Communication interventions on conditional requesting or rejecting skills for individuals with moderate to severe developmental disabilities: a scoping review

Abstract

Objective: To synthesize communication intervention studies that addressed conditional requesting or rejecting skills for individuals with moderate to severe developmental disabilities from 1980 to 2017.

Methods: A systematic search was conducted and the yielded 20 studies were coded from three aspects, including participant characteristics, intervention characteristics, and quality of study parameters.

Results: The identified studies involved 31 participants who ranged from 4 to 45 years of age and had a diagnosis of a developmental disability with moderate to severe disabilities. These individuals with moderate to severe disabilities were able to learn to request or reject conditionally, with maintenance and generalization reported in some studies. In general, the identified studies differed in their specific intervention goals related to conditional requesting or rejecting skills. The methodological adequacy of the studies identified warrants improvement.

Conclusion: There is the need for future research on conditional communication intervention for individuals with developmental disabilities to refine and expand the existing evidence base.

Communication Interventions on Conditional Requesting or Rejecting Skills for Individuals with Moderate to Severe Developmental Disabilities: A Scoping Review Communication development is a common focus in the treatment of individuals with moderate to severe developmental disabilities. Although a number of evidence-based communication interventions have been developed over the years to enable these persons to communicate more effectively, identifying interventions that promote a learner’s conditional use of communication remains limited (Chadsey-Rusch and Halle 1992, Johnston et al. 2012, O’Neill and Reichle 1993). Conditional use of communication refers to ‘the ability to discriminate among the specific stimulus conditions that should and should not result in a communicative utterance’ (Reichle and Johnston 1999, p. 324). That is, individuals need to know when, where, and to whom they should and should not emit certain communicative acts. Conditional communication use is sometimes referred to as conditional communication or discriminated communication (Kuhn et al. 2010). Examples of conditional use of communication abound in everyday life: We greet our close friends in a different manner than we greet strangers; We talk to younger people in a different manner than we talk to older people; We speak at home in a different manner than we speak in a conference. Under all of these scenarios, communicative acts are discriminated by certain antecedent stimuli (e.g. communicative partners, environment) to be more functional.

Conditional communication is not only crucial for typically developing people, but also important for individuals with moderate to severe developmental disabilities (Reichle and Wilkinson 2012). Without the conditional use of communicative behavior the quality of life among individuals with moderate to severe developmental disabilities may be diminished. Existing evidence suggests that persons with significant communicative challenges often have difficulty using newly acquired communicative behavior conditionally (Albin and Horner 1988, Johnston et al. 2012, Reichle et al. 1984). The relatively limited literature addressing conditional communicative use among persons with moderate and severe developmental disabilities suggests a need for further investigation. To more clearly elucidate existing evidence in this area, there is a need to conduct a systematic literature review of extant communication intervention literature addressing the conditional use of communication among persons with moderate and severe developmental disabilities.

Targeting conditional requesting/rejecting in the current review

The current review will focus on studies examining the conditional use of the fundamental communicative production acts of requesting and rejecting. The rationale for this choice is threefold. First, typically, expressive requesting and rejecting are among the earliest communicative functions to emerge for both typically developing children and children with developmental delays/disabilities (Bijou and Baer 1965, Carpenter et al. 1983, Crais et al. 2004). Communicative requesting is defined as behavior that enables the speaker to gain or maintain access to objects, activities, or interactions through the mediation of a listener (Sigafoos and Mirenda 2002), while communicative rejecting is defined as behavior that enables the speaker to escape from or avoid objects, activities, or interactions through the mediation of a listener (Sigafoos et al. 2002). From the standpoint of Skinner’s (1957) verbal behavior, requesting and rejecting can be considered as the two verbal behaviors that constitute mands, which Skinner defined as ‘a verbal operant in which the response is reinforced by a characteristic consequence and is therefore under the functional control of relevant conditions of deprivation or aversive stimulation’ (pp. 35–36). Requesting helps a learner to access positive reinforcement while rejecting helps a learner to access negative reinforcement (i.e. avoidance or removal of aversive or non-preferred stimuli).

Second, among communication intervention programs for individuals with moderate to severe developmental disabilities, requesting and rejecting skills are usually the target functional communication skills during treatment (Reichle et al. 2002, Sigafoos et al. 2004). For example, Light et al. (2002) defined one of the four major communication goals as the regulation of someone else’s behavior so as to obtain desired objects or actions or to reject undesired objects or actions. In particular, teaching a learner with moderate to severe developmental disabilities to request or reject using socially acceptable communicative means is often part of a proactive or reactive approach to the treatment of problem behavior (e.g. Carr and Durand 1985, Langdon et al. 2008, Tiger et al. 2008; and a plethora of others).

Finally, along with the emphasis on teaching requesting and rejecting for learners with moderate to severe developmental disabilities there has been increasing attention given to facilitating conditional use of requests and rejections. For instance, among persons who utilize augmentative and alternative communication (AAC) using natural gestures, manual signs, picture-based communication board, and speech-generating devices, teaching the conditional use of communicative modes to request or reject as a function of speaker and context variables has been advocated (Mirenda 1999, Reichle et al. 2003, Sigafoos and Drasgow 2001). For example, a sign which is quick to produce might be appropriate when one wants to communicate as fast as possible with someone who signs but a slower to emit graphic symbol may be a better choice with a listener who does not sign.

Also, as functional communication training (FCT) developed as an evidence-based approach to addressing problem behavior for individuals with developmental disabilities since mid-1980s, recently more attention has been paid to the issue of how to make their communicative requesting or rejecting skills learned through FCT more functional and socially appropriate (e.g. Fisher et al. 2000, LeBlanc et al. 2002). For example, a child with developmental disabilities who has learned how to request preferred items through an intervention using a continuous reinforcement schedule in an FCT program may start to initiate requests at an undesirably high rate, which is impractical to maintain due to practical restrictions such as (a) the item may not always be readily available; (b) continuous consumption of the item may not be healthy or may interrupt other important daily living activities; and (c) the high-rate requests may not be age appropriate (Sidener et al. 2006). Similar challenges may occur when a child learned to reject the consumption of some food (e.g. vegetables) or access to certain activities (e.g. house-keeping chores) through FCT but started to initiate the rejections too frequently. Under scenarios like these, it becomes helpful if not urgent to teach the child how to request or reject conditionally.

In summary, given the importance of requesting and rejecting skills for individuals with moderate to severe developmental disabilities, as well as an increasing call for more attention to the issue of conditional use of requesting or rejecting skills, the current review will focus on synthesizing communication intervention studies that addressed conditional use of requesting or rejecting skills for individuals with moderate to severe developmental disabilities.

