ESTABLISHING VERBAL REPERTOIRES IN CHILDREN WITH AUTISM USING FUNCTION-BASED VIDEO MODELING: DIRECT AND COLLATERAL EFFECTS By Joshua Benjamin Plavnick A DISSERTATION Submitted to the Michigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Special Education 2010 ABSTRACT ! ESTABLISHING VERBAL REPERTOIRES IN CHILDREN WITH AUTISM USING FUNCTION-BASED VIDEO MODELING: DIRECT AND COLLATERAL EFFECTS By Joshua Benjamin Plavnick Young children with autism may demonstrate severe language impairment including the absence of vocal speech. This may limit the ability to mand (i.e., request) for preferred items or events and can lead to the development of problematic behavior that functions as a mand. The purposes of the present investigations were to (a) identify the function of gestures emitted by young children with autism, (b) examine differences in outcomes of mand training when the function of gestural behavior was either incorporated or not incorporated into mand training procedures, and (c) examine collateral effects of the mand training conditions on related but nontargeted behaviors. In Experiment 1, gestural behavior emitted by 5 nonvocal children with autism was functionally analyzed using a single subject alternating treatment design. Repeated applications of test and control conditions revealed gestures functioned as mands for attention for 1 participant and as mands for assistance obtaining a preferred item or event for 4 participants. Experiment 2 examined the effects of a video modeling (VM) intervention on acquisition of vocal or picture exchange mands under 2 experimental conditions. Function-based VM targeted response forms that were functionally equivalent to the gestures examined during Experiment 1 whereas nonfunction-based VM targeted response forms that were not related to participants’ gestural mands. Differential effects of the VM conditions were examined using an alternating treatment within multiple probe across behaviors design. Participants acquired, generalized, and maintained mands taught during function-based ! VM but not during nonfunction-based VM. Additionally, the function-based condition resulted in clear improvements to problem behaviors and slight improvements to listener behaviors when compared to the nonfunction-based condition. Results are discussed in terms of the empirical and applied implications for function-based interventions and VM. ! DEDICATION Dedicated to Amy for 10 years of unparalleled love and support. iv ACKNOWLEDGEMENTS I would like to thank my advisor, Dr. Summer Ferreri, for her constant support over the past three years. I have learned a tremendous amount from her about research methods, behavior analysis, special education, and maintaining professionalism under all conditions. Thank you for your patience and for always creating optimal environmental conditions under which I could behave. I would like to thank my dissertation committee members, Dr. Troy Mariage, Dr. John Carlson, and Dr. Jodene Fine for their extensive guidance and support throughout this project. Your insights along the way were valuable learning opportunities for this dissertation and future research projects. Thank you to the Leadership Fellows who helped develop the functional analysis methodology used in the present investigation. I am especially grateful to Tami Mannes for her tireless data collection efforts. I am very thankful of all the educators, parents, and students who participated in this research study. I will forever be impacted by my experiences in your classrooms and with your children. I owe a special thank you to Mary Mariage for three wonderful years of collaboration and support. This dissertation would not exist were it not for the support of my family along the way. Thanks Mom, Dad, Matt, and Becky for a lifetime of support and for your helpful perspectives during my graduate training. Finally, I am deeply indebted to Amy and the girls for tireless patience and so much more. Thanks to Abby and Paige for reminding me that life is usually about the journey and not the destination. Amy, words cannot possibly express my devotion and appreciation for you or my v gratitude for the past 10 years. Thank you for bringing purpose to my life and teaching me about the importance of making real differences in the lives of others. vi TABLE OF CONTENTS LIST OF TABLES………………………………………………………………….. xi LIST OF FIGURES………………………………………………………………… xii CHAPTER 1 INTRODUCTION………………………………………………………………….. Autism………………………………………………………………………. Verbal Behavior…………………………………………………………….. Functional Behavior Assessment and Function-based Intervention………... Video Modeling………………………………………………………..…… Purpose of the Study………………………………………………………... 1 1 3 5 6 7 CHAPTER 2 LITERATURE REVIEW…………………………………………………………... Autism………………………………………………………………………. Characteristics………………………………………………………. Diagnosis……………………………………………………………. Communication Impairment………………………………………... Treatment…………………………………………………………… Applied Behavior Analysis and Autism Treatment………………………… Terminology and Early Applications of ABA to Children with Autism……………………………………………………………... Experimental Research Designs in ABA…………………………... Analysis of Data in ABA…………………………………………... Internal Validity…………………………………………….. External Validity……………………………………………. Summary……………………………………………………………. Verbal Behavior…………………………………………………………….. Terminology and Conceptualization………………………………... Mand………………………………………………………... Tact………………………………………………………….. Echoic……………………………………………………….. Intraverbal…………………………………………………... Applications to Children with Autism……………………………… Previous Research…………………………………………………... Functional Communication Training……………………………………….. Terminology and Conceptualization………………………………... Functional behavior assessment……………………………... Functional analysis……………………………………. Previous Research…………………………………………………... Future Research……………………………………………………... Video Modeling…………………………………………………………….. Conceptualization…………………………………………………… vii 8 8 8 9 12 13 13 14 17 23 24 24 25 26 26 27 27 27 28 28 31 36 36 36 38 40 45 48 48 Previous Research……………………………………………………. Future Research…………………………………………………….... Purpose of the Present Investigation………………………………………... Research Questions…………………………………………………………. Research Question 1…………………………………………………. Research Question 2…………………………………………………. Research Question 3…………………………………………………. Research Question 4…………………………………………………. Research Question 5…………………………………………………. 49 51 52 53 53 53 54 55 55 CHAPTER 3 METHOD…………………………………………………………………………… Experiment 1: Assessment………………………………………………….. Participants………………………………………………………….. Experimenter and Observers………………………………………... Materials…………………………………………………………….. Pre-experimental Procedure……………………………………….... Functional Assessment Inventory…………………………... Preference Assessment……………………………………… Functional Assessment Observations……………………….. Definition and Measurement of Dependent Variables……………… Experimental Design………………………………………………... Experimental Procedure…………………………………………….. Action……………………………………………………….. Escape………………………………………………………. Attention…………………………………………………….. Play………………………………………………………….. Interobserver Agreement……………………………………………. Procedural Integrity…………………………………………………. Data Analysis……………………………………………………….. 57 57 57 61 63 63 63 63 65 67 69 72 72 73 74 75 75 76 76 CHAPTER 4 RESULTS…………………………………………………………………………... Experiment 1: Assessment………………………………………………….. Fuller………………………………………………………………... Bailey……………………………………………………………….. Victor……………………………………………………………….. Morris……………………………………….………………………. Matthew…………………………………………………………….. 78 78 78 80 82 84 86 CHAPTER 5 DISCUSSION: ASSESSMENT……………………………………………………. Research Question 1:……………………………………………………….. Implications for Functional Analyses………………………………………. Implications for Mand Training…………………………………………….. Alternative Communication Systems……………………………….. 88 88 89 91 93 viii Limitations and Future Research…………………………………………… 94 CHAPTER 6 METHOD…………………………………………………………………………… Experiment 2: Intervention…………………………………………………. Participants and Setting……………………………………………... Materials…………………………………………………………….. Definition and Measurement of Dependent Variables……………… Procedure…………………………………………………………… Baseline……………………………………………………... Video Modeling: Phase 1…………………………………… Video Modeling: Phase 2…………………………………… Video Modeling: Phase 3…………………………………… Training for Generalization…………………………………. Generalization Probes………………………………………. Follow-up…………………………………………………… Experimental Design………………………………………………... Reliability and Validity……………………………………………... Measurement Reliability……………………………………. Internal Validity…………………………………………….. External Validity……………………………………………. Social Validity………………………………………………. Data Analysis………………………………………………. ………. 98 98 98 101 102 107 111 112 118 119 120 120 120 120 122 122 125 125 125 127 CHAPTER 7 RESULTS…………………………………………………………………………... Experiment 2: Intervention…………………………………………………. Fuller………………………………………………………………... Bailey……………………………………………………………….. Victor……………………………………………………………….. Morris……………………………………….………………………. Matthew…………………………………………………………….. 131 131 133 139 145 150 155 CHAPTER 8 DISCUSSION………………………………………………………………………. Intervention Research Questions…………………………………………… Question 2…………………………………………………………... Question 3…………………………………………………………... Question 4…………………………………………………………... Question 5…………………………………………………………... Video Modeling…………………………………………………………….. Implications for Practice……………………………………………………. Limitations and Future Research…………………………………………… 160 160 160 167 170 171 173 174 176 APPENDICES………………………………………………………………….…... 180 ix APPENDIX A: Observer Training Procedures……………………….…….. APPENDIX B: Inventory of Potential Communicative Acts………… APPENDIX C: Caregiver Preference Inventory………….…………….…... Paired Stimulus Preference Assessment Data Collection Instrument………….…………….…………….…………... APPENDIX D: Inventory of Potential Communicative Acts: Checklist for Observation #1………….…………….…….. APPENDIX E: Operational Definitions for Potential Communicative Behaviors………….…………….…………….…………… APPENDIX F: Operational Definitions of Potential Environmental Variables………….…………….………….. APPENDIX G: Environmental Observation Instrument for Observation #2 APPENDIX H: Example of Functional Analysis Procedures……………… APPENDIX I: Procedural Integrity Checklists for Functional Analysis…… APPENDIX J: Example of Video Modeling Procedures…………………… APPENDIX K: Procedural Integrity Checklist for Video Modeling……….. APPENDIX L: Social Validity Questionnaire………….……………….….. APPENDIX M: Parent Permission for Child to Participate………….…….. APPENDIX N: Teacher Consent for Participation………….…………….... 180 183 193 REFERENCES……………………………………………………………………… 234 x 197 198 200 205 207 209 213 218 221 223 226 230 LIST OF TABLES Table 2.1: Autism diagnostic criteria…………………………………………..….. 11 Table 3.1: Characteristics of participants………………………………………….. 62 Table 3.2: Ratings of preferences for each participant…………………………….. 65 Table 3.3: Operationally defined target behaviors………………………………… 69 Table 3.4: Functional analysis conditions with programmed variables…………… 71 Table 6.1: Target mands and topographies for each participant…………………… 104 Table 6.2: Timeline of research procedures for all participants…………………… 110 Table 6.3: Evocative events and consequences delivered for each target behavior 114 Table 6.4: IOA during baseline and video modeling conditions…………………... 124 Table 6.5: Social validity ratings for all experimental conditions…………………. 126 Table 7.1 Mean responding for each participant across all dependent measures and conditions………………………………………………………….. 132 Table E.1: Operational definitions for potential communicative behaviors identified during pre-experimental observation……………………….. 201 Table F.1: Operational definitions of conditions observed during pre-experimental observation……………………………………………………………... 206 xi LIST OF FIGURES Figure 4.1: Fuller’s rate of gestural behavior during FA conditions………………. 79 Figure 4.2: Bailey’s rate of gestural behavior during FA conditions……………… 81 Figure 4.3: Victor’s rate of gestural behavior during FA conditions……………… 83 Figure 4.4: Morris’ rate of gestural behavior during FA conditions………………. 85 Figure 4.5: Matthew’s rate of gestural behavior during FA conditions…………… 87 Figure 6.1: Sequence of conditions and phases……………………………………. 109 Figure 7.1: Percent of accurate manding for Fuller………………………………... 134 Figure 7.2: Percent of listener behaviors for Fuller ……………………………….. 136 Figure 7.3: Percent of problem behaviors for Fuller………………………………. 138 Figure 7.4: Percent of accurate manding for Bailey……………………………….. 140 Figure 7.5: Percent of listener behaviors for Bailey……………………………….. 142 Figure 7.6: Percent of problem behaviors for Bailey……………………………… 144 Figure 7.7: Percent of accurate manding for Victor……………………………….. 146 Figure 7.8: Percent of listener behaviors for Victor……………………………….. 148 Figure 7.9: Percent of problem behaviors for Victor………………………………. 149 Figure 7.10: Percent of accurate manding for Morris……………………………... 151 Figure 7.11: Percent of listener behaviors for Morris……………………………... 152 Figure 7.12: Percent of problem behaviors for Morris…………………………….. 154 Figure 7.13: Percent of accurate manding for Matthew…………………………… 156 Figure 7.14: Percent of listener behaviors for Matthew…………………………… 158 Figure 7.15: Percent of problem behaviors for Matthew…………………………... 