, M. D., & Reichle, J. (2004). The impact of aided lan- guage stimulation on symbol comprehension and production in children with moderate cognitive disabilities. American Journal of Speech-Language Pathology, 13(2), 155–167. https://doi.org/10.1044/1058-0360(2004/016)

Hedges, L. V. (1981). Distribution theory for Glass’s estimator of effect size and related estimators. Journal of Educational Statis- tics, 6(2), 107–128. https://doi.org/10.3102/10769986006002107

Holyfield, C., Drager, K. D., Kremkow, J. M., & Light, J. (2017). Systematic review of AAC intervention research for adolescents and adults with autism spectrum disorder. Aug- mentative and Alternative Communication, 33(4), 201–212. https://doi.org/10.1080/07434618.2017.1370495

Hong, E. R., Ganz, J. B., Neely, L., Boles, M., Gerow, S., & Davis, J. L. (2016). A meta-analytic review of family- implemented social and communication interventions for indi- viduals with developmental disabilities. Review Journal of Autism and Developmental Disorders, 3(2), 125–136. https:// doi.org/10.1007/s40489-016-0071-3

Horner, R. H., Carr, E. G., Halle, J., McGee, G., Odom, S., & Wolery, M. (2005). The use of single-subject research to identify evidence-based practice in special education. Exceptional Children, 71(2), 165–179. https://doi.org/10. 1177/001440290507100203

Horner, R. H., Eberhard, J. M., & Sheehan, M. R. (1986). Teach- ing generalized table bussing: The importance of negative teaching examples. Behavior Modification, 10(4), 457–471. https://doi.org/10.1177/01454455860104005

Hustad, K. C., & Miles, L. K. (2010). Alignment between aug- mentative and alternative communication needs and school- based speech-language services provided to young children with cerebral palsy. Early Childhood Services, 4(3), 129–140. https://pubmed.ncbi.nlm.nih.gov/22194705/

Johnston, S. S., Reichle, J., Feeley, K. M., & Jones, E. A. (2012). AAC strategies for individuals with moderate to severe disabil- ities. Brookes.

Kerkhoff, A. D., Farrand, E., Marquez, C., Cattamanchi, A., & Handley, M. A. (2022). Addressing health disparities through implementation science—A need to integrate an equity lens from the outset. Implementation Science, 17(1), 13. https://doi. org/10.1186/s13012-022-01189-5

Koegel, R. L., & Koegel, L. K. (2019). Pivotal response treatment for autism spectrum disorders. Brookes.

Kratochwill, T. R., Horner, R. H., Levin, J. R., Machalicek, W., Ferron, J., & Johnson, A. (2021). Single-case design standards: An update and proposed upgrades. Journal of School Psychol- ogy, 89, 91–105. https://doi.org/10.1016/j.jsp.2021.10.006

Leaf, R. B., & McEachin, J. (1999). A work in progress: Behavior management strategies and a curriculum for intensive behav- ioral treatment of autism. DRL Books.

Logan, K., Iacono, T., & Trembath, D. (2017). A systematic review of research into aided AAC to increase social- communication functions in children with autism spectrum

disorder. Augmentative and Alternative Communication, 33(1), 51–64. https://doi.org/10.1080/07434618.2016.1267795

Maenner, M. J., Shaw, K. A., Baio, J., Washington, A., Patrick, M., DiRienzo, M., Christensen, D. L., Wiggins, L. D., Pettygrove, S., Andrews, J. G., Lopez, M., Hudson, A., Baroud, T., Schwenk, Y., White, T., Rosenberg, C. R., Lee, L.-C., Harrington, R. A., Huston, M., . . . Dietz, P. M. (2020). Prevalence of autism spectrum disorder among children aged 8 years—Autism and developmental disabilities monitoring network, 11 sites, United States, 2016. MMWR Surveillance Summaries, 69(4), 1–12. https://doi.org/10.15585/mmwr.ss6904a1

Magaña, S., Lopez, K., Aguinaga, A., & Morton, H. (2013). Access to diagnosis and treatment services among Latino chil- dren with autism spectrum disorders. Intellectual and Develop- mental Disabilities, 51(3), 141–153. https://doi.org/10.1352/ 1934-9556-51.3.141

McGee, M. (2005). Self-help, Inc.: Makeover culture in American life. Oxford University Press on Demand.