The review that follows constitutes a scoping review to evaluate the extent of a body of literature on a topic, identify the potential gaps in the existing research, and ensure that further research in that area would be beneficial (Landa et al. 2011). Specifically, this review was designed to:

• (a)profile the range of participants identified in the current study through briefly describing some major participant characteristics, including gender, age, intelligence level, disability diagnosis, expressive communication levels, imitation ability, and the status of problem behavior.

These participant characteristics were selected because all of them have been frequently extracted to represent the profile of participants in other literature reviews on communication interventions (e.g. Mancil 2009, Millar et al. 2006, Walker and Snell 2013).

• (b)identify the characteristics related to the interventions on conditional requesting and/or requesting skills applied to learners with moderate to severe developmental disabilities, including (1) experimental design, implementer, setting, antecedent variable(s) manipulated to make the requesting/rejecting skill conditional, positive and negative teaching conditions, target requesting/rejecting skill(s) under positive teaching condition(s), and target skill(s) under negative teaching condition(s).

Since conditional communication requires at least two different communicative conditions, it is reasonable to investigate how the antecedent variable(s) were manipulated to create at least two different conditions, what the different conditions were and the respective target skills for each condition.

• (c)evaluate the quality of conditional requesting or rejecting research studies that were included in the current review. Descriptive summaries of study quality allow for the evaluation of study credibility and research adequacies. Five study quality characteristics were evaluated in the current study, each recognized as a quality indicator for single-case research (Horner et al. 2005), including: (1) reliability; (2) intervention fidelity; (3) social validity; (4) maintenance; and (5) generalization.

Prior to a description of methodology used in the review, the major empirical foundation for conditional use of communication (i.e. stimulus control) will be briefly discussed in the next section, so as to provide some theoretical backgrounds for the current review.

Empirical foundation for conditional communication: stimulus control

To a certain extent, conditional use of communication can be viewed on a continuum of antecedent stimuli conditions that set the occasion for the communicative act to occur and not to occur. Therefore, conditional communication production (e.g. conditional requesting or rejecting) can be better understood from a stimulus control perspective. Stimulus control refers to a functional relationship between presentation of an antecedent stimulus and change in the probability of a response (i.e. antecedent stimulus → response [R]; Terrace 1966). Typically, there are two types of antecedent stimulus, that is, establishing operation (EO) and discriminative stimulus (S^D). Specifically, the concept of establishing operation was initially termed by Keller and Schoenfeld (1950) but later refined by Michael (1982, 1993, 2000). Michael (1982, 1993, 2000) defined it as any stimulus, condition, or event that (a) momentarily alters the value of some stimulus as a reinforcer and (b) evokes all responses that have produced that reinforcer in the past. The term establishing operation was later replaced with motivating operation (MO) to encompass both the establishing and abolishing operations so as to more accurately represent the bidirectional effects of establishing and abolishing the value of reinforcement and the corresponding evoking or abating of the relevant behavior (Laraway et al. 2003). In contrast with EO or MO, discriminative stimulus (S^D) signals the (un)availability of reinforcement (e.g. Tiger et al. 2006). In particular, discriminative stimulus correlated with the availability of reinforcement is designated as S+,while discriminative stimulus correlated with the unavailability of reinforcement is designated as S-. Although motivating operation and discriminative stimulus are different in nature, both of them are able to exert stimulus control over responses.

Therefore, from the perspective of stimulus control, a conditional communication production involves at least one positive condition (i.e. the presence of certain MO and/or S^D) under which a communicative act should/could occur, and at least one negative condition (i.e. the absence of certain MO and/or S^D) under which the communicative act should/could not occur or an alternative act should/could occur. For instance, as described in the study by Reichle and Johnston (1999), a participant learned to request assistance in getting preferred snacks out of reach and learned to get the snacks by himself/herself when the preferred snacks were placed within his/her reach. In this scenario, the distance between the preferred snacks and the participant functioned as SDs and exerted stimulus control over the participant’s requesting behavior. On the basis of the empirical foundation, a systematic search and synthesis of communication intervention literature related to conditional requesting or rejecting for individuals with moderate to severe developmental disabilities was presented in the sections to follow.

Method

Search procedures

Two searches were completed — the original search in July 2013 and another in July 2017 to broaden the scope of analysis through the inclusion of more recent studies. For both searches, a systematic three-step search was implemented to identify articles related to conditional requesting or rejecting. First, systematic searches were conducted in four electronic databases including Academic Search Premier, Education Resources Information Center (ERIC), PsychINFO, and Education Full Text. The search was limited to English language peer-reviewed studies and studies published between January 1980 and July 2013 for the first search, and between July 2013 and July 2017 for the second search. The year range from 1980 to 2017 was selected because few experimental studies on conditional communication use were reported prior to 1980s (Reichle et al. 1991). The key words used in these searches were listed in Table 1. After excluding the duplicates and irrelevant articles, the search resulted in a sample of 587 studies for the first search and 272 studies for the second search to which inclusion and exclusion criteria (further described below) were applied. Of the 859 studies, 15 studies (13 studies from the first search and 2 studies from the second search) were included in the review. Next, hand or electronic searches of the tables of contents of six journals (i.e. Journal of Applied Behavior Analysis; The Analysis of Verbal Behavior; Journal of Developmental and Physical Disabilities; Research in Developmental Disabilities; Journal of the Association for Persons with Severe Handicaps; Language, Speech, and Hearing Services in Schools) between 1980 and 2017 (with issues of the journals published during these years) were conducted using the same key words listed in Table 1. These six journals were selected because they produced at least two included articles that were identified through the initial electronic search. Last, ancestral searches of the references cited in all identified articles were conducted. The three-step search process was designed to obtain a maximum number of relevant studies for the review. Through the three-step search, a total of 20 articles (19 from the first search and 1 from the second search) meeting the inclusion criteria were identified for review.

Table 1. Search terms used in electronic search.