159 xii CHAPTER 1 INTRODUCTION The following dissertation examines the behaviors children with autism use to communicate in lieu of vocal speech and how understanding those behaviors can help inform teaching procedures that lead to useful communication and potentially prevent the development or worsening of problematic behavior. The present chapter provides a brief overview of the topics that will be covered in greater detail in the remaining chapters of this dissertation. This includes an introduction to (a) autism characteristics and prevalence, (b) applied behavior analysis as a treatment for children with autism, (c) early communication development from a behavioral perspective (i.e., verbal behavior), (d) functional approaches to understanding early communicative behavior, and (e) video modeling as a method for teaching early communication. The introduction concludes with a statement of purpose for the dissertation. Autism The autism spectrum disorders (ASD) are neurodevelopmental disorders characterized by impairments in social interaction and communication along with the demonstration of restricted, repetitive, and stereotyped patterns of behavior (American Psychiatric Association [APA], 2000). Over the past 15 years, there has been an epidemic-like increase in the number of children diagnosed with ASD; current statistics suggest that 1 out of 110 children in the United States are diagnosed with an ASD (Kogan et al., 2009). Autism, or autistic disorder, is a severe form of ASD that affects approximately 1 in 769 individuals (Fombonne, 2007). The majority of individuals diagnosed with autism are educated in segregated settings, do not obtain high school diplomas or post-school employment, have minimal social relationships, do not live independently as adults, and demonstrate severe language impairment as adults (Howlin, Goode, 1 Hutton, & Rutter, 2004). Though long-term outcomes can be difficult to predict, longitudinal studies suggest children who acquire functional communication by age 5 are more likely to experience some autonomy as adults (Dawson, 2008; Howlin et al.). Functional communication is defined as the ability to consistently produce five or more communicative responses that can be understood by others (Yoder & Stone, 2006; Howlin et al., 2004). Children who do not develop or are not successfully taught a form of communication that allows them to convey basic needs and wants to others are at risk for lifelong impairments across multiple adaptive domains and for the development of problematic behaviors such as severe aggression or self-injury (Carr & Durand, 1985; Dawson, 2008; Iwata, Dorsey, Slifer, Bauman, & Richman, 1982; Tager-Flusberg et al., 2009). Early educational interventions that enable the acquisition of speech and language repertoires or provide a usable alternative (e.g., sign language, pictorial exchange) are therefore critical for young children with autism (National Autism Center Standards Report [NSR], 2009). Due to the proliferation of treatments for children with autism over the past 10 to 15 years, parents, educators, speech pathologists and other service providers may struggle to identify treatments likely to produce positive outcomes (Wilczynski & Christian, 2008). In response to this issue, several agencies have developed guidelines for evaluating the scientific merit of treatments for individuals with autism (NSR, 2009; Romanczyk & Gillis, 2008). Scientific merit is determined by evaluating quality and quantity of research supporting a specific treatment. If research for a treatment meets established quality and quantity indicators, consumers can take some solace that the treatment is likely to be effective. Collectively, treatments that meet this standard are referred to as evidence-based practices (EBPs). An overwhelming number of EBPs for teaching communication to children with autism (i.e., greater 2 than 85%) come from the field of applied behavior analysis (National Research Council [NRC], 2001; NSR; New York State Department of Health, 1999). Applied behavior analysis (ABA) involves the systematic application of behavioral principles to improve socially relevant behaviors (Cooper, Herron, & Heward, 2007). ABA interventions have been validated primarily through single subject research, which is an experimental approach to research that allows for causal inferences about environment-behavior relations (Kratochwill et al., 2010). Critical features of ABA include an emphasis on the behavior of an individual, reliance on intervention approaches based on empirical principles of individual behavior, and analysis of outcomes involving behaviors that impact the day-to-day life of the person receiving treatment (Cooper et al.). The focus of an ABA intervention is to produce visible improvement in socially relevant behavior such that other people in an individual’s life can identify positive behavioral change (Baer, Wolf, & Risley, 1968; 1987; Cooper et al.). Verbal Behavior From a behavior analytic perspective, communication can be conceptualized based on Skinner’s (1957) functional description of verbal behavior. Skinner functionally defined verbal behavior as “behavior reinforced through the mediation of other persons.” (p. 14). This includes all behavior that involves a social interaction between a speaker and a listener (i.e., communicative partner) wherein the speaker gains access to reinforcement through the listener’s behavior (Sundberg, 2007). Speech, sign language or other gestures, pictorial communication, and problem behaviors could all be examples of verbal behavior. If a child asks a parent for juice by emitting the vocalization “juice,” points to a jar of juice in the refrigerator, or stands in front of the refrigerator screaming, they are engaging in verbal behavior if the response occurs as a result of 3 a learning history whereby listeners respond to the behavior by delivering juice. By contrast, nonverbal behavior includes responses that have an effect on the physical environment but not other people in the environment. If the child goes to the refrigerator and obtains juice, this would be classified as nonverbal behavior. Within Skinner’s (1957) framework, verbal behavior is broken down into a series of operants (i.e., class of behaviors); each one defined according to the functional relation between the behavior and specific antecedents and consequences. Skinner developed unique names for these operants to reduce confusion with terms previously used to classify the function of communicative behavior such as request, label, or imitate. The verbal operant described in the juice example above is termed a “mand” and is, loosely speaking, a request for an item, action, or information. Other verbal operants relevant to early language training include the tact (similar to labels), the echoic (similar to vocal imitations), and the intraverbal (similar to responding to communication of others) (Lerman et al., 2005). The mand is critical in early communication training for individuals with deficient verbal repertoires as it is the only verbal operant that allows an individual to control his or her environment (Sundberg & Michael, 2001). For example, an individual can indicate when he does or does not want attention from a caregiver or indicate when he needs assistance with a particular task. Teaching children with autism to emit mands in a form others can understand is critical for future development and yet, some children are not taught effective verbal repertoires or fail to respond to current treatments (Bodfish, 2004; Dawson, 2008). Remarkably, some of these children develop idiosyncratic behavior that functions as a verbal operant: most likely as a type of mand (Bijou & Baer, 1965; Keen, 2005; Siegel-Causey & Guess, 1989; Sigafoos et al., 2000; Sundberg, 2007; Wetherby, Warren, & Reichle, 1998). Topographical examples include 4 problematic behaviors such as aggression, pica, or self-injury and nonproblematic gestures such as directed eye gaze, hand gestures, or grabbing others. Functional Behavior Assessment and Function-based Intervention During the late 1970’s and early 1980’s, researchers examining problematic behavior (e.g., aggression, self-injury, property destruction) of children with and without severe disabilities were focused on using behavioral assessment to better understand the function, or loosely speaking, the purpose of problem behaviors (Carr & Durand, 1985; Carr, Newsome, & Binkoff, 1976; Iwata et al., 1982/1994). A functional analysis is a type of behavioral assessment that manipulates environmental variables to identify antecedents and consequences that produce problem behavior. Once the environmental variables that evoke and maintain problematic behavior are identified, interventions can be designed to interrupt the relation between the behavior and the reinforcing consequence (Iwata et al.). A commonly used example of a function-based intervention (i.e., intervention based on the results of a functional analysis) is functional communication training (FCT). The general premise of FCT is to replace a problem behavior with a mand that yields the same or better outcomes for the student and is acceptable to caregivers (Carr & Durand, 1985). For example, a child who engages in aggressive behavior maintained by escape from task demands may be taught to say “break” whenever difficult tasks are assigned. Manding for a break using the new vocal response is then consequated by the delivery of a break whereas aggression is placed on escape extinction (i.e., a break from the task is not provided). FCT is an extremely reliable technology for teaching mands as an alternative to problem behavior and is an evidence-based practice for treating problem behavior emitted by children with autism (NSR, 2009; Tiger, Hanley, & Bruzek, 2008). 5 Applications of the FCT logic to verbal behavior that is not problematic (e.g., idiosyncratic gestures) may demonstrate the potential for teaching mands that an individual can use in a variety of environments. Such an approach could be of benefit to children who lack an effective mand repertoire, but do not engage in problematic behavior with enough frequency or severity to warrant a functional analysis of problem behavior. Extending FCT to gestural behavior constitutes a proactive approach to early intervention for children with autism and might be effective in preventing the development or worsening of problematic behavior. Video Modeling Once the function of a gesture is identified, intervention agents could shape or model an alternative mand that listeners are more likely to understand. Each of these has potential disadvantages as shaping language is a slow process and traditional modeling can create awkward response chains or prompt dependency that inhibit independent responding (Bourret, 2004). Video modeling (VM) might allow for rapid acquisition, similar to other modeling approaches, yet reduce the likelihood that a child becomes dependent on the model to emit the target response. Though VM has been identified as an EBP for children with autism, participants in most investigations are taught social or play skills and are identified as having mild or moderate autism (NSR, 2009). Additional research is needed to better understand the extent to which VM is effective for teaching new skills to children with severe autism (Rayner, Denholm, & Sigafoos, 2009). Purpose of the Study The following investigations examine a functional approach to the assessment and subsequent treatment of deficient verbal repertoires demonstrated by children diagnosed with autism. The purpose of the first investigation was to identify the functional properties of gestures 6 emitted by nonvocal children with autism. Specifically, the experimental procedure assessed for the function of each child’s gesture as a mand for assistance obtaining a preferred item or activity, a mand for attention, or a mand for the removal of an aversive stimulus. The purpose of the second investigation was to examine the degree to which novel mands were acquired, generalized, and maintained following the administration of VM during two distinct conditions; one condition utilized the pre-existing functional relation identified during the first experiment and the second condition utilized nonfunction-based procedures. A secondary purpose of this investigation was to examine the extent to which participants engaged in behavior that has been correlated with mand acquisition during each experimental condition. Correlated behavior included listener responses (i.e., orienting to speaker, following directions) and problem behavior (e.g., head-banging, scratching, yelling). 7 CHAPTER 2 LITERATURE REVIEW The following chapter reviews research literature emphasizing behavioral approaches to teaching functional communication to children with autism who demonstrate severe communicative deficits. A broad overview of autism is followed by a description of ABA as an approach to treatment. Communication training for children with severe autism is discussed within a framework consistent with Skinner’s (1957) analysis of verbal behavior. The application of functional analysis and FCT to assess and treat problem behavior is reviewed and theoretically extended to children who do not emit conventional mand topographies, but who also do not engage in frequent episodes of severe problem behavior. Video modeling is then discussed as a potential procedure for teaching new mands to children with autism. The chapter concludes with specific research questions and reviews of the literature most relevant to those questions. Autism Characteristics Autism spectrum disorders (ASD) are neurodevelopmental disorders characterized by a series of behavioral excesses and deficits (Ospina et al., 2008). An individual with an ASD may be specifically diagnosed with Autistic disorder (i.e. autism), Asperger’s syndrome (AS), pervasive developmental delay not otherwise specified (PDD-NOS), Rett’s syndrome, or childhood disintegrative disorder. Current estimates suggest 1 in 110 individuals are diagnosed with some form of ASD (Kogan et al., 2009). The term spectrum refers to the heterogeneity of individuals diagnosed with an ASD (Volkmar & Lord, 2007). An individual with AS might develop a full language repertoire and live independently yet demonstrate impairments in social interaction that isolate them from 8 others and can lead to depression (McPartland & Klin, 2006). Alternatively, a child with autism may not emit any language and might require intensive lifelong support for a variety of independent living tasks such as shopping, eating, and navigating a daily schedule (Dawson, 2008). Kanner (1943) published the first description of individuals with autism in a paper examining 11 children described as having disturbances of affective behavior. The children were characterized by a deficiency in affective contact, a desire for sameness, a fascination with objects, and severe impairments to communicative competence. Some of the children were mute and those that spoke frequently demonstrated echolalia, literalness, and pronoun