McLay, L., van der Meer, L., Schäfer, M. C. M., Couper, L., McKenzie, E., O’Reilly, M. F., Lancioni, G. E., Marschik, P. B., Green, V. A., Sigafoos, J., & Sutherland, D. (2015). Com- paring acquisition, generalization, maintenance, and preference across three AAC options in four children with autism spectrum disorder. Journal of Developmental and Physical Disabilities, 27(3), 323–339. https://doi.org/10.1007/s10882-014-9417-x

McNulty, M., Smith, J. D., Villamar, J., Burnett-Zeigler, I., Vermeer, W., Benbow, N., Gallo, C., Wilensky, U., Hjorth, A., Mustanski, B., Schneider, J., & Brown, C. H. (2019). Imple- mentation research methodologies for achieving scientific equity and health equity. Ethnicity and Disease, 29(Suppl. 1), 83–92. https://doi.org/10.18865/ed.29.S1.83

Millar, D. C., Light, J. C., & Schlosser, R. W. (2006). The impact of augmentative and alternative communication inter- vention on the speech production of individuals with develop- mental disabilities: A research review. Journal of Speech, Lan- guage, and Hearing Research, 49(2), 248–264. https://doi.org/ 10.1044/1092-4388(2006/021)

Morin, K. L., Ganz, J. B., Gregori, E. V., Foster, M. J., Gerow, S. L., Genç-Tosun, D., & Hong, E. R. (2018). A systematic quality review of high-tech AAC interventions as an evidence- based practice. Augmentative and Alternative Communication, 34(2), 104–117. https://doi.org/10.1080/07434618.2018.1458900

Neely, L. C., Ganz, J. B., Davis, J. L., Boles, M. B., Hong, E. R., Ninci, J., & Gilliland, W. D. (2016). Generalization and maintenance of functional living skills for individuals with autism spectrum disorder: A review and meta-analysis. Review Journal of Autism and Developmental Disorders, 3(1), 37–47. https://doi.org/10.1007/s40489-015-0064-7

O’Neill, T., Light, J., & Pope, L. (2018). Effects of interventions that include aided augmentative and alternative communica- tion input on the communication of individuals with complex communication needs: A meta-analysis. Journal of Speech, Language, and Hearing Research, 61(7), 1743–1765. https:// doi.org/10.1044/2018_JSLHR-L-17-0132

Ouzzani, M., Hammady, H., Fedorowicz, Z., & Elmagarmid, A. (2016). Rayyan—A web and mobile app for systematic reviews. Systematic Reviews, 5(1), 210. https://doi.org/10.1186/ s13643-016-0384-4

Page, M. J., McKenzie, J. E., Bossuyt, P. M., Boutron, I., Hoffmann, T. C., Mulrow, C. D., Shamseer, L., Tetzlaff, J. M., Aki, E. A., Brennan, S. E., Chou, R., Glanville, J., Grimshaw, J. M., Hróbjartsson, A., Lalu, M. M., Li, T., Loder, E. W., Mayo-Wilson, E., McDonald, S., . . . Moher, D. (2021). The PRISMA 2020 statement: An updated guideline

Reichle et al.: Systematic Review of Variable Instruction in AAC 1753

• • •

for reporting systematic reviews. BMJ, 372, 71. https://doi. org/10.1136/bmj.n71

R Core Team. (2022). R: A language and environment for statisti- cal computing. R Foundation for Statistical Computing. https://www.R-project.org/

Reichle, J., O’Neill, R. E., & Johnston, S. S. (2021). Advances in AAC intervention: Some contributions related to applied behavior analysis. Augmentative and Alternative Communication, 37(3), 206–216. https://doi.org/10.1080/07434618.2021.1962405

Reichow, B., Volkmar, F. R., & Cicchetti, D. V. (2008). Develop- ment of the evaluative method for evaluating and determining evidence-based practices in autism. Journal of Autism and Developmental Disorders, 38(7), 1311–1319. https://doi.org/10. 1007/s10803-007-0517-7

Schlosser, R. W., & Koul, R. K. (2015). Speech output technologies in interventions for individuals with autism spectrum disorders: A scoping review. Augmentative and Alternative Communication, 31(4), 285–309. https://doi.org/10.3109/07434618.2015.1063689