Search terms

‘conditional use’ AND ‘communication’; ‘discriminative use’ AND ‘communication’; ‘conditional communication’; ‘discriminated communication’; ‘conditional manding’; ‘discriminated manding’; ‘conditional requesting’; ‘discriminated requesting’; ‘establishing operation’ AND ‘request’; ‘establishing operation’ AND ‘mand’; ‘motivating operation’ AND ‘request’; ‘motivating operation’ AND ‘mand’; ‘discriminative stimulus’ AND ‘request’; ‘discriminative stimulus’ AND ‘mand’; ‘conditional discrimination’ AND ‘developmental disability’; ‘matching to sample’ AND ‘developmental disability’; ‘stimulus control’ AND ‘request’; ‘stimulus control’ AND ‘mand’; ‘stimulus control’ AND ‘developmental disability’; ‘mand’ AND ‘developmental disability’; ‘request’ AND ‘developmental disability’; ‘discriminated functional communication training’; ‘mand training’; ‘rejecting’ and ‘developmental disability’; ‘teaching rejecting’; ‘protesting’ and ‘developmental disability’; ‘teaching protesting’; ‘yes-no responding’; ‘interrupted behavior chain’; ‘wrong-item format’; ‘question-asking’ AND ‘developmental disability’; ‘multiple schedule’; ‘schedule-correlated stimuli’; ‘general-case instruction’; ‘multiple-exemplar training’

Inclusion and exclusion criteria

Studies only were included that: (a) employed an experimental single-case research design (e.g. reversal, multiple baseline or multiple probe, alternating treatment, and combined design), a quasi-experimental/pre-experimental single-case research design (e.g. AB design), or an experimental or quasi-experimental group design. A preliminary search of literature yielded few group design studies related to interventions on conditional requesting or rejecting, so the current study only focused on single-case research design studies; (b) were published between January 1980 and July 2017; (c) served at least one learner with a developmental disability who had moderate to severe intellectual delay/disability; (d) addressed the conditional requesting or rejecting skill(s) in the intervention, that is, involving at least one positive teaching condition (i.e. a teaching condition in which a target requesting or rejecting behavior should/could occur) and one negative teaching condition (i.e. a teaching condition in which the target requesting or rejecting behavior should/could not occur or a target alternative appropriate behavior should/could occur); and (e) pointed out the antecedent variables (i.e. motivating operations and/or discriminative stimulus) that were manipulated in the intervention to make the requesting or rejecting skill(s) conditional; and (f) reported outcome measures that can reflect the learner’s communicative performance on both the positive and negative teaching conditions (either in a separate or combined manner). Studies were excluded if a description of participants’ characteristics (e.g. age, gender, intelligence level) was absent. Also, during the coding phase when the full article was examined, if one or more participants within any given study did not meet the inclusion criteria, those participants’ data were excluded from participant-level descriptive analyses.

Reliability of search procedures

To ensure the search accuracy, a second independent observer conducted an independent search in two randomly selected databases using the same keywords, which yielded 100% inter-observer agreement (IOA, agreements divided by agreements plus disagreements multiplying by 100) for both the search conducted in 2013 and the one conducted in 2017. Subsequently, two journals reviewed were randomly selected, and the independent observer repeated the application of the search criteria. This resulted in 96 and 100% IOA computed as agreements divided by agreements plus disagreements 100 for the two searches, respectively. Next, among the initial pool of 587 articles from electronic searches, a random sample of 20% of the articles were identified for two observers to determine whether each of them should be included or not based on the inclusion/exclusion criteria. This yielded an IOA of 98% computed as agreements divided by agreements plus disagreements multiplied by 100. The same procedures were conducted for the pool of 272 articles from electronic searches conducted in July 2017, which yielded an IOA of 96%. Finally, the independent observer implemented an ancestral search of 20% of the articles that met the selection criteria. Using the same item-by-item agreement formula, an IOA of 100% resulted for both searches. Throughout the reliability process, disagreements were resolved through discussion and a final consensus was reached.

Coding procedures

Coding instrument

An Excel spreadsheet was created to code the identified studies. The coding instrument contained a total of 21 items organized under three categories that included: (a) Participant characteristics (i.e. gender, age, intelligence, disability diagnosis, expressive communication level, imitation ability, and the status of problem behavior); (b) Intervention characteristics (i.e. experimental design, implementer, setting, antecedent variable[s] manipulated to make the requesting/rejecting skill conditional, positive teaching condition[s], negative teaching condition[s], target requesting/rejecting skill[s] under positive teaching condition[s], and target skill[s] under negative teaching condition[s]); and (c) Quality of study parameters (i.e. reliability, fidelity, social validity, maintenance, generalization [type and its measurement]). A list of the operational definitions for the coding items was presented in Table 2.

Table 2. Operational definitions of coding items and corresponding coding schema.