reversal (Volkmar & Lord, 2007). Kanner’s original descriptions of ‘early infantile autism’ were consistent with some of the present day descriptions of children with more severe forms of ASD (Bodfish, 2004; Tager-Flusberg, 2009). The prevalence of autism is approximately 1 in 769 individuals (Fombonne, 2007). It is a pervasive and lifelong neurodevelopment disorder with a diverse range of associated behavioral characteristics in three core areas: social interaction, communication, and restricted or repetitive interests (Caronna, Milunsky, & Tager-Flusberg, 2008). The extreme heterogeneity among individuals with autism is likely due to co-occurring diagnoses of cognitive impairment, language impairment, or both in some cases (Dawson, 2008; Volkmar & Lord, 2007). This diversity persists throughout the lifespan as adults with autism may be independent and present few observable characteristics or demonstrate severe autistic symptoms and are fully dependent on others (Mailick Seltzer, Shattuck, Abbeduto, & Greenberg, 2005). Diagnosis 9 Autistic disorder is diagnosed based on the characteristics presented in Table 2.1 (APA, 2000). There are three broad criteria that must be met with several additional requirements within each category. Generally speaking, children must demonstrate qualitative impairments in the social and communicative domains as well as demonstrating a pattern of restricted, repetitive, and stereotyped behaviors all before the age of 3 to be diagnosed with autism (Snow, Lecavalier, & Houts, 2009). Additionally, these characteristics cannot be better explained by Rett’s disorder or childhood disintegrative disorder. Standardized observation and rating instruments based on the three-domain model have been developed and are often used by trained psychologists or psychiatrists to diagnose autism in young children (Snow et al., 2009). Identification of autism using this model is made by a diagnostician’s professional judgment. Accurate diagnosis is critical as several related disorders, such as developmental delays or specific language impairment, can appear similar but may not be indicative of a similar treatment pattern (Goldstein, Naglieri, & Ozonoff, 2008). 10 Table 2.1 Autism diagnostic criteria Category Autism Diagnostic Criteria A At least six items from (1), (2), and (3), with at least two from (1), and one each from (2) and (3) (1) Qualitative impairment in social interaction, as manifested by at least two of the following: • marked impairment in the use of multiple nonverbal behaviors such as eye-to-eye gaze, facial expression, body posture, and gestures to regulate social interaction • failure to develop peer relationships at developmental level • a lack of spontaneous seeking to share enjoyment, interests, or achievements with other people, (e.g., by a lack of showing, bringing, or pointing out objects of interest to other people) • lack of social or emotional reciprocity (2) Qualitative impairments in communication as manifested by at least one of the following: • delay in, or total lack of, the development of spoken language (not accompanied by an attempt to compensate through alternative modes of communication such as gesture or mime) • in individuals with adequate speech, marked impairment in the ability to initiate or sustain a conversation with others • stereotyped and repetitive use of language or idiosyncratic language • lack of varied, spontaneous make-believe play or social imitative play appropriate to developmental level (3) Restricted repetitive and stereotyped patterns of behavior, interests and activities, as manifested by at least one of the following: • encompassing preoccupation with one or more stereotyped and restricted patterns of interest that is abnormal either in intensity or focus • inflexible adherence to specific, nonfunctional routines or rituals • stereotyped and repetitive motor mannerisms (e.g., hand or finger flapping or twisting, or complex whole-body movements) • persistent preoccupation with parts of objects B Delays or abnormal functioning in at least one of the following areas, with onset prior to age 3 (1) social interaction (2) language as used in social communication (3) symbolic or imaginative play C The disturbance is not better accounted for by Rett's disorder or childhood disintegrative disorder 11 Communication Impairment A core characteristic of individuals with autism is a qualitative impairment in communicative behavior (APA, 2000). Some children do not acquire basic speech or alternative communicative behavior to convey needs and wants (Dawson, 2008). As a result of the communicative impairment associated with autism, some children come to rely on gestural and in some cases, problematic behavior to convey communicative functions (Keen, 2005; Keen, Sigafoos, & Woodyatt, 2001; Sigafoos et al., 2000; Sundberg, 2007). Caregivers for children with severe communication impairment may struggle to recognize the occurrence of a communicative behavior; and thereby fail to deliver reinforcement following a child’s communicative attempts (Halle & Meadan, 2007; Keen et al., 2001; Meadan, Halle, Watkins, & Chadsey, 2006). This can lead to fewer opportunities to engage in social interactions with others and the development of severe problem behavior to convey basic needs and wants (Halle & Meadan). The subset of children with autism who also demonstrate a severe impairment in communication are at great risk for social isolation, lifelong dependence on others, and the development of problematic behavior (Cederlund, Hagberg, Billstedt, Gillberg, & Gillberg, 2008; Hartley, Sikora, & McCoy, 2008; Howlin et al., 2004). Effective treatment for this group of children is critical. One of the first steps toward addressing the needs of children with autism and severe communication impairment is teaching them to request basic needs and wants from a variety of listeners in their everyday environments (Yoder & Stone, 2006). 12 Treatment The National Autism Center recently released the NSR (2009) to provide consumers with information about EBPs for individuals with autism. The NSR established criteria for evaluating intervention research studies and for determining whether a sufficient number of rigorous studies exist to classify an intervention as evidence-based. This criteria was similar to other recent guidelines in psychology (Chambless et al., 1996), school psychology (Stoiber & Kratochwill, 2000), and special education (Odom et al., 2005). Research that met the highest standard of evidence employed random assignment of participants to treatment and control groups or single subject designs with at least three comparisons of the intervention and control conditions (NSR). Following an extensive review of over 7000 research studies from 1957-2007, the NSR (2009) identified 775 peer-reviewed research studies with sufficient methodological rigor to be included in an analysis of EBPs for children with ASD. Though no single treatment emerged as effective for all children all the time, over 85% of the established EBPs came from behavioral literature or used principles of behavior to produce positive outcomes for children (these procedures are collectively referred to as ABA). These findings were similar to previous reviews of intervention effectiveness for children with autism (NRC, 2001; New York State Department of Health, 1999). Applied Behavior Analysis and Autism Treatment The environmental influences that lead to behavioral excesses and deficits demonstrated by children with autism can be identified and manipulated to alter behavior using experimental research methods from ABA (Wacker, Berg, & Harding, 2008). Research in ABA examines the effectiveness of interventions based on behavioral principles to improve socially meaningful behavior for individual participants (Cooper, Herron, & Heward, 2007, p. 20). The interventions 13 are applied systematically to identify functional, or causal, relations between a specific independent variable and the behavior of interest (Kratochwill et al., 2010). The primary experimental methodology used to understand these functional relations and identify EBPs for children with autism is single-subject design (SSD; NSR, 2009). The purpose of the present section is to define critical terms in ABA and explain the SSD research methodology. Where applicable, relevant examples from the extant literature involving ABA interventions for children with autism will be discussed. Terminology and Early Applications of ABA to Children with Autism ABA investigations identify functional relations between environmental events occurring antecedent and consequential to a socially significant behavior (Cooper et al., 2007). Based on basic behavioral research pioneered by Skinner (1938), consequences are now known to be the driving force behind behavior. The terms used to describe functional relations between a consequence and a behavior include reinforcement, punishment, and extinction. A reinforcing consequence immediately follows a behavior and increases the future frequency of the behavior under similar conditions. The terms positive and negative are used to describe reinforcers that involve stimulus delivery and removal, respectively. A punishing consequence immediately follows a behavior and decreases the future frequency of the behavior under similar conditions. Similar to reinforcement, punishment can involve the presentation or removal of stimuli and is termed positive or negative punishment, respectively. Extinction is also a consequence that reduces a behavior, though it does so for different reasons than punishment (Cooper et al., 2007). Extinction is the removal of reinforcement following a behavior that has previously been reinforced. This results in an eventual decrease in 14 the behavior until reaching prereinforced levels. Sometimes, the immediate process of extinction is followed by an extinction burst, a brief increase in the frequency of the response. Understanding the effects of consequences on behavior has been particularly informative in the establishment of educational and behavioral interventions for children with autism (Ahearn, Clark, MacDonald, & Chung, 2007; Halle & Meadan, 2007; NSR, 2009). For example, early ABA research showed that adult attention in the form of reprimands following problem behavior increased the amount of problem behavior emitted by a child with autism (Wolf, Risley, & Mees, 1964). Despite attempting to punish the behavior, hospital staff had inadvertently strengthened the future probability of the response. However, by identifying attention as a reinforcing consequence, the experimenters accurately presumed that delivering attention contingent on an acceptable behavior and withholding attention following problem behavior (i.e., extinction) would lead to a change in the child’s behavior. Antecedent variables are also important in ABA research as they can control behavior through a previous history of being paired with specific consequences. One type of antecedent variable is a discriminative stimulus (SD). An SD is a stimulus in the presence of which a given response is reinforced and in the absence of which the same responses have occurred but have not been reinforced (Cooper et al., 2007). The presence or absence of this stimulus controls behavior through a history of association with differential reinforcement. Simply speaking, an SD signals the availability of reinforcement contingent on a specific behavior. Early investigations of procedures used to teach new skills to children with autism relied upon behavior coming under control of specific antecedent stimuli (Tarbox & Najdowski, 2008). In one example, Lovaas, Berberich, Perloff, and Schaeffer (1966) taught 2 mute children who demonstrated characteristics of autism to imitate successive approximations of vocal verbal 15 behaviors. The experimenters started by teaching each child to emit any sound within a few seconds of the adult emitting a sound. Over time, reinforcement was only delivered when the child’s sound approximated the adult’s model. The adult’s model was an SD that signaled the availability of reinforcement contingent on the child’s vocal imitation. The child learned this discrimination by receiving reinforcement following a correct imitation and not receiving reinforcement for an incorrect or no imitative response. After 23 days of training, participants had learned to imitate numerous sounds and as many as 31 words. Another type of antecedent variable that has an important effect on behavior is a motivating operation (MO; Laraway, Snycerski, Michael, & Poling, 2003). Michael (2007) defined a MO as an antecedent variable that temporarily alters the value of an item or event as a reinforcer and therefore temporarily alters the probability of a response that has been followed by that item or event in the past. The value of an item as a reinforcer and subsequent behavior can be increased or decreased depending on states of deprivation, satiation, or aversive stimulation. Simply speaking, the MO is what makes a person “want something” and produces behavior that has previously gotten whatever was “wanted” (Michael, 2004). Charlop, Schreibman, & Thibodeau (1985) provided one of the first examples of manipulating MOs to teach children with autism to request preferred items. In order to teach participants to spontaneously request, the experimenters held a preferred item in the child’s view but only gave the item to the child if they asked for it vocally. As children acquired prompted responses, the experimenters began increasing the amount of time between the presentation of the stimulus and the delivery of a prompt. The inaccessible preferred item was a MO. That is, presentation and withholding of a preferred item increased the value of the item and increased behavior that had been followed by delivery of the item in the past. 16 Laraway and colleagues (2003) noted that MOs could both increase or decrease the value of a reinforcer as well as behavior that had previously been followed by the reinfocer. A MO that temporarily increases the value of an item or event as a reinforcer and corresponds to an increase in behavior, as in the preceding example, is called an establishing operation (EO). An abolishing operation (AO) decreases the value of an item or event as a reinforcer and corresponds to a decrease in behavior. Experimental Research Designs in ABA In 1968, the Journal of Applied Behavior Analysis was established to publish empirical applications of the principles of behavior described above to improve behaviors of social importance. In the inaugural issue, Baer, Wolf, and Risley (1968) documented the dimensions of ABA research that would define the field. In addition to emphasizing the applied nature of behavioral research, Baer and colleagues noted the importance of research methods that would allow for a clear demonstration of the effects of behavioral principles on individual behavior. Briefly summarized, the critical dimensions of ABA include (a) the observable improvement of behaviors that make a difference in the day-to-day life of consumers (e.g., learning adaptive skills, reading, social skills, communication); (b) the collection and analysis of data prior to, during, and following intervention to assess the degree to which improvement has occurred and can be connected to the intervention; (c) interventions or teaching procedures based on principles of behavior (e.g., reinforcement, punishment, antecedent manipulations); and (d) outcomes that bring behavior to practical levels and generalize across caregivers, settings, and time (Baer et al., 1968; 1987). ABA research can be evaluated based on the extent to which it fulfills these dimensions (Cooper et al.). 