Schreibman, L., Dawson, G., Stahmer, A. C., Landa, R., Rogers, S. J., McGee, G. G., Kasari, C., Ingersoll, B., Kaiser, A. P., Bruinsma, Y., McNerney, E., Wetherby, A., & Halladay, A. (2015). Naturalistic developmental behavioral interventions: Empirically validated treatments for autism spectrum disor- der. Journal of Autism and Developmental Disorders, 45(8), 2411–2428. https://doi.org/10.1007/s10803-015-2407-8

Shire, S. Y., & Jones, N. (2015). Communication partners sup- porting children with complex communication needs who use AAC. Communication Disorders Quarterly, 37(1), 3–15. https://doi.org/10.1177/1525740114558254

Sievers, S. B., Trembath, D., & Westerveld, M. (2018). A system- atic review of predictors, moderators, and mediators of aug- mentative and alternative communication (AAC) outcomes for children with autism spectrum disorder. Augmentative and Alternative Communication, 34(3), 219–229. https://doi.org/10. 1080/07434618.2018.1462849

Steinbrenner, J. R., McIntyre, N., Rentschler, L. F., Pearson, J. N., Luelmo, P., Jaramillo, M. E., Boyd, B. A., Wong, C., Nowell, S. W., Odom, S. L., & Hume, K. A. (2022). Patterns in reporting and participant inclusion related to race and eth- nicity in autism intervention literature: Data from a large- scale systematic review of evidence-based practices. Autism, 26(8), 2026–2040. https://doi.org/10.1177/13623613211072593

Stokes, T. F., & Baer, D. M. (1977). An implicit technology of generalization. Journal of Applied Behavior Analysis, 10(2), 349–367. https://doi.org/10.1901/jaba.1977.10-349

Tager-Flusberg, H., & Kasari, C. (2013). Minimally verbal school-aged children with autism spectrum disorder: The neglected end of the spectrum. Autism Research, 6(6), 468– 478. https://doi.org/10.1002/aur.1329

Therrien, M. C., & Light, J. (2016). Using the iPad to facilitate interaction between preschool children who use AAC and their peers. Augmentative and Alternative Communication, 32(3), 163–174. https://doi.org/10.1080/07434618.2016.1205133

Therrien, M. C., Light, J., & Pope, L. (2016). Systematic review of the effects of interventions to promote peer inter- actions for children who use aided AAC. Augmentative and Alternative Communication, 32(2), 81–93. https://doi.org/10. 3109/07434618.2016.1146331

van der Meer, L., Didden, R., Sutherland, D., O’Reilly, M. F., Lancioni, G. E., & Sigafoos, J. (2012). Comparing three aug- mentative and alternative communication modes for children with developmental disabilities. Journal of Developmental and Physical Disabilities, 24(5), 451–468. https://doi.org/10.1007/ s10882-012-9283-3

van der Meer, L., Sutherland, D., O’Reilly, M. F., Lancioni, G. E., & Sigafoos, J. (2012). A further comparison of manual signing, picture exchange, and speech-generating devices as communication modes for children with autism spectrum dis- orders. Research in Autism Spectrum Disorders, 6(4), 1247– 1257. https://doi.org/10.1016/j.rasd.2012.04.005

Warren, S. F., Fey, M. E., & Yoder, P. J. (2007). Differential treatment intensity research: A missing link to creating opti- mally effective communication interventions. Mental Retarda- tion and Developmental Disabilities Research Reviews, 13(1), 70–77. https://doi.org/10.1002/mrdd.20139

Wattanawongwan, S., Ganz, J. B., Hong, E. R., Dunn, C., Yllades, V., Pierson, L. M., & Foster, M. (2022). Interventions for improving social-communication skills for adolescents and adults with ASD: A meta-analysis. Journal of Autism and Developmental Disorders, 1–15. https://doi.org/10.1007/s40489- 021-00300-8

What Works Clearinghouse. (2019). Response to comment from the public on update version 4.1 of the WWC procedures and WWC standards handbook. Institute of Education Sciences. https://ies.ed.gov/ncee/wwc/handbooksWhat