Coding item	Definition
Participant characteristics
Gender	Coded as 1 = Male; 2 = Female, and 3 = Unspecified
Age	Defined as the chronological age reported for the individual participant
	Coded in years to one decimal
Intelligence	Defined as the degree of intellectual disability reported for the individual participant (Note: The higher degree of intellectual disability is coded if a range of degrees is provided). For example, ‘severe’ intellectual disability is coded if ‘moderate to severe’ intellectual disability is reported
	I.Q. ranges associated with terms included: 1= Moderate (35 ≤ IQ < 50, or 40 ≤ IQ < 55); 2 = Severe (20 ≤ IQ < 35, or 25 ≤ IQ < 40) (American Psychiatric Association 2013)
Disability Diagnosis	Defined as the handicapping conditions reported for individual participant
	Coded as described in the article for future categorization in data analysis
Expressive Communication Level	Defined as the participant’s ability to express verbal and/or nonverbal communicative acts
	Coded as described in the article for future categorization in data analysis
Imitation Ability	Defined as the participant’s ability to reproduce different verbal and/or nonverbal communication behaviors
	Coded as: 1 = Vocal (sound approximation that was not an intelligible word); 2 = Gestural (natural gesture, manual sign); 3 = Verbal (spoken word or combination of spoken words); 4 = Unspecified
Status of Problem Behavior	Defined as the status of the presence or absence of challenging behavior (e.g. self injury, aggression, property destruction, off-task behavior, tantrum, repetitive behaviors) for the individual participant
	Codes as: 1 = The presence of problem behavior reported (specify the function[s] of problem behavior if identified or reported); 2 = The absence of problem behavior reported; 3 = Unspecified
Intervention characteristics
Experimental Design	Defined as the specific single-case research design utilized for the investigation of conditional requesting or rejecting
	Coded as: 1 = Quasi-experimental single case design (specify); 2 = Reversal design (specify); 3 = Alternating treatment design; 4 = Multiple baseline/multiple probe design (specify); 5 = Combined design (specify the designs that are combined); 6 = Other (specify)
Implementer	Defined as the person(s) who implemented the intervention of conditional requesting/rejecting
	1 = Professional researchers (specify); 2 = Parents; 3 = Teachers; 4 = Therapists; 5 = Paraprofessionals; 6 = Siblings; 7 = Unspecified
Setting	Defined as the setting where the intervention was implemented
	Coded as it is described in the study. If not reported, codes it as ‘9999.’
Antecedent Variables Manipulated	Defined as the specific antecedent variables that were manipulated to make the communicative requesting or rejecting skill(s) conditional in the intervention
	Coded as: 1 = Motivating operations (MOs; defined as any environmental variable that alters the value of the reinforcer and evokes the responses that have produced that reinforcer in the past [Michael 1982, 1993, 2000]); 2 = Discriminative stimuli (SDs; defined as any environmental variable that signals the [un]availability of the reinforcer); 3 = Both MOs and SDs (Note: This is coded when the impacts of MOs and SDs are not explicitly separated and both of the impacts are possible.)
Positive Teaching Condition(s)	Defined as the condition(s) in which a target requesting/rejecting behavior should/could occur
	Coded as it is described
Negative Teaching Condition(s)	Defined as the condition in which a target requesting/rejecting behavior should/could not occur or an alternative appropriate behavior should/could occur
	Coded as it is described
Target Skill(s) Under the Positive Teaching Condition(s)	Defined as the specific functional type[s] of communicative requesting or rejecting behavior under the condition in which the target behavior should/could occur
	Coded as: 1= requesting items; 2 = requesting activities; 3 = requesting attention; 4 = requesting break; 5 = requesting information; 6 = requesting assistance; 7 = requesting clarification; 8 = requesting work check; 9 = requesting permission; 10 = requesting alternative; 11 = rejecting (specify); 12 = Other (specify)
Target Skill(s) Under the Negative Teaching Condition(s)	Defined as the specific target behavior that should or could occur under the negative condition
	Coded as: 1 = Silence (i.e. the absence of the target skill[s] under the positive condition); 2 = Alternative appropriate behavior (enter the target behavior specified in the article). Note: Codes of ‘1’ and ‘2’ are mutually exclusive
Quality of study parameters
Reliability	Defined as the measurement of acceptable interobserver agreement (set at 80% item by item or a Kappa of at least 0.60) associated with each dependent variable
	Coded as: 1 = IOA > 80%; 2 = IOA < = 80%; 3 = Kappa no less than 60%; 4 = Kappa less than 60%; 5 = Unspecified
Fidelity	Defined as the measurement of treatment integrity for each independent variable
	Coded as: 1 = Interviews (specify the results); 2 = Rating scales (specify the results); 3 = Direct observation (specify the results); 4 = Unspecified
Social Validity	Defined as the measurement of the social significance of the goals (i.e. whether the behavioral goals are what the society/consumers wants), the social appropriateness of the procedures (i.e. whether the consumers consider the procedures acceptable), and the social importance of the effects (i.e. whether consumers are satisfied with the results) (Wolf 1978)
	Coded as: 1 = Reported (specify); 2 = Unreported
Maintenance	Defined as the examination of skill presence after the intervention is ended
	1 = Examined (specify the interval between the time when the intervention ended and the maintenance probe[s]); 2 = Unexamined
Generalization_ Type	Defined as the specific parameters on which the target skill[s] of the positive and/or negative conditions were measured in the absence of direct intervention
	Coded as: 1 = Settings (specify); 2 = Interventionists (specify); 3 = Tasks/materials/activities (specify); 4 = Other parameters as specified directly in the article; 5 = No report of any parameter of generalization
Generalization_ Measurement	Defined as the way generalization probes were assessed during baseline and intervention
	Coded as: 1 = All generalization probes were conducted during baseline and intervention TP and TN separately; 2 = Some generalization probes were measured during baseline and intervention for TP and TN separately (specify the probes ); 3 = All generalization probes were measured only during intervention for TP and TN separately; 4 = All generalization probes were measured during baseline and intervention for TP and TN collectively; 5 = Some generalization probes were measured during baseline and intervention for TP and TN collectively (specify the probes); 6 = All generalization probes were measured only during intervention for TP and TN collectively; 7 = Other (specify); 8 = N/A (i.e. no generalization probes)

It is notable here that in terms of the coding for positive and negative teaching conditions, if only two teaching conditions were involved and the two teaching conditions consisted of one requesting teaching condition (i.e. the condition in which certain requesting skill should/could occur) and one rejecting condition (i.e. the condition in which certain rejecting skill should/could occur), the requesting teaching condition was coded as the positive teaching condition and the rejecting teaching condition was coded as the negative one. If more than two teaching conditions were involved, which consisted of only one requesting teaching condition or only one rejecting teaching condition, the requesting or rejecting teaching condition was coded as the positive teaching condition, while the rest were coded as the negative one(s). If more than two teaching conditions were involved, which consisted of at least one requesting teaching condition and at least one rejecting teaching condition, then all the requesting teaching condition(s) were coded as the positive teaching condition(s), while all the rejecting teaching condition(s) and other teaching condition(s) were coded as the negative ones, with one exception as follows. That is, when there was at least one positive and one negative teaching condition for requesting and/or rejecting, the positive and negative teaching condition(s) were coded for requesting and/or rejecting, respectively, with the rest teaching condition(s) (if any) also coded as negative one(s).

Coding reliability and analysis

From the 20 qualifying studies, four (20%) were randomly selected to assess inter-observer agreement (IOA). Each coder independently coded the studies using the coding instrument. IOA was calculated as agreements divided by agreements plus disagreements 100. On average, IOAs for the coding items of participant characteristics, intervention characteristics, and quality of study parameters were 96.4% (range = 85.7–100%), 93.8% (range = 87.5–100%), and 87.5% (range = 83.3–100%). All cases of disagreement were discussed until resolved and the correct code was agreed upon.

After all identified studies were coded, descriptive data analysis which contained a summary of participant characteristics, intervention characteristics, and quality of study parameters was conducted in Excel spreadsheet.

Results

Overall, the 20 identified studies qualitatively suggested that participants with moderate to severe developmental disabilities were able to learn the conditional requesting and/or rejecting skills that researchers set out to teach them, with only one participant in the study by Duker et al. (1994) reported to have limited discriminated responding over the conditional communication intervention. In the sections to follow, results will be described by participant characteristics, intervention characteristics, and quality of study parameters, respectively. Notably, the participant characteristics will only be briefly described with the aim to provide a general profile of the range of participants involved in the current study. The major results for each identified article were displayed in Tables 3, 4, and 5.

Table 3. Summary of intervention characteristics, including number of participants and number who met inclusion criteria, research design, implementer, and setting.