17 As the emphasis of research in ABA is on individual behavior, the experimental procedures are somewhat dissimilar to the comparison of group outcomes that is more common within the social sciences (Cooper et al., 2007; Neef, 2006). ABA investigations typically begin with a question of curiosity as opposed to an explicit hypothesis or prediction (Johnston & Pennypacker, 2008). Therefore, ABA research does not attempt to confirm or disconfirm a hypothesis through experimentation. Instead, questions are posed to discover the extent to which an intervention clearly changes multiple characteristics of behavior and the process by which the intervention is effective (Franklin, Allison, & Gorman; Johnston & Pennypacker). Early research in ABA demonstrates how this inductive approach to experimental research is carried out. Ayllon and Michael (1959) taught nurses in a psychiatric hospital to respond differentially to the problematic and appropriate behavior of patients. Nurses were trained to withhold attention following problem behavior and to deliver attention and preferred items such as food and cigarettes contingent on an appropriate alternative behavior selected by the hospital staff. Problem and alternative behaviors were repeatedly measured during a baseline condition prior to the intervention and during the intervention condition. Problem behaviors decreased and appropriate alternative behaviors increased after the intervention was implemented. Though the single manipulation of experimental conditions from baseline to intervention was not sufficient to declare a functional relation, the possibility that contingent attention could rapidly alter behavior in applied settings was quickly assimilated into rigorous ABA investigations (Wolf et al., 1964). Wolf and colleagues (1964) extended the work of Ayllon and Michael (1959) by conducting one of the first experimental analyses in ABA to understand the reinforcing effects of attention on problem behavior. The experimenters taught a 3-year-old boy with autism, mental 18 retardation, and cataracts causing blindness to wear glasses that could reduce the effects of cataracts. After learning to wear glasses, the boy began throwing the glasses approximately twice each day, which frequently resulted in broken glasses and the delivery of attention in the form of reprimands from hospital staff. To reduce glasses throwing behavior, Wolf and colleagues instructed hospital staff to interact minimally with the boy while placing him in his room for 10 min immediately following an instance of throwing glasses; the behavior rapidly decreased. This first half of the investigation was similar to the method used by Ayllon and Michael (1959) in that a baseline phase was followed by an intervention phase. However, hospital employees responsible for administering the intervention were not certain that the time out procedure was responsible for decreasing the behavior. Therefore, the experimenters terminated the time out procedure following glasses throwing, which allowed for measurement of the behavior under the same environmental conditions as the original baseline. Glasses throwing behavior increased to original levels. The experimenters once again implemented the intervention and glasses throwing decreased to zero instances per day after six days. The investigation conducted by Wolf and colleagues (1964) was remarkable for the procedure used to establish internal validity. When the participant’s behavior covaried with the repeated application and removal of the independent variable, it became less likely that extraneous variables such as maturation or other events occurring in the hospital at that time were responsible for any changes in behavior. Thus, it was most likely that the independent variable was responsible for observed changes in behavior. This SSD was termed a reversal design as behavior was reversed when conditions alternated between baseline and intervention. A reversal design continues to be used in ABA today and is the most straightforward design for 19 demonstrating experimental control through manipulation of the independent variable (Cooper et al., 2007). The reversal design was the most common design during the early years of ABA, though two issues limited its applicability in some research contexts (Cooper et al., 2007). First, behaviors such as asking for preferred items might prove irreversible, as the response would come into contact with contingencies that maintained the behavior in the natural environment. Second, the baseline conditions that produce minimal levels of desired behavior or higher levels of problem behavior must be reinstituted following an effective intervention (Cooper et al., 2007). This approach may not be acceptable to practitioners in applied settings or could be dangerous when severe problem behavior is targeted. Risley and Hart (1968) provided one of the first systematic methods for identifying functional relations without using a reversal design. The experimenters repeatedly measured appropriate play and reporting of play activities by preschool children during a baseline condition. Following baseline, a reinforcement contingency was sequentially applied to each behavior. First, children received snack for answering questions about what they did with specific toys during free playtime. Next, the participants received snack only when they actually engaged in the behavior they reported. The results showed that participants demonstrated an increase in both behaviors only when the reinforcement contingency was applied. The Risley and Hart (1968) investigation was critical in establishing the use of ABA in school settings and for the application of the multiple baseline design (MBD). The MBD provides a different demonstration of experimental control than the reversal design. A MBD is useful for determining functional relations when an independent variable cannot be removed (e.g., language acquisition) or would be unethical to remove (e.g., a procedure that reduces 20 severe self-injurious behavior) in order to demonstrate a replication of the original baseline (Neef, 2006). The logic of the MBD is to demonstrate a change in responding when the intervention is sequentially applied to two or more independent behaviors. Independent behavior refers to (a) two or more different behaviors emitted by the same participant, (b) the same behavior emitted across two or more settings, or (c) the same behavior across two or more participants. Experimental control is established by demonstrating a steady state baseline for all independent behaviors. The intervention is then applied to one independent behavior at a time. A MBD allows for prediction of what that first behavior would look like if not for the application of the intervention following a steady baseline state. This prediction is then verified when the second independent behavior continues to demonstrate steady state responding in the baseline condition. Finally, replication of intervention outcomes are established when the intervention is applied to the second independent behavior. The multiple probe design (MPD) is a variant of the MBD and was developed by Horner and Baer (1978) to measure behavior that is likely to be extremely steady during baseline. For example, if a child has not yet acquired a behavior, it is likely to remain at zero during baseline conditions. Therefore, behavior that is subject to baseline conditions can be probed intermittently as opposed to measured during every experimental session. As SSD and behavioral research methods developed in the 1970s, additional design tactics were needed to answer questions such as “What environmental conditions reliably produce a specific behavior” or “Which of two interventions will produce the greatest outcomes for an individual student?” (Cooper et al., 2007). In order to meet this need in applied research, 21 Ulman and Sulzer-Azaroff (1975) translated Sidman’s (1960) description of a multielement design used to compare schedules of reinforcement in basic research. Iwata and colleagues (1982/1994) provided an early demonstration of the multielement design by comparing rates of problem behavior during several environmental conditions. The experimenters rapidly alternated between experimental conditions on a session-by-session basis. Frequency of problem behavior was measured during each condition, which allowed for a direct comparison of the conditions most likely to produce problem behavior. By identifying the environmental variables that produced problem behavior, the researchers could develop procedures to reduce the behavior. The multielement design is now commonly referred to as the alternating treatments design (ATD) due to the rapid alternation between conditions and the fact that it is frequently used to compare different treatments (Cooper et al., 2007; Kratochwill et al., 2010). The logic of the ATD is very similar to the reversal design as several rapidly applied reversal designs are used to demonstrate differential effects of two or more experimental conditions (Cooper et al., 2007). However, transition from one treatment to another does not require a demonstration of steady state responding following the application of the first treatment (Cooper et al.). Thus, a baseline is not necessary to demonstrate functional relations when using an ATD (Cooper et al.). Instead, treatments are applied in an alternating manner and according to a predetermined schedule. The ATD can therefore be a very efficient design for identifying the most effective treatment when selecting from several potential options (Neef, 2006). As in a reversal design, the experimenter predicts, verifies, and replicates behavioral patterns using an ATD. However, a single data point is used in ATDs to predict future behavior when that same treatment is applied, verify predicted behavior the last time that treatment was 22 applied, and replicate previous levels of behavior under the same treatment conditions (Cooper et al., 2007). Functional relations are identified when the data series obtained during one condition is clearly separated from the data series obtained in another condition (i.e., visible difference in level, trend, and nonoverlapping data) (Neef, 2006). The benefit of this design is that it allows a researcher, consultant, or practitioner to apply and test various treatments before selecting one for further use. The reversal, MBD/MPD, and ATD are commonly used to demonstrate experimental control and identify functional relations in a variety of present-day applied settings (Cooper et al., 2007; Kratochwill et al., 2010). These were the first designs pioneered by early ABA researchers to extend the principles of behavior learned in basic animal research to applied settings with humans (Risley, 2005). The common standard for identifying functional relations using these designs is three manipulations of the independent variable (Kratochwill et al.). Using a reversal, this means the experiment must include at least one replication of both the baseline and intervention conditions. A MBD must include baseline measurement and the sequential application of an intervention to three separate behaviors to meet the same criteria. An ATD must compare two or more conditions with at least five alternations between the conditions. Analysis of Data in ABA Baer (1977) boiled the analysis of data in ABA down to the identification of extremely potent variables that are highly effective in solving real problems. If an intervention leads to a difference in behavior, one should be able to clearly see such a difference. In alignment with this framework, visual analysis of graphically displayed data during repeated applications of various conditions is the most commonly used method for analyzing data in SSD research (Cooper et al., 23 2007; Horner et al., 2005; Kratochwill et al., 2010; Neef, 2006). This type of analysis allows for a full interpretation of the overall effectiveness of the intervention (Cooper et al.). Visual analysis consists of an interpretation of the level, trend, and variability of behavior across experimental conditions (Horner et al., 2005). The level is interpreted by comparing the mean rate of behavior during the various conditions. When analyzing trend, the researcher looks for acceleration or deceleration of behavior within a condition and compares the general trends across conditions. Variability refers to the fluctuation of behavior around the mean identified for each condition. In order to identify a functional relation from a SSD, an experimenter