What Works Clearinghouse. (2020). What Works Clearinghouse standards handbook (Version 4.1). U.S. Department of Educa- tion, Institute of Education Sciences, National Center for Education Evaluation and Regional Assistance. https://ies.ed. gov/ncee/wwc/handbooks

Yoder, P. J., & Layton, T. L. (1988). Speech following sign lan- guage training in autistic children with minimal verbal lan- guage. Journal of Autism and Developmental Disorders, 18(2), 217–229. https://doi.org/10.1007/BF02211948

Zablotsky, B., Black, L. I., & Blumberg, S. J. (2017). Estimated prevalence of children with diagnosed developmental disabilities in the United States, 2014–2016. https://www.cdc.gov/nchs/ products/databriefs/db291.htm?fbclid=IwAR0MwhW

1754 American Journal of Speech-Language Pathology Vol. 32 1734–1757 July 2023

Appendix (p. 1 of 3)

Definitions of Participant Characteristics, Instructional Characteristics, and Communication Outcomes

Participant Characteristics

Communication Modes Prior to Intervention (i.e., natural gesture, manual sign, low-tech aided system, mid- to high-tech speech-generating device, vocal, verbal). The medium through which communicative behavior is expressed. If one DV is categorized that contains multiple modes (e.g., “requesting” that includes verbal and AAC), check all of them.

• Manual sign language: an unaided system that relies on no equipment and instead relies on the learner’s own body to produce communicative acts. An action that includes a specific handshape, location where the sign is produced, and movement pattern that adheres to a sign language or sign system (e.g., ASL, Signed English).

• Verbal: Intelligible words or word approximations

• Natural gesture: Natural gestures that may or may not include a facial expression (e.g., head shake yes or no) but exclude intelligible manual signs or sign approximations. May include the muscles beneath the skin of the face that purposely conveys emotional state to a communicative partner (e.g., frowning, smiling).

• Vocalization: Production of sound, sound combinations that are not intelligible word approximations. Examples: sound—air passing that vibrates the vocal cords that can be heard. This excludes wheezing, snorting, grunting, and whistling.

• Manual sign language: unaided communicative acts that rely on the learner’s own body and that are part of a formal sign language or system (e.g., American Sign Language, Signed English, Seeing Exact English)

• Low-tech aided AAC: An application of a graphic communication mode that does not require electrical power or bat- teries to operate and do not have the capability to produce synthesized or digitized speech (e.g., graphic symbols housed in a wallet, a laminated card housing graphic symbols, a three-ring binder housing graphic symbols).

• Mid- to high-tech aided AAC: An application of an aided communication system. High-tech applications involve the use of electrical or battery power. Typically, they permit the use of digitized (mid-tech). Additionally, high tech allows synthesized text to speech, environmental control and may support email and computer access applications. Addition- ally, they allow unlimited vocabulary, encoding capability, prediction, a variety of access methods, and permit linking any symbol to any other symbol location displayed (e.g., Tobii Dynavox, Prentke-Romich products).

Imitation Use Prior to Intervention

• Vocal/verbal imitation: The replication or partial replication of sound, sound combinations, or spoken word approxi- mations that was produced after a spoken model. To be coded it should be reported as a pre-intervention assessment (or baseline) implemented or a reliable report that occurred prior to the implementation of the independent variable.

• Gestural imitation: The replication or partial replication of an action produced by another after modeled behavior. To be coded, it should be reported as a pre-intervention assessment (or baseline) implemented or a reliable report that occurred prior to the implementation of the independent variable.

Reichle et al.: Systematic Review of Variable Instruction in AAC 1755

Appendix (p. 2 of 3)

Definitions of Participant Characteristics, Instructional Characteristics, and Communication Outcomes

• • •

Instructional Characteristics

Instructional Features

• Graphic prompt: Interventionists provide graphic stimuli (e.g., picture, printed word, product logo) that provide learners with information about how to engage in the target behavior (e.g., task analysis checklist, following a recipe transition picture card).

• Modeling: Modeling of communication used (e.g., verbally demonstrating contextual words, demonstrating use of AAC to make a comment): Interventionist produces a behavior offering an opportunity for the learner to replicate the behav- ior. The dependent measure should specify whether the learner is required to produce an exact or partial replication.