Citation	# partici-pants	# who met criteria	Research design	Implementer	Setting
Anderson et al. (2010)	2	1	Alternating treatment design	Experimenter	In a lab room
Brown et al. (2000)	4	2	Alternating treatment design	Trained graduate students (for 1subject); parents for the other subject	For 1 subject: inpatient unit; For the other: home
Duker et al. (1994)	6	3	Reversal design	Teachers	In a classroom at a residential school
Fisher et al. (1998)	2	1	Reversal design	Therapists	An inpatient unit
Grow et al. (2010)	1	1	Alternating treatment design	Therapists	In the living room of the group home
Horner et al. (1986)	5	3	Split-multiple baseline design	Trained student in special education	In 4 training grocery stores and 1 probe store
Hung (1980)	2	2	Quasi-experimental (AB design)	Teachers	In the class
Kreibich et al. (2015)	1	1	Multiple baseline design across tasks incorporating a changing criterion design	Behavior analysis-trained interventionists	In an individual classroom in the school
Kuhn et al. (2010)	2	1	Multiple baseline design across exemplar pairs	Therapists	In an individual therapy room or in a bedroom on the inpatient unit
Magpusao and Cipani (1989)	1	1	Multiple baseline design across behaviors	Graduate student in SPED or qualified SPED teacher	At the sink area of the classroom
Reichle et al. (1984)	1	1	Multiple probe design across communicative functions	Teachers	In the classroom
Reichle and Brown (1986)	1	1	Quasi-experimental (AB design)	Teachers	In the classroom
Reichle and Johnston (1999)	2	2	Quasi-experimental (AB design)	Teachers	In the classroom during the regularly occurring snack time activity
Reichle et al. (2005)	1	1	Within-subject multiple probe design across task components	Experimenter	In the common area of the residence
Sigafoos et al. (1989)	3	1	Multiple probe design across mand and tact embedded within a multiple-baseline design across object sets	Program staff; undergraduate research assistants	In the kitchen area of a center-based sheltered workshop
Sigafoos et al. (1990)	2	2	ABA design within object sets embedded within a multiple-baseline design across object sets	Interventionist	In the kitchen area of the habilitation center
Sigafoos and Reichle (1992)	4	3	Multiple-probe design across stimulus sets	Senior direct-service staff	At the dining room table of the group home
Sigafoos (1998)	1	1	Combined design (AB, reversal, alternating treatment design)	Research assistant also a qualified SPED teacher	In the classroom of a preschool program
Sigafoos and Roberts-Pennell (1999)	2	2	Multiple baseline design across subjects	Research assistant also a qualified SPED teacher	At a table in the small-group work area in the classroom
Sigafoos and Drasgow (2001)	1	1	Quasi-experimental design (ABCD design) + Alternating treatment design	PhD-level clinician	In a therapy room

Table 4. Summary of intervention characteristics, including antecedent variable(s) manipulated, positive/negative teaching conditions, and target skill(s) for positive/negative conditions.

Citation	Antecedent variable(s) manipulated	Positive teaching condition(s)	Negative teaching condition(s)	TP (target skill[s] for positive condition)	TN (target skill[s] for negative condition)
Anderson et al. (2010)	Discriminative stimuli (SDs)	Presence of communication card without an X over it	Presence of communication card with an X over it	Requesting terminating punishment for stereotypy	Silence
Brown et al. (2000)	Motivating operations (MOs)	EO present	EO absent	Requesting items & attention (sub1)	Requesting ‘work done’
				Requesting ‘work done’ (sub2)	Requesting tangible
Duker et al. (1994)	Motivating operations (MOs)	Conditioned EO present (incomplete items)	Conditioned EO absent	Requesting items or activities	Silence
Fisher et al. (1998)	Discriminative stimuli (SDs)	One drawing indicating the availability of interaction with adults	One drawing indicating the availability of tangible items	Requesting attention	Requesting tangible items
Grow et al. (2010)	Discriminative stimuli (SDs)	Not wearing a black lanyard	Wearing a black lanyard	Requesting attention	Silence
Horner et al. (1986)	Both MOs and SDs	Non-matched items	Matched items	Rejecting the item	Select the item
Hung (1980)	Both MOs and SDs	Presence of preferred items, asking ‘Do you want ##?’	Presence of aversive items, asking ‘Do you want ##?’	Requesting the item	Rejecting the item
Kreibich et al. (2015)	Discriminative stimuli (SDs)	The delay cues were all provided (a full tube)	The delay cues were not yet fully provided (a not yet full tube)	Requesting a break	Engaging in the task
Kuhn et al. (2010)	Discriminative stimuli (SDs)	The therapist is not busy	The therapist is busy	Requesting attention	Silence
Magpusao and Cipani (1989)	Motivating operations (MOs)	One item is missing	No item is missing	Requesting the missing item	Not requesting the present item
Reichle et al. (1984)	Both MOs and SDs	Present preferred items, ask ‘What do you want?’	Present the neutral to undesired item	Requesting the preferred item	Rejecting the neutral to undesired item
			Present the item, ask ‘What is it?’		Labeling the item
Reichle and Brown (1986)	Discriminative stimuli (SDs)	Ask ‘What do you want?’	Ask ‘What is it?’	Requesting the item	Labeling the item
Reichle and Johnston (1999)	Discriminative stimuli (SDs)	Snack items placed close to the teacher	Snack items placed close to the learner	Requesting the item	Take the item
Reichle et al. (2005)	Both MOs and SDs	When the assembly is difficult for the learner	When the assembly is easy for the learner	Requesting assistance	Doing the task independently
Sigafoos et al. (1989)	Both MOs and SDs	Food or beverage beyond reach & utensil out of sight	Holding up one item and asking ‘What is this?’	Requesting the item	Labeling the item
Sigafoos et al. (1990)	Both MOs and SDs	Food or beverage beyond reach & utensil out of sight	Holding up one item and asking ‘What is this?’	Requesting the item	Labeling the item
Sigafoos and Reichle (1992)	Discriminative stimuli (SDs)	Offering one explicit item	Offering one generalized item	Requesting the item using the explicit symbol	Requesting the item using the proper generalized symbol
Sigafoos (1998)	Discriminative stimuli (SDs)	Preferred item out of reach	Preferred item within reach	Requesting the item	Reaching the item
Sigafoos and Roberts-Pennell (1999)	Both MOs and SDs	Offered with the wrong item	Offered with the correct item	Rejecting the wrong item	Reaching the correct item
Sigafoos and Drasgow (2001)	Discriminative stimuli (SDs)	The presence of SGD	The absence of SGD	Requesting the items or activities using SGD	Requesting the items or activities using manual signs

Table 5. Summary of quality of study parameters, including maintenance, generalization, reliability, intervention fidelity, and social validity.