must provide a compelling display of data that shows (a) large differences in mean across conditions, (b) minimal latency to behavior change when conditions change, and (c) consistent trends in data each time the independent variable is applied and/or removed (Horner et al., 2005). Internal validity. The design and measurement of SSDs can rule out many threats to internal validity (e.g., time, history, unidentified variables) (Neef, 2006). This demonstration of experimental control occurs through multiple and repeated manipulations of the independent variable to show that behavior not only changes when the intervention is applied, but that it does so in a reliable and predictable manner (predictable based on the original application of the independent variable). Because of this emphasis on experimental control, effectively designed SSDs that demonstrate functional relations have a high level of internal validity (Johnston & Pennypacker, 2008). External validity. External validity within SSD occurs when the original results are replicated in (a) future experiments, (b) with different participants, (c) in different locations, or (d) with slight changes to the independent variable (Sidman, 1960; see also Cooper et al., 2007). Direct and systematic replications of initial experiments are the process by which external 24 validity is established (Cooper et al., 2007; Sidman, 1960). A direct replication is an exact replication of the original investigation and can include the same or new participants (Sidman). When a research study in ABA includes more than one participant, each participant beyond the first provides an instance of direct replication and demonstrates the reliability of the intervention (Cooper et al.). The rationale for this is that each participant represents his or her own control and therefore represents one complete experiment (Neef, 2006). This is the reason why only one participant is needed to demonstrate functional relations in behavioral research. Though direct replication provides a demonstration of the reliability of a functional relation, it does not speak to the generality of results to new environments (Sidman, 1960). The demonstration of generality using SSDs requires systematic replication (Sidman). Systematic replication is the process of replicating an experiment with variations in contextual variables such as characteristics of participants or settings in which the original experiment is applied (Sidman). When an experimenter intentionally alters conditions of previous experiments and obtains similar results, any identified functional relation can be said to have both reliability and generality (Cooper et al., 2007). This can occur within a single investigation when a design is created to include participants with varied characteristics, the intervention is applied across multiple settings, the intervention is applied across multiple target behaviors, or the administration of the design varies across applications. Systematic replication provides the rationale behind variations in participant characteristics frequently demonstrated in SSDs. When functional relations are demonstrated across participants with varied characteristics, the boundaries of generality increase (Neef, 2006; Sidman, 1960). Summary 25 ABA and its corresponding research methodology, SSD, represent a powerful approach for identifying EBPs in educational settings (Cooper et al., 2007; Horner et al., 2005; Johnston & Pennypacker, 2008; Kratochwill et al., 2010; NSR, 2009; NCR, 2001; Neef, 2006). Similar to all methods of scientific research, the clear demonstration of internal validity must precede any assessment of external validity of the findings (Neef). As such, the design of experiments in ABA is constructed and applied to rigorously control for any confounds that could decrease the internal validity of the investigation (Neef). Any modifications made to study characteristics in replications of the original investigation clarify the boundaries to which one can infer application of the procedure with similar results (Neef; Sidman). Verbal Behavior Terminology and Conceptualization Skinner (1957) wrote Verbal Behavior to inductively account for the controlling relations of speakers and listeners using empirically derived principles of behavior (Sundberg, 2007). Verbal behavior includes any behavior that is reinforced through the mediation of another person’s behavior (Skinner). When an individual emits a verbal response, he is termed the speaker and the mediator of reinforcement is the listener. The topography of verbal behavior can take on any form, so long as it meets the functional requirements of the definition. Thus, verbal behavior does not refer only to vocal speech but can also include sign language, written text, and in some cases, problem behavior (Sundberg). Dependent variables in a verbal behavior analysis are verbal operants, defined as the functional relation between a class of behavior and the controlling environmental variables (Skinner, 1957). Verbal operants are differentiated from one another based on the antecedents and consequences that evoke and maintain a particular class of responding. Four of these 26 operants are especially relevant to teaching young children with autism to emit conventional response topographies (Lerman et al., 2005). Mand. The mand is an operant under the control of MOs and specific reinforcers. Essentially, the mand allows an individual to ask for things he wants (Sundberg, 2007). When a child with autism walks past a classroom and sees his favorite toy train through a window in the door, he may go into the classroom to get it. If the door is locked, the child may look to an adult and mand for the door to be opened. In this case, the train and the locked door functioned as EOs in control of a mand such as “open door!” The value of adult assistance as a reinforcer increased because the door was locked and the probability of behavior that has been followed by an adult unlocking a door also increased. Tact. The tact is an operant under the control of a nonverbal discriminative stimulus and maintained by generalized conditioned reinforcers, such as praise (Sundberg, 2007). Tacts name items, actions, or events that a speaker contacts through any of the five senses. Many children with autism develop an extensive tact repertoire despite the absence of other verbal operants (Sundberg). For example, a child may learn to emit the names of favorite foods upon seeing pictures and can therefore tact pictures of a favorite meal such as pancakes, blueberry, and syrup. Most listeners within the verbal community respond to this type of behavior with a response such as “yes, those are pancakes.” The pictures are nonverbal stimuli that occasion the verbal response “pancakes, blueberry, syrup” and listeners reinforce the response by praising the child’s accurate tacts. Echoic. The echoic is an operant under the control of a verbal discriminative stimulus and generalized conditioned reinforcers (Sundberg, 2007). The beginning, middle, and end of the echoic response must be the same as the model. This is referred to as point-to-point 27 correspondence. Additionally, the response must have formal similarity to the model, which means that the response is emitted using the same sense modality. Many children with autism engage in echoic behavior under faulty stimulus control. For example, when a teacher holds two preferred items in front of a student and tells the student to “pick one,” the student might echo the response “pick one” but fail to select a preferred item to play with. Similar to tacts, echoics are maintained by consequences such as praise or the correspondence to the model provided by the listener. Intraverbal. The intraverbal is an operant under the control of a verbal discriminative stimulus and generalized conditioned reinforcers (Sundberg, 2007). The intraverbal is differentiated from the echoic as it does not have formal similarity to the verbal discriminative stimulus. Teachers often provide hints as prompts that lead to intraverbal behavior. For example, when a teacher shows a child a picture of a dog and says “it’s not a cat, but a...,” the child’s response “dog” is an intraverbal. Applications to Children with Autism Early applications of ABA to teach language to children with autism transferred principles of operant conditioning such as stimulus control and reinforcement to the psycholinguistic structure emphasized in research and practice in the 1960s (Hewett, 1965; Lovaas et al., 1966). Reinforcement was systematically applied to sequential developmental targets including looking toward teachers, following directions, demonstrating receptive identification (e.g., “touch head”), imitating a teacher’s vocal response, and eventually naming objects and pictures (Lovaas, 1977). This developmental sequence is often referred to as a receptive-expressive skill hierarchy; receptive language skills are taught first followed by expressive language skills (LeBlanc, Esch, Sidener, & Firth, 2006). 28 The logic of the traditional approach was that potent rewards could be used to quickly teach children a broad repertoire of linguistic targets (e.g., auditory comprehension, imitation, labels). Purveyors hoped that once a target was acquired, children would be able to emit the target in natural contexts with little to no additional teaching (Lovaas, Koegel, Simmons, & Stevens Long, 1973). That is, if a child could tact “juice” when asked to label a cup full of orange liquid during training, the child would be able to mand for juice when thirsty and in the presence of a listener who could provide the beverage. Research has since shown that many children with autism – especially those with severe language impairment – are not able to generalize the responses acquired during training (Bodfish, 2004; LeBlanc et al., 2006). This issue is thought to be a result of two phenomena related to generalization. First, verbal operants proved to be functionally independent; for example, a topographical response often needs to be independently trained as a tact and mand (Hall & Sundberg, 1987; Lamarre & Holland, 1985; Twyman, 1996). Second, many participants did not emit trained operants under novel stimulus conditions (Bodfish, 2004; LeBlanc et al., 2006). Each environment in which a response would be used required additional training. Language training programs based on Skinner’s (1957) analysis of verbal behavior emerged in the 1980s to address the problems with generalization following traditional behavioral interventions for children with autism (LeBlanc et al., 2006). Whereas traditional communication training taught children according to the receptive-expressive developmental sequence established in psycholinguistics, verbal behavior training utilized the developmental sequence theorized by Skinner. The verbal behavior approach teaches individuals to mand prior to teaching other verbal operants (Drash, High, & Tudor, 1999; Sundberg & Michael, 2001). 29 The logic of early mand training was that children would learn target behaviors faster and be more likely to generalize responding because the targets allowed a child to regulate his environment (LeBlanc et al., 2006). Additionally, because mand training was presumed to be highly reinforcing, collateral behaviors such as responding to others and compliance with instruction could be affected (Koegel, O’Dell, & Dunlap, 1988). Although terminology varies across disciplines (i.e., mands are referred to as requests or spontaneous initiations), targeting mands before other verbal operants is consistent with recent empirical literature from the fields of developmental psychology, early childhood special education, and communication disorders (Halle & Meadan, 2007; Keen, Sigafoos, & Woodyatt, 2001; Tager-Flusberg et al., 2009). In order to effectively teach mands, an interventionist must identify preferred items, events, or activities and then capture or contrive a MO in order to evoke a response from the participant (Michael, 1982; 1988; 2000; 2007; see also Greer & Ross, 2008; Iwata, Smith, & Michael, 2000; Laraway et al., 2003; Sundberg & Michael, 2001). For example, deprivation from a highly preferred toy could increase the value of obtaining the toy for a young child. The placement of the toy on a high shelf could evoke climbing behavior by a child who sees that toy on top of the shelf. Alternatively, if a tall listener were present and the child had learned to say “get toy” while pointing to the item, the listener could mediate the delivery of the toy. Deprivation is a common EO that evokes behavior; the converse of deprivation is satiation, which is an AO that suppresses behavior as it temporarily decreases the value of an item as a reinforcer (Laraway et al., 2003). In the example above, if a child had recently played with the toy before it was placed on a shelf, she may be less likely to climb the shelf or mand for the toy because of the abative effects of the AO. The manipulation of MOs is therefore a critical 30 component of mand training for children with autism (Greer & Ross, 2008; Sundberg & Michael, 2001). Previous Research Though not used to teach language to children with autism, Hart and Risley (1975) developed a methodology for capturing naturally occurring EOs to increase the language production of 11 economically disadvantaged preschoolers. Instead of immediately helping children who demonstrated the need for assistance, researchers trained teachers to wait for a child to initiate some type of social interaction. The teacher could then expand the interaction and teach the child to use language to mand for adult assistance. For example, if a child could not reach a preferred toy, the experimenters taught teachers to wait for the child to look at them and then prompt the child to mand for help obtaining the toy. A MBD across listeners was used to demonstrate that language use increased as a function of the listener capturing the EO and waiting for the child to emit a vocal mand. Halle, Marshall, and Spradlin (1979) extended the strategy of capturing naturally occurring EOs to teach teenagers with mental retardation to mand for food items at breakfast. Six participants who were accustomed to receiving meal trays and food items by progressing through a line were exposed to an experimental condition involving a 15-s time delay prior to the delivery of each item. Mands for the item prior to the elapsed time delay resulted in a more rapid acquisition of the item. Two of the 6 participants emitted target mands when in the breakfast line and generalized the responses to the lunch line following implementation of the time delay procedure. The remaining participants required more intensive training including the delivery of a vocal model by cafeteria staff at the end of the 15-s time delay (e.g., “tray please”). All participants eventually acquired and generalized target mands. 