• Physical prompts: Interventionist comes in physical contact with the learner to help them produce a behavior being taught (e.g., hand-over-hand assistance to make an SGD selection, tapping a learner’s hand to cue him to begin to pick up an object).

• Prompt fading: Prompt-fading strategies used in which the delivery of a prompt is delayed affording the learner an increased period of time in which to emit an independent target response (e.g., time delay, least-to-most prompt fad- ing, most-to-least)

• Systematic arrangement: The interventionist systematically arranged the environment (use of communicative tempta- tions, such as enticing, having preferred materials in sight)

• Verbal prompts: Verbal prompts provided to prompt communication: Interventionist produces statements to direct a learner’s behavior. Examples: You might need to try it a different way‚ Write your name. “What do you want?” “Say, ‘____.’” “Let me know if you want anything.” “Tell me if you need anything.” A typical SD required to initiate the task is NOT a verbal prompt (e.g., “point to ___,” “put the ball on ___,” “choose/pick ___”).

• Reinforcement: Reinforcement provided to promote communication (e.g., social praise for speaking, providing items requested).

• Preference assessment: Preference/reinforcer assessment was conducted before or throughout intervention.

Distribution of Communication Outcomes

Communication Modes During Intervention

• Low-tech aided AAC: Aided communication that does not require electrical power or batteries to operate and does not have the capability to produce synthesized/ digitized speech (e.g., graphic symbols housed in a wallet, board, notebook, folder, etc.).

• Mid- to high-tech aided AAC: Relies on graphic symbols displayed in a battery-powered system that produces digi- tized or synthesized speech.

• Manual sign: unaided communicative acts that rely on the learner’s own body and that are part of a formal sign lan- guage or system (e.g., American Sign Language, Signed English, Signing Exact English).

• Natural gestures: head shake yes or no, frowning, smiling, pointing, proffering a cup to have it refilled; leading an indi- vidual to an object/event and more idiosyncratic gestures such as putting fist on one’s nose to communicate (“need a tissue”).

• Verbalization: Intelligible words or easily guessed word approximations

• Vocalization: Production of sound(s), sound combinations that are not intelligible word approximations (excludes wheezing, snorting, and whistling that do not require use of vocal cords)

1756 American Journal of Speech-Language Pathology Vol. 32 1734–1757 July 2023

Appendix (p. 3 of 3)

Definitions of Participant Characteristics, Instructional Characteristics, and Communication Outcomes

Communication Systems

• Comprehension: Deriving meaning (understanding) utterances produced by others using speech, gestures/sign, or graphic symbols (e.g., parent says, “get your shoes,” and child travels to bedroom and retrieves shoes). The participant demonstrates receptive understanding through action in response to the interventionist’s language output.

• Production: The emission of sounds, sound combinations, spoken words, gestures (including facial expression), man- ual signs, graphic symbols (e.g., miniature objects), object remnants (e.g., a button representing a shirt), photos, pic- tures, product logos, printed words, or combinations of the preceding to influence a communicative partner’s behavior

Communication Functions

• Behavioral regulation: Communicative act emitted to obtain or maintain access to an object, activity, or person; or to escape or avoid contact with an object, activity, or person (e.g., requesting a hug, asking to go for ice cream, protest- ing bath time, protesting the offer of a food item, requesting help, requesting a break).

• Joint attention: In a joint attention communicative act emitted to direct a partner’s attention to an object or event external to the communicative partners (e.g., providing requested information that is not in the context of an effort to increase turn taking comment, naming objects in the environment, requesting information). A joint attention act can also spontaneously comment on or name, for example, an object (e.g., “dog”), person (e.g., “dog”), action (communi- cating “run” as a dog runs by the speaker or communicative partner), attribute (while looking at a group of green apples the learner sees one red one and communicates “red”), or adverbial (a dog is moving slowly and learner com- municates “slow”).

• Social interaction: A communicative act emitted to direct attention to oneself. Communicative act emitted to direct a partner’s attention to oneself (e.g., reciting story passages [where the objective is more taking turns than the impor- tance of the information], telling a knock-knock joke, greetings, calling—“hey look at me,” requesting another’s attention).

Reichle et al.: Systematic Review of Variable Instruction in AAC 1757

Copyright of American Journal of Speech-Language Pathology is the property of American Speech-Language-Hearing Association and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use.