Citation	Maintenance	Generalization	Reliability	IF	SV
Anderson et al. (2010)	–	–	+	–	–
Brown et al. (2000)	–	–	+	+	–
Duker et al. (1994)	–	–	+	–	–
Fisher et al. (1998)	–	–	+	–	–
Grow et al. (2010)	6 months	Across staff from group home (TP only during intervention)	+	+	–
Horner et al. (1986)	–	Across 10 positive items in a nontrained store (only TN during both BL and intervention)	+	–	+
Hung (1980)	Interval NS	Across new sets of yes-no items; (TP & TN collectively only during intervention)	+	–	–
Kreibich et al. (2015)	–	Across untaught long engagement tasks; (TP & TN separately during both BL and intervention)	+	+	–
Kuhn et al. (2010)	–	Across other busy and non-busy activities (TP & TN separately only during intervention)	+	+	–
Magpusao and Cipani (1989)	–	–	+	–	–
Reichle et al. (1984)	–	Across natural activities and teaching assistants (TP & TN separately only during intervention)	+	–	–
Reichle and Brown (1986)	–	Across new staff and in day activity center environment (TP & TN separately only during intervention)	+	–	–
Reichle and Johnston (1999)	–	Across other times and locations different from the intervention (TP & TN separately only during intervention)	+	+	–
Reichle et al. (2005)	–	Across two task components (TP & TN separately during baseline)	+	–	–
Sigafoos et al. (1989)	–	Across untaught items (TP & TN separately only during intervention)	+ (<80%)	–	–
Sigafoos et al. (1990)	–	–	+	–	–
Sigafoos and Reichle (1992)	Interval NS	–	+	+	–
Sigafoos (1998)	–	–	+	–	–
Sigafoos and Roberts-Pennell (1999)	2 weeks, 3 months	Across 2 untaught pairs of items; across new trainer (only the probes for untaught pairs were assessed for TP & TN separately during both BL and intervention)	+	+	–
Sigafoos and Drasgow (2001)	–	–	+	–	–

Notes: IV refers to implementation fidelity, and SV refers to social validity; TP refers to target skill(s) under positive teaching condition(s), and TN refers to target skill(s) under negative teaching condition(s); NS refers to not specified.

Participant characteristics

A total of 31 participants meeting the inclusion criteria participated in the 20 studies identified. Twenty-seven (87.1%) participants were male, while 4 (12.9%) were female. The mean participant age was 18.3 years old (range: 4.3 — 45 years old; SD = 12.6 or about 12 years and 7 months). Ten (32.3%) of the included 31 participants were reported to have a moderate intellectual delay/disability, while the rest 21 (67.7%) participants were reported to have a severe intellectual delay/disability. With respect to the disability diagnosis, ten (32.3%) were diagnosed with autism, five (16.1%) with Down’s syndrome, two (6.5%) with a possible seizure disorder, two (6.5%) with cerebral palsy, two (6.5%) with communication impairment, two (6.5%) with spastic quadraparesis (not ambulatory), two (6.5%) with microcephaly, one (3.2%) with disruptive behavior disorder (not otherwise specified), one (3.2%) with profoundly deaf and legally blind, one (3.2%) with ADHD, one (3.2%) with developmental disability, one (3.2%) with several minor physical disabilities, and one (3.2%) with pervasive developmental disorder.

The expressive communication level prior to intervention was reported for 30 (96.8%) participants. Five (16.1%) produced sentences; three (9.7%) produced single words; four (12.9%) used manual signs; five (16.1%) communicated through pre-linguistic communicative acts (e.g. babbling, vocalizations, gestures, facial expressions); six (19.4%) produced a combination of pre-linguistic communicative acts and other conventional communicative acts (e.g. manual signs, graphic symbol); and three (9.7%) were nonverbal. Besides, three (9.7%) were reported to produce speech; and one (3.2%) had no systematic learning of symbol use.

Although imitation skill has been reported as an important predictor for language development (e.g. Yoder and Layton 1988, Johnston et al. 2012), its status was reported for only four (12.9%) participants. Anecdotally, one participant was reported to engage in gestural imitation, one in verbal word imitation, while two participants were reported to be lacking in vocal and motor imitation. Last, no information about problem behavior was reported for the 18 (58.1%) participants, the remaining 13 (41.9%) were reported to engage in problem behavior. Four (12.9%) engaged in problem behavior for access to attention and/or tangible items, three (9.7%) for escape from demands or activities, one (3.2%) for automatic reinforcement, and five (16.1%) with the function(s) of problem behavior unspecified.

Intervention characteristics

Experimental design

For the 20 studies identified, three (15%) studies implemented a quasi-experimental design (e.g. AB design); seven (35%) used a multiple baseline or probe design; three (15%) used an alternating treatment design; and two (10%) used a reversal design. In addition, five (25%) studies adopted a combined design (e.g. the combination of AB, reversal, and alternating treatment designs; the combination of a multiple probe design across communicative functions and a multiple baseline design across object sets; the combination of a multiple probe design and a changing criterion design).

Implementer

Among the 20 identified studies, the teachers implemented the intervention in five (25%) studies, therapists in four (20%) studies, experimenters in three (15%) studies, research assistants who were also qualified special education teachers in two (10%) studies, trained graduate students in one (5%) study, trained graduate students or parents in one (5%) study, trained graduate students or qualified special education teachers in one (5%) study, program staff and undergraduate research assistants in one (5%) study, senior direct-service staff in one study, and a PhD-level clinician in one (5%) study.

Setting

With respect to the contexts where the intervention was conducted, nine (45%) of the 20 studies were implemented in schools (e.g. classroom), followed by five (25%) in residential facilities (e.g. group home, sheltered workshop, habilitation center), four (20%) in clinical settings (e.g. laboratory room, inpatient unit, therapy room), one (5%) in community (i.e. stores), and one (5%) in two different settings (specifically, inpatient unit for one participant and home for the other participant in the study).

Antecedent variable[s] manipulated

As shown in Table 4, in terms of the range of antecedent variable(s) that were manipulated to make the target requesting or rejecting skill conditional, ten (50%) of the 20 identified studies manipulated the discriminative stimuli (S^Ds) to create different teaching conditions. In general, there are two different ways of presenting overt signals. One is the presentation of overt signals for each of the teaching conditions involved. For instance, in the study by Fisher et al. (1998), a drawing of a boy with toys indicated a toy-requesting teaching condition with the signal of available toys, and a drawing of a boy on an adult’s shoulders indicated an attention-requesting condition with the signal of available attention. The other is the presentation of overt signal(s) for certain teaching condition(s) but the absence of the overt signal(s) for the other teaching condition(s). For example, in the study by Sigafoos and Drasgow (2001), the presence of a speech-generating device (SGD) indicated the teaching condition of using the SGD to request items or activities, while the absence of the SGD indicated the teaching condition of using manual signs to make the request. Similarly, in the study by Grow et al. (2010), the absence of a black lanyard indicated the attention-requesting teaching condition with the signal of available attention, while the presence of a black lanyard indicated the no-request teaching condition with the signal of unavailable attention.

Three (15%) studies manipulated the motivating operations (MOs) to create different teaching conditions. Recall that MOs were defined as any stimulus, condition, or event that momentarily alters the value of some stimulus as a reinforcer and evokes all responses that have produced that reinforcer in the past. Specifically, among the three (15%) studies, establishing operations were manipulated to be present for one teaching condition but to be absent for the other teaching condition. For example, in the study by Duker et al. (1994), incomplete items (EO-present) were presented to the learner for the learner to request the items (i.e. a requesting teaching condition), while the same but complete items (EO-absent) were presented for the learner not to request the items (i.e. a no-requesting teaching condition). Similarly, in the study by Magpusao and Cipani (1989), the toothbrush and toothpaste were alternated between being missing and being present across teaching opportunities. Thus, during each teaching opportunity, the absence of one item (EO-present) sets up the teaching condition for the learner to request that item, while the presence of the other item (EO-absent) sets up the teaching condition for the learner not to request the other item.

In addition, the remaining seven (35%) studies manipulated both S^Ds and MOs to create different teaching conditions in that the potential impacts of S^Ds and MOs were not explicitly separated. For instance, in the early study by Hung (1980), when a preferred item was presented, the learner was supposed to request it (a requesting teaching condition); but when a non-preferred item was presented, the learner was supposed to reject it (a rejecting teaching condition). In this case, it is possible that both the reinforcing value of the items involved in the two teaching conditions (MOs-manipulated) and the physical characteristics of these items (S^Ds-manipulated) exerted stimulus control over the learner’s discriminated responding between the two conditions. Similarly, in the study by Sigafoos and Roberts-Pennell (1999), after the learner reached for an item, he was offered either the item reached for (i.e. the correct item) or the other item (i.e. the wrong item). When offered the correct item, the learner was supposed to accept it; but when offered the wrong item, he was supposed to reject it. In this study, the reinforcing value of the items offered to the learner were not explicitly measured, therefore, both MOs and S^Ds had been potentially manipulated for the accepting and rejecting teaching conditions.

Positive and negative teaching condition(s)

As shown in Table 4, nineteen (95%) of the 20 studies considered only one positive and one negative teaching conditions, while the remaining one study (Reichle et al. 1984) considered one positive teaching condition (i.e. requesting teaching condition in which a preferred item was presented for the learner to request), and two negative teaching conditions (i.e. rejecting teaching condition in which a neutral to undesired item was presented for the learner to reject, and labeling teaching condition in which an item was presented for the learner to label). Besides, in only one (5%) study (Hung 1980), the two teaching conditions consisted of one requesting teaching condition that was designated as the positive teaching condition and one rejecting condition that was designated as the negative one.

Target requesting/rejecting skill[s] under the positive teaching condition[s]

For the target requesting or rejecting skill(s) under the positive teaching condition(s), ten (52%) studies targeted the skill of requesting items, three (15%) for the skill of requesting attention, two (10%) for the skill of requesting items or activities, one (5%) for the skill of requesting assistance, one (5%) for the skill of requesting a break, one (5%) for the skill of requesting termination of punishment, and one (5%) for the skill of rejecting the wrong item which was offered to the learner different from the one he had reached for. For the remaining one (5%) study, the target skill under the positive teaching condition for one participant was to request items or attention, and to request ‘work done’ for the other participant.

Target skill[s] under the negative teaching condition[s]

Four (20%) of the 20 identified studies adopted silence (i.e. the absence of the target requesting or rejecting skill[s] under positive teaching condition) as the target skill under the negative teaching condition (i.e. Duker et al. 1994, Anderson et al. 2010, Grow et al. 2010, Kuhn et al. 2010). In the remaining 16 (80%) studies, alternative appropriate behavior was targeted as the skill under the negative teaching condition[s], which could be different communicative functions (e.g. labeling the item [Reichle and Brown 1986]; rejecting the item [Hung 1980, Reichle et al. 1984]), different communicative tools or modes (e.g. requesting using generalized symbol instead of explicit symbols [Sigafoos and Reichle 1992]; requesting using manual signs instead of using an SGD [Sigafoos and Drasgow 2001]), or other proper behavioral responses (e.g. taking, selecting, or reaching for the item on the learner’s own [e.g. Horner et al. 1986, Sigafoos 1998, Sigafoos and Roberts-Pennell 1999]; engaging in the task [e.g. Kreibich et al. 2015]).

Quality of study parameters

Reliability

All of the 20 studies reported measures of reliability, and only one (5%) study (i.e. Sigafoos et al. 1989) reported reliability lower than 80%, which was due to disagreements on the absolute number of multiple responses to a single symbol.

Fidelity and social validity

Only seven (35%) of the 20 studies reported intervention fidelity, while only one (5%) study (i.e. Horner et al. 1986) documented social validity, which was conducted through direct observation.

Maintenance

Four (20%) of the 20 studies addressed the maintenance of the target skills for positive and/or negative teaching conditions. Although the maintenance effects were reported to be positive, only two (10%) of them specified the concrete intervals between the end of the intervention and the maintenance probe(s) (i.e. 6 months [Grow et al. 2010]; 2 weeks and 3 months [Sigafoos and Roberts-Pennell 1999]).

Generalization

Eleven (55%) of the 20 identified studies investigated the generalization of the conditional requesting or rejecting intervention effects, with two (10%) across untaught items, one (5%) across interventionists, one (5%) across settings, one (5%) across untaught tasks, one (5%) across untaught task components, one (5%) across activities, and four (20%) across more than one dimensions such as interventionists, settings, activities, and/or untaught items.

In spite of the seemingly positive generalization effects reported, among the 11 (55%) studies that addressed generalization, seven (35%) reported the performance of the target skills for the positive and/or negative teaching conditions during generalization probes only throughout intervention condition[s]. One (5%) reported the performance of the target skills for the positive and negative teaching conditions during generalization probes separately throughout both baseline and intervention conditions. One (5%) reported the performance of the target skill for the negative teaching condition during generalization probes throughout both baseline and intervention conditions. One (5%) reported the performance of the target skills for the positive and negative teaching conditions during generalization probes separately only throughout the baseline condition. The remaining one (5%) study reported two types of generalization effects (i.e. across untaught items, and across new trainer), but it only reported the performance of the target skills for the positive and negative conditions separately throughout baseline and intervention conditions for the generalization across untaught items.

Discussion

This scoping review synthesized communication intervention studies that addressed conditional requesting or rejecting skills for individuals with moderate to severe developmental disabilities. A systematic search identified 20 studies published between 1980 and 2017. The evidence base yielded found that, overall, individuals with moderate to severe developmental disabilities learned to request and/or reject conditionally. Additionally, maintenance and generalization over a variety of parameters were reported in one fifth and about half of the studies identified, respectively. Further discussion was elaborated with regard to main findings, limitations of this review, and directions for future research, respectively, in the following sections.

Summary of main findings

In relation to the first goal of the current study on profiling the range of the participants involved in this study, only a relatively small number of participants (n < 50) with a broad age range between 4 and 45 years old with various diagnoses, diverse pre-intervention expressive communication skills and moderate to severe intellectual delays/disabilities were involved in the pool of 20 studies. Information on imitation skills and the status of problem behavior was unreported for a majority of the participants analyzed, although both of them were reported as important predictors for communication acquisition (e.g. McCathren et al. 1999, Rogers et al. 2003). The age range of the participants with moderate to severe developmental disabilities in the review suggested that virtually no information is currently available on conditional communicative production (at least on conditional requesting or rejecting) among preschool children with moderate to severe developmental delays.

With respect to the second goal of the current study on identifying the characteristics of conditional requesting and/or rejecting interventions for learners with moderate to severe developmental disabilities, several major findings were notable. First, a majority of the interventions were implemented by teachers or trained researchers (e.g. trained graduate students, research assistants) in school or residential settings. Second, most of the studies manipulated discriminative stimuli (S^Ds), or both S^Ds and motivating operations (MOs) to make the target requesting or rejecting skill conditional, while relatively less studies only manipulated MOs to create conditions for discriminated responding. Third, in general the pool of the studies identified had different specific intervention goals in terms of conditional requesting and/or rejecting, and thus had diverse target skills for the positive and negative teaching conditions. To be specific, the target skills for the positive and negative teaching conditions could be making requests versus being silent (Duker et al. 1994, Anderson et al. 2010, Grow et al. 2010, Kuhn et al. 2010), requesting versus labeling (Reichle and Brown 1986, Sigafoos et al. 1989, 1990), requesting versus rejecting (Hung 1980), requesting versus rejecting and commenting (Reichle et al. 1984), requesting through manual sign versus requesting through an SGD (Sigafoos and Drasgow 2001), requesting through a generalized symbol versus requesting through an explicit symbol (Sigafoos and Reichle 1992), requesting attention versus requesting tangible items (Fisher et al. 1998), requesting a missing item versus not requesting a present item (Magpusao and Cipani 1989), requesting an item when it was out of reach but taking it when it was placed within reach (Sigafoos 1998, Reichle and Johnston 1999), requesting assistance when the task was difficult but completing it independently as proficiency developed (Reichle et al. 2005), rejecting an item offered wrongly but taking an item offered correctly (Horner et al. 1986, Sigafoos and Roberts-Pennell 1999), and requesting a break versus engaging in the task (Kreibich et al. 2015).

With the participant and intervention characteristics evaluated, the last goal of the current study was to evaluate the quality of the studies included. The results from this review suggested that the included studies may lack some methodological components. First, some of the identified studies utilized a quasi-experimental design. Second, few studies reported intervention fidelity and social validity, although most of the identified studies scrutinized reliability. Without a demonstration of adequate fidelity, it is not clear whether the procedures can be replicated. Without a demonstration of social validity, it is not clear whether the implementation of the procedures results in a practical change in terms of the outcomes that the interventionists purport to be establishing. Last, only few studies documented maintenance and generalization effects. In particular, the demonstration of generalization effects was less substantiated since a relatively large number of studies that reported generalization effects did not measure the target skills for positive and/or teaching conditions on the generalization probes across both baseline and intervention conditions.

Limitations of the current literature review

There are several limitations of this literature review. First, the current review only focused on two functional communicative functions that included requesting and rejecting. Consequently, generalizing the findings from this study to other communicative functions (e.g. commenting, labeling, answering questions) is somewhat limited. Similarly, because studies reviewed addressed individuals with more moderate/severe disabilities, it would be unwarranted to generalize the findings from this study to individuals with mild or profound developmental disabilities. This review was more focused on describing the extent of the literature on the specific topic of conditional requesting and/or rejecting for individuals with moderate to severe developmental disabilities and identifying potential research gaps to inform future research. Consequently, this study did not attempt to quantify intervention outcomes and detect the potential correlations between participant as well as intervention characteristics and participants’ intervention outcomes. Finally, it merits pointing out that, in spite of rigorous search procedures, it is still possible that not all studies meeting the inclusion criteria have been identified and included in this review and the studies published after July 2017 were not included in the current review.

Directions for future research

Notwithstanding the limitations inherent in the current review, the findings from this study provided some implications for future research direction. One potential research direction is to expand the current review by exploring ways to quantify the intervention outcomes and scrutinizing the study quality utilizing the full breadth of the quality indicators delineated by Horner and colleagues (2005), which are not the focus of the current review. Another reasonable research direction may be to conduct an analysis similar to the one conducted in this review but with focus on conditional use of other expressive communicative skills (e.g. answering questions, commenting, greeting). For instance, it may be important for a listener to know when it is and when it is not required to respond to a speaking partner’s utterance. The partner saying ‘Gee I wonder if it will snow’ does not require a response from the listener; but saying ‘What time is it’ do require a response. Similarly, it may be important for a learner to know how to greet people appropriately (e.g. saying ‘How are you’ on the playground but waving or smiling in the library). Similar reviews on these other communicative skills may further enrich evidence in this field. In addition, it may be helpful to extend the review scope to include individuals with mild or profound developmental disabilities so as to gain a more integrated overview over this field, especially for individuals with developmental disabilities.

Given that generalization was not well investigated among most of the studies identified and a well-mastered conditional communication production under balanced stimulus control should concurrently address both stimulus discrimination (i.e. differentially responding to two or more stimuli) and stimulus generalization (i.e. producing the same response to two or more stimuli) (Horner et al. 1982), one prominent research direction may be to systematize the conditional communication intervention with both the two aspects of conditional communication (i.e. stimulus discrimination and stimulus generalization) fully considered. More specifically, in the context of conditional communication use, a systematic conditional communication intervention should be able to demonstrate how a learner performs on taught positive and negative teaching exemplars and how s/he performs over the other untaught positive and negative exemplars.

In general, the conditional communication use has been sparingly examined during the period of 1985–2017, and more research on this regard is needed in the future considering the potentially theoretical and practical implications for individuals with developmental disabilities.