31 Capturing naturally occurring MOs has since been applied to teaching language to children with ASD and continues to be an effective method for incorporating learning opportunities into regularly scheduled activities (Charlop, Schreibman & Thibodeau, 1985; Hancock & Kaiser, 2002; Koegell, O’Dell, & Koegel, 1987; Yoder & Stone, 2006). The child’s motivation to emit the target response combined with a vocal model delivered by the interventionist leads to rapid acquisition of the target responses for many children. However, as originally established by the traditional language training programs (e.g., Lovaas et al., 1973), children with severe autism require more learning opportunities than natural events may allow (Sundberg & Michael, 2001). Explicit manipulation of MOs to increase learning opportunities differentiates a verbal behavior approach from other procedures used to teach language to children with autism (LeBlanc et al., 2006). Hall and Sundberg (1987) provide one of the first examples of a method for contriving MOs to increase the opportunities for 2 deaf teenagers with mental retardation to mand for preferred items using sign language responses. The experimenters first taught participants to complete a chain of behaviors leading to a preferred consequence (e.g., operating a vending machine). After training several chains, the experimenters contrived an EO by withholding an item needed to complete the chain (e.g., a quarter). Withholding an item altered the value of that item as a consequence and therefore functioned to evoke a response that had obtained that consequence in the past (e.g., approaching or look toward adults). The experimenters quickly prompted participants to emit a sign language response to mand for the missing item. By contriving EOs, this investigation showed that intervention agents could supplement naturally occurring MOs to increase learning opportunities for individuals with severe disabilities. 32 Bondy and Frost (1994) developed the Picture Exchange Communication System (PECS) to extend the strategy of contriving EOs to teach children with autism who had not demonstrated speech capacity. Children were taught to mand for preferred items by placing a picture of the item into a listener’s hand. Using the PECS, EOs are contrived by briefly withholding preferred items prior to mand training. During training, interventionists hold a preferred item in the visual field of the child and place a picture of the item on a table and directly in front of the child. When a child initiates an interaction by reaching for the item, a second interventionist standing behind the child manually guides the child’s hand to pick up the picture of the item and hand it to the listener. The listener then responds by looking at the picture, saying “oh, you want the toy”, and handing the item to the child. Following PECS training, many children develop a mand repertoire sufficient to communicate with a variety of listeners in multiple environments and some children acquire vocal verbal behavior (Yoder & Stone, 2006). Drash and colleagues (1999) used an approach similar to the PECS procedures to contrive EOs and shape vocal mands emitted by very young children with autism. When the child reached for an item, the experimenters asked a question such as “What do you want?” to prompt a vocalization. Initially, any vocalization other a scream was reinforced. As the child met predetermined criteria only successive approximations to the target mand were reinforced. All 3 participants acquired target mands in a relatively brief period of time. This investigation, along with previous mand training research (Bondy & Frost, 1994; Hall & Sundberg, 1987), offer systematic methods for teaching individuals with severe language impairment to mand for items using a topography that could be understood by a variety of listeners in an individual’s environment. Unfortunately, early analyses of mand training did not employ experimental designs that allowed for the discovery of functional relations. 33 Charlop-Christy, Carpenter, Le, LeBlanc, & Kellet (2002) addressed the limitation of experimental control in previous research by teaching the PECS to 3 children with autism using a MBD across participants. The results of this investigation have important implications as all participants rapidly acquired target mands and several behaviors not explicitly targeted were also affected following the implementation of training. In addition to acquiring vocal speech following the PECS training, all participants demonstrated an increase in several social behaviors such as responding to initiations of others, joint attention, and making eye contact. Further, all participants demonstrated a decrease in problematic behavior. Though previous research had established an inverse relation between mand acquisition and problem behavior (Carr & Durand, 1985), this investigation was the first study to document this effect when mand acquisition, as opposed to problem behavior, was the targeted response. Jennett, Harris, and Delmolino (2008) provided another experimental demonstration of vocal mand acquisition following manipulation of EOs. The experimenters used MPDs across participants to compare mand training to traditional language training for 6 children with autism. Jennett and colleagues compared a child’s rate of mands when an EO was present (i.e., mand training) and when it was not (i.e., traditional). During mand training, the EO was contrived for each trial by allowing participants to select from a number of common toys they could not access such as a preferred item in a container the child could not open. During traditional training, the experimenter randomly selected toys for each trial and asked the participant “what do you want?” During both conditions, the instructor held the item in view of the participant and provided a vocal model of the target response (e.g., “I want crayon”). Results of the investigation by Jennett and colleagues (2008) have several important implications for teaching verbal behavior to children with autism. First, 5 of the 6 participants 34 demonstrated higher rates of manding during the mand training versus traditional condition. The final participant did not acquire mands under either condition. Second, though not empirically controlled all participants demonstrated higher rates of eye contact during the traditional versus mand training condition. Third, 4 of the participants demonstrated reduced levels of problem behaviors during the mand versus traditional training condition; problem behaviors were also not experimentally controlled. Though these results demonstrate the effectiveness of mand training procedures, they need to be interpreted with caution as all participants demonstrated some vocal mand capacities prior to training. It is not known if the outcomes would be similar for individuals who had not yet demonstrated the capacity for vocal speech. Though mand training can be effective, several potential challenges may prevent the establishment of a functional relation between the desired response and environmental variables that evoke and maintain that response. First, EOs with sufficient strength to evoke initiating behaviors can be difficult to contrive or capture (Sundberg & Michael, 2001). Second, consistent functional reinforcers can be extremely challenging to identify for children with severe autism (Jennett et al., 2008). Third, it can be difficult to discern whether a child with autism has the capacity to emit the desired topography (e.g., vocal speech). Fourth, prompts must be inserted into teaching trials to teach the desired mand topography; prompts must be faded over time to ensure the child does not become dependent on the prompt to emit the response (Jennett et al., 2008). An error in any of the above processes can lead to ineffective mand training. Postintervention error analysis can be time consuming and leaves the child without the ability to mand until errors have been identified. Thus, procedures that limit the potential for error or reduce the variables responsible for any error that may occur can be extremely valuable when first teaching mands to children with autism. 35 Functional Communication Training Terminology and Conceptualization FCT is a specific type of mand training whereby an interventionist identifies the effect a problem behavior has on the environment and teaches a child to emit an alternative response that has the same effect as the problem behavior (Herzinger & Campbell, 2007). For example, if a child with autism flops to the floor and yells when instructed to complete a disliked task and the effect this behavior has is that it delays or eliminates starting the task, an alternative response could be teaching the child to ask for a break. Taking a break has the same effect on the environment as problem behavior: removal of task demands. In this case, the term function is used to describe the effect a behavior has on the environment (Hanley et al., 2003). The function of problem behavior in the preceding example is to escape or avoid task demands. The example above offers a simplified version of FCT. Several components need further explanation including how the function of the problem behavior is determined and how alternative responses are selected and taught to a student. The identification of function is discussed below and the procedures for teaching are quite similar to mand training procedures discussed previously. However, by identifying the function of problem behavior prior to mand training, FCT offers a systematic approach to capturing or contriving MOs and identifying reinforcers for an individual child: two of the potentially challenging aspects of mand training (Jennett et al., 2008). Functional behavior assessment. A functional behavior assessment (FBA) is an operant assessment methodology used to identify the function of problem behavior (Herzinger & Campbell, 2007). The purpose of conducting a FBA is to obtain information about 36 environmental events that evoke and maintain problem behavior. The information can be used to develop an intervention that is likely to be effective for an individual student. The example of the student flopping on the floor when tasks are assigned may be helpful in considering the value of information derived from FBA. Teaching the child to ask for a break, as in the example above, would only be functionally equivalent to flopping and yelling if escape from task demands were the function of the problem behaviors. Alternatively, if the function of the problem behavior were to obtain adult attention, teaching her to mand for a break is not likely to be an effective intervention. By identifying the function of behavior prior to intervention, an intervention agent can systematically manipulate antecedents and consequences that are likely to change behavior (Carr & Durand, 1985; Hanley et al., 2003; Iwata et al., 1982/1994). There are three distinct methods for conducting a FBA (Herzinger & Campbell, 2007). First, caregivers can be asked to complete an informational checklist or questionnaire that describes what happens immediately before and after the behavior. A second approach to FBA is a direct observation of the student in her natural environment. A coding system is used to identify the events that occur just before and after an instance of the problem behavior. The goal of both the indirect checklist and direct observation is to obtain a sample of instances of problem behavior and to look for patterns in the temporal sequence of events that may provide clues as to the function of the response. A third approach to FBA is to conduct an experimental functional analysis (FA). Whereas indirect checklists and direct observations reveal correlations, a FA employs a SSD to reveal functional relations between the problem behavior and environmental variables (Hall, 2005). A FA is the most accurate FBA procedure for identifying the function of problem behavior (Hall; Herzinger & Campbell; Thompson & Iwata, 2007). 37 Functional analysis. Hanley and colleagues (2003) define a FA of behavior as any “...direct observation and measurement of behavior under test and control conditions in which some environmental variable is manipulated” (p. 149). The FA simulates various conditions that can maintain or increase problem behavior in the natural environment. These are called test conditions. The FA also simulates a condition in the natural environment that is likely to be void of problem behavior, such as play or snack time. This is called a control condition, though it is sometimes referred to as a play condition because it simulates a free play situation for the child (Iwata et al., 1982/1994). The test conditions of the FA assess for behavior maintained by social positive reinforcement, social negative reinforcement, and automatic reinforcement (Iwata et al., 1982/1994). The term “social” refers to reinforcement mediated by the behavior of another person; problematic behavior maintained by social reinforcers is therefore, verbal behavior. Automatic reinforcement indicates the behavior is maintained by the sensory stimulation it provides. Many children with autism engage in repetitive behavior such as rocking, jumping, or hand-mouthing that is often maintained by automatic reinforcement (Bodfish, 2004). Each test condition within the FA combines a source of potential reinforcement with related MOs and discriminative stimuli (Iwata et al., 1982/1994). If the problem behavior is maintained by the consequences within a specific condition, the behavior should occur multiple times within a FA session for that condition. Across FA sessions, the frequency of problem behavior during that condition should increase or at least be maintained. The most common social positive reinforcement condition in FA tests for contingent attention as a maintaining consequence (Hanley et al., 2003). Problem behavior maintained by attention in the natural environment is most likely to occur when the student is deprived of 38 attention for a certain period of time (Carr & Durand, 1985). Therefore, the student is instructed to sit by himself to contrive an EO that increases the value of attention as a reinforcer during the attention test condition of a FA. The experimenter sits on the other side of the room pretending to read or complete paperwork and delivers attention contingent on problem behavior. The most common social negative reinforcement condition in FA tests for escape from task demands as a maintaining consequence (Hanley et al., 2003). Problem behavior maintained by escape in the natural environment is not likely to occur if there are no demands placed on a student, as there would be nothing to escape (Carr & Durand, 1985). Therefore, a test condition for escape combines the delivery of task demands as an antecedent with the contingent removal of task demands as a consequence. The automatic reinforcement condition tests for behavior maintained by sensory stimulation. Problem behavior maintained by sensory stimulation could occur at any time, though it is most likely when the child experiences minimal stimulation in the environment (Iwata et al., 1982/1994). Therefore, a student is placed by herself in an empty room during this test condition. Behavior that occurs during this condition is likely evoked by the barren environment and assumed to be maintained by the stimulation it provides. The control condition is used to show that the experimenter has identified the variables responsible for producing the behavior (Iwata et al., 1982/1994). For example, demands are not delivered to remove the EO for problem behavior maintained by escape. Attention is constantly delivered to create an AO for problem behavior maintained by attention. Finally, interesting toys are included in the control condition to provide alternative sources of stimulation. Behavior should occur at low levels during the control condition as the variables likely to evoke and maintain the response are not present. 39 As the participant is exposed to the various conditions, his behavior will often reveal the environmental variables responsible for evoking and maintaining the problem behavior (Iwata et al., 1982/1994). Based on principles of behavior, the strength of a response will either be maintained due to reinforcing consequences or it will not as the consequences are not reinforcing and the behavior is placed on extinction. Therefore, if behavior occurs at a higher rate in a single condition compared to all other conditions, the problem behavior is maintained by the consequence manipulated during the condition with high responding (Hanley et al., 2003). In other words, the function of the problem behavior is to obtain whatever consequence was delivered during the condition with a high level of responding. If multiple test conditions show a high rate of responding, the behavior has multiple functions (Hanley et al.). If responding occurs at a high rate during the control condition, it can be indicative of idiosyncratic functions that were not captured during the FA or an automatic reinforcement function (Hanley et al.). The benefit of a FA is that the information is more likely to be accurate than information derived from other types of FBA. Consider again the student who flops and yells when a task is assigned. If the event is always preceded by an assigned task and followed by temporary or complete task avoidance in addition to attention from an adult in the classroom, direct observation would reveal that escape from task demands and attention reliably follow problem behavior. It is only through experimental manipulation where each consequence is delivered independently and behavior is measured over several sessions that the consequence responsible for maintaining the behavior can be discovered. Previous Research Iwata and colleagues (1982/1994) conducted one of the first systematic FAs used as an assessment prior to the treatment of self-injurious behavior (SIB) for 9 children between the ages 40 of 3 and 17. The assessment included one control condition and the three typical test conditions (i.e., obtain attention, escape demands, and alone). Six to 13 sessions for each condition were individually administered to each participant using an ATD. The results showed that 6 of the 9 participants reliably engaged in SIB during a specific test condition. The three remaining participants did not demonstrate differential levels of responding across conditions. The investigation by Iwata and colleagues (1982/1994) was seminal for several reasons. First, it showed that an individual’s SIB did not randomly occur, but was a result of specific environmental variables immediately preceding and following the behavior. Second, as the function varied across participants, the results showed that a standard intervention for treating self-injury may be effective for some individuals yet contraindicated for others. Third, it explicitly described an empirical methodology for identifying variables that evoke and maintain a problem behavior. In another approach to FA, Carr and Durand (1985) manipulated the level of task difficulty and the amount of adult attention to identify antecedent variables that functioned to evoke the self-injurious and aggressive behavior of 4 children with cognitive impairment and developmental delays. By manipulating MOs (i.e., attention and work difficulty), the investigators could infer the maintaining consequence based on identifying the condition with the highest rate of problem behavior. That is, when behavior occurred most frequently during an easy work and low attention condition, delivery of attention was hypothesized to be the maintaining consequence. Conversely, when behavior occurred most frequently during a high attention and hard work condition, removal of task demands was hypothesized to be the maintaining consequence of problem behavior. 41 Unlike the investigation conducted by Iwata et al. (1982/1994), the procedure used by Carr and Durand (1985) inferred reinforcers based on the presence of a MO that evoked problem behavior. Because consequences were inferred as opposed to empirically derived, this methodology is more likely to lead to an erroneous identification of the function of a behavior (Hanley et al., 2003). The experimenters nevertheless demonstrated the utility of the procedure as the first application of FCT was successfully implemented for all participants following the assessment. All 4 participants in the Carr and Durand (1985) investigation demonstrated a reduction in problem behavior to near zero levels during the FCT condition though continued to engage in problem behavior during control conditions. These results showed that children who already had vocal verbal repertoires could learn to emit vocal mands for specific actions (e.g., assistance, attention) and these mands could replace previous problematic mand topographies. However, it was not clear whether the extinction of problematic responses or reinforcement for competing responses led to the change in behavior. Wacker and colleagues (1990) conducted a component analysis of FCT to determine which behavioral principle (i.e., extinction or reinforcing a competing response) was responsible for changes in behavior. When the problem behavior and the competing response were followed by reinforcement, both behaviors were maintained. However, by administering a time-out following problem behavior, experimenters could suppress problem behavior and maintain only the alternative response at high levels. Thus, both components of FCT were necessary to decrease problem behavior and increase a competing response. This has important implications for mand training for individuals reliant on gestural behavior because it shows that established responses continue to occur if reinforcement for those responses is available. As some listeners 42 in a child’s environment are likely to respond to and reinforce gestures, the child may be less likely to acquire and reliably emit a new response if the gesture is not placed on extinction during mand training. Marcus, Vollmer, Swanson, Roane, & Ringdahl (2001) extended previous FAs by including a social positive reinforcement test condition that manipulated access to tangible items as a consequence. This condition was added to the three test conditions and control condition used in the Iwata et al. (1982/1994) investigation. In the tangible condition, deprivation from preferred tangible items such as food and toys replaced deprivation from attention as the EO and access to items contingent on the target response replaced the delivery of attention as a consequence. Similar to the attention condition, the tangible condition assessed for social positive reinforcement as a maintaining consequence because the experimenter delivered the preferred tangible item. Thus, problem behavior that occurred consistently during the tangible condition would also be a mand for the item delivered following the behavior. The results of the FA conducted by Marcus and colleagues (2001) revealed that 4 of the 8 participants engaged in aggressive behavior maintained by access to tangible items. Three of the participants aggressive behavior was maintained by escape from task demands and the final participant appeared to engage in aggressive behavior maintained by peer attention, though this was not directly tested during the FA. An important contribution of this investigation was that the majority of participants engaged in aggressive behavior maintained by social reinforcers. These results confirmed the assumptions of Carr and Durand (1985) that aggression was often a type of mand for children with developmental disabilities. Systematic reviews of multiple FA cases also suggest problem behavior frequently functions as a type of mand (Hanley et al., 2003; Iwata et al., 1994). Iwata and colleagues 43 analyzed results from 145 previously conducted FAs to identify common trends in the function of problem behavior emitted by children with developmental disabilities. In 68% of the cases, problem behavior was maintained by a social reinforcer. Hanley and colleagues conducted a more recent review, analyzing the results of 536 graphed FAs. The researchers found that at least 80% of the FA cases revealed problem behavior maintained by social reinforcers. The reviews of FA outcomes are important as they show that problem behavior functions as a mand more often than it does not. Thus, FCT could likely be an effective intervention for many children with developmental disabilities who engage in problematic behavior. Additionally, it might be possible to prevent the development of those problem behaviors if children acquired alternative mands at an earlier age. Since the inception of FA as a pretreatment approach to assess SIB and aggression, researchers have replicated and extended this methodology to successfully develop treatments for bizarre vocalizations (Durand & Crimmins, 1987), noncompliance (Reimers et al., 1993), stereotypy (Mace, Browder, & Lin, 1987), hair pulling (Miltenberger, Long, Rapp, Lumley, & Elliot, 1998), tantrums (Vollmer, Northup, Ringdahl, LeBlanc, & Chauvin, 1996), elopement (Piazza et al., 1997), pica (Piazza, Hanley, & Fisher, 1996), and other forms of problematic behavior (Hanley et al., 2003). The majority of participants involved in these investigations had a developmental disability and engaged in problem behavior that functioned as a type of mand (Hanley et al.; Iwata et al., 1994). Therefore, FCT is an intervention that can be applied in many situations to children with developmental disabilities who engage in problematic behavior. The FCT procedures have been systematically replicated across diverse groups of participants, research sites, and forms of problematic behavior (e.g., Casey & Merical, 2006; Dunlap, Ester, Langhans, & Fox, 2006; Keen et al., 2001; Wacker, Harding, & Berg, 2008). FCT 44 is effective because interventionists can recreate an evocative event (i.e., MO and discriminative stimuli) that typically leads to problematic behavior. Before problem behavior has a chance to occur, interventionists can prompt an individual to emit an acceptable alternative behavior and provide the same consequence that was previously delivered following problem behavior. If, on the other hand, the antecedents and consequences that produce problem behavior are not known, interventionists are left to their “best guess” of all possible environmental manipulations that could sufficiently change the target behavior(s) (Horner & Carr, 1997; Iwata et al., 1982/1994). That is, each potentially reinforcing consequence and evocative antecedent would need to be paired in a trial and error approach until the correct combination sufficiently alters behavior. FCT has drastically improved the overall effectiveness and efficiency of interventions targeting problem behavior functioning as a mand over the past 25 years (Hanley et al., 2003; Horner & Carr, 1997; Tiger et al., 2008). This technology is so widely recognized as an effective approach to understanding the communicative behavior of individuals with developmental disabilities that it has been included in many major reviews of evidence-based practices (NRC, 2001; NSR, 2009), federal legislation, and educational policy (Individuals with Disabilities Education Improvement Act, 2004). Despite this highly effective technology for teaching acceptable and conventional mand topographies, children must first develop frequently occurring problematic behaviors before receiving FCT. Children who demonstrate severe deficits in verbal behavior but do not demonstrate severe problem behavior are not candidates for one of the most empirically validated methods of teaching verbal behavior. Future Research Recent research has extended the use of pretreatment FA in a manner that may inform future applications of FCT prior to the establishment of problem behavior. Langdon and 45 colleagues (2008) found that children with developmental disabilities often emit precursor behaviors: innocuous or less severe forms of problem behavior that, when not reinforced, reliably preceded the occurrence of severe problem behaviors such as self-injury and aggression. The behaviors were functionally equivalent mands for the removal of work demands. Interestingly, precursers occurred reliably across conditions though problem behavior only occurred when the precursor behavior was not reinforced. A response class hierarchy had formed wherein the member of the response class requiring the least effort was emitted and only followed by a high effort response if the first response did not yield reinforcement. Essentially, the removal of work demands following precursor behavior could prevent the occurrence of other behaviors in the response class. The FA of precursor behaviors (e.g., Langdon et al., 2008) has important implications for applications of FCT to teach mands to children with autism without an emphasis on problem behavior. Based on reinforcement history, a series of functionally equivalent behaviors may develop into a verbal response class for a child (Halle et al., 2004). For example, a nonvocal child with autism might walk toward and look at his mother to be picked up when at home. At school, this same behavior may receive a different consequence, such as a redirection to his assigned seat to complete a task. The child then begins to whine, as whining can often produce additional attention. However, the attention may not be physical and therefore, it does not produce a consequence that is similar to being picked up at home. The child might then hurt himself or another individual as dangerous behavior is frequently followed by some type of physical intervention. Walking toward and gazing, whining, and aggression are all members of the same response class in the previous example and each is a mand for physical attention. Walking 46 toward and gazing gestures might be maintained in the home environment where an attentive and experienced care provider reliably reinforces those gestures. However, in environments that include an inattentive or inexperienced care provider who does not reliably reinforce gestures, a child may develop precurser and problematic behaviors to mand for preferred items, events, and activities. Once the response class described above is constructed, a child might develop a response class hierarchy wherein gestures are emitted in some situations and precursor or problematic behaviors are used in other situations (Langdon et al., 2008; Shabani, Carr, & Ingeborg Petursdottir, 2009). If this were the case, teaching conventional mands may decrease the value of engaging in problematic behaviors. That is, the value of an item, action, or event as a reinforcer is reduced because a child has already obtained the preferred consequence. Most importantly, this approach would only require a FA of gestural behavior, which means it could be used as a proactive approach to mand training and may prevent the development or worsening of problem behavior. Ferreri & Plavnick (in press) recently extended the FA methodology to gestural behavior. Results indicated gestural responding was maintained by access to preferred items for one participant and by the delivery of information about novel stimuli for a second participant. This intersubject variability exemplified the potential need for subtle differences in the evocative events and consequences used to teach mands to children. However, interventions based on this assessment were not formally developed and reported; as such, the approach could not be evaluated for its educational effectiveness as an assessment procedure. Implementation of function-based interventions may support or refute the need to functionally assess idiosyncratic verbal behavior and may also inform necessary variations to the design of the assessment. 47 Several of the challenges related to mand training might be addressed by a pretraining FA of gestural behavior. Interventionists may be more accurate when contriving MOs and identifying reinforcers for mand training following a FA of gestural behavior. Thus, the mand training challenges related to contriving MOs or identifying reinforcers would be less likely to lead to ineffective training. Such an approach could also be beneficial to identifying whether a previously nonvocal child is able to emit vocal verbal behavior. A preintervention FA could help ensure optimal antecedents and consequences were used to evoke mands, thus suggesting the absence of vocal behavior is more likely related to an individual’s vocal capacity rather than motivative variables. Video Modeling Conceptualization Video modeling (VM) is a procedure designed to teach new skills to an individual by showing a video of a model engaging in the target behavior and then providing the observer with an opportunity to engage in the same behavior (Darden-Brunson, Green, & Goldstein, 2009). Though the mechanisms that make VM effective have not been empirically identified, current research credits Bandura’s (1977) social learning theory as the likely explanation (DardenBrunson et al.). In order to learn through observation, Bandura posited that an individual must be able to attend to the behavior of others and outcomes of those behaviors, retain the behavior for future use, reproduce the behavior under stimulus conditions similar to those observed, and be motivated to engage in the target response. Based on a social learning conceptualization (Bandura, 1977), VM is thought to work when a child is interested in the model and has the cognitive capacity to store and retrieve the sequence of events that occur in the video. Interest in the physical characteristics of the model is 48 hypothesized to increase the child’s likelihood of paying attention and to increase the child’s motivation to imitate the model at a later time (Darden-Brunson et al., 2009). Another aspect of VM thought to be influential for individuals with autism is a preference for attending to television or computer screens over attending to live models (Charlop-Christy, Le, & Freeman, 2000). Though this account relies on some mentalistic concepts rather than offering a comprehensive behavioral analysis (see Deguchi, 1984 or Greer, Dudek-Singer, & Gautreaux, 2006 for a behavioral account of observational learning), it has provided the foundation for a number of VM research studies; these studies show that VM can be an extremely effective method for teaching new skills or facilitating improved performance of already acquired skills by children with ASD. Previous Research Haring, Kennedy, Adams, and Pitts-Conway (1987) compared the effects of grocery shopping training in a single grocery store to VM with a variety of stores depicted in the videos. In this first application of VM to individuals with autism, Haring and colleagues discovered training led to skill acquisition without generalized responding whereas VM led to generalized skill acquisition across environments and without prompts for all participants. VM proved to be a more efficient intervention than training in a single environment. Charlop and Milstein (1989) extended VM to younger children with autism and found that high functioning boys with autism increased and maintained their conversational statements following exposure to VM conditions. Additionally, prompts did not have to be delivered and both participants demonstrated generalized responding across settings, conversational partners, and stimuli. Numerous replications of VM interventions have demonstrated the generality of this procedure for teaching a variety of skills to children with mild or moderate autism (Charlop- 49 Christy et al., 2000; Gena, Couloura, & Kymissis, 2005; Hine & Wolery, 2006; Lowy Apple, Billingsley, & Schwartz, 2005; Paterson & Arco, 2007; Reagon, Higbee, & Endicott, 2006; Sherer et al., 2001). However, very few research studies have investigated the benefits of VM for teaching mands to children with severe autism. Wert & Neisworth (2003) used video self-modeling to teach children with autism who emitted few spontaneous requests (i.e., mands) to increase requests during the school day. Video self-modeling is a type of VM procedure where careful editing allows adults to prompt the participant to emit the target behavior and then remove prompts from the video once it is captured. Participants viewed videos of themselves emitting target mands prior to going to school. All 4 participants demonstrated an increase in mands following VM and maintained mands at a higher level than baseline up to 6 weeks following the termination of VM. However, because all participants acquired mands during training periods when the video clips were created (that is the only way video self- modeling can occur), it is not clear whether response training, VM, or both caused the change in behavior. Nikopoulos and Keenan (2007) examined the effects of VM on social initiations (i.e., mands) emitted by 2 children with severe autism. Participants viewed a 25- to 37-s video clip of a typically developing 10-year-old boy grabbing an adult’s hand and saying “let’s play” to the adult listener. The participant was then immediately given an opportunity to engage in the target behavior under conditions with high stimulus similarity to the video clip (i.e., same room, materials, and adult listener). Grabbing an adult’s hand or saying “let’s play” were scored as occurrences of the target response. Both participants demonstrated decreased latency to social initiations though it was not clear whether either participant had emitted the vocal response during any of the trials (i.e., could have only grabbed adult’s hand). It is possible that VM led to 50 the acquisition of a potentially ambiguous mand rather than a mand that more specifically identifies the reinforcer. Future Research Despite positive effects of VM for children with autism, the majority of research examines complex social and play skills for children with mild or high functioning ASD. There are few investigations of this procedure for children with severe autistic symptoms. From an applied perspective, more research is needed to better understand the effects of VM on skill acquisition and performance for children with severe autism. Since children with severe autism tend to demonstrate deficient mand repertoires, this may be an ideal target behavior for future research. VM provides a method for modeling the target response without inserting the model into the natural antecedent-behavior-consequence relation as is typically done in mand training (Greer et al., 2006; Jennett et al., 2008). Research literature on VM for children with autism has yet to identify the behavioral mechanisms that lead to positive outcomes (McCoy & Hermansen, 2007; Rayner et al., 2009). A common assumption is that certain physical characteristics of the model impact the behavior of the observer (Buggey, 2005). This has led to research examining whether adults are better models than peers or if the child himself is the best model for VM interventions (McCoy & Hermansen). Other researchers have questioned whether the point of view of the modeled sequence impacts the outcomes of the procedures (Hine & Wolery, 2006). When the sequence is filmed from a first person perspective, the observer only sees listeners and not the person whose behavior she is to imitate. This is contrasted to the more traditional third person perspective where the observer sees all participants within a scene. 51 The above questions stem from Bandura’s (1977) theoretical framework, which suggested that an observer would be more likely to attend to and imitate a model if it is similar to him or herself. These uncertainties can be categorized as structural explanations in that each one emphasizes the possibility that structural aspects of the model (or modeled sequence) affect the probability of the observer engaging in the behavior. Less attention has been paid to another tenant of Bandura’s theory, which was that observational learning occurs when the observer views the behavior of others and the outcomes of those behaviors. Future research is needed to better understand those mechanisms that lead to positive outcomes following VM for children with autism. No previous research has examined a functional conceptualization of VM; therefore, experimental analyses of the functional components of VM may provide important information currently absent from the extant literature. Purpose of the Present Investigations The ability to produce functional communicative responses by the age of five is an important outcome indicator for children with autism (Howlin et al., 2004; Tager-Flusberg et al., 2009). From an ABA perspective, the acquisition of functional communication means children are able to mand in a variety of settings and for a variety of consequences (Sundberg, 2007). Many children with autism can learn to mand when EOs are contrived and the target response is prompted by an interventionist (Charlop-Christy et al., 2002; Jennett et al., 2008). However, some children do not acquire, generalize, or maintain mands following evidence-based mand training procedures (Dawson, 2008; NSR, 2009). Research is needed to better understand why some children with autism do not respond favorably to otherwise effective mand training procedures (NSR, 2009). It is possible that some treatments are simply not going to be effective for certain individuals. Another possibility is that 52 manipulable components of interventions can lead to positive outcomes if intervention agents knew ahead of time which components needed to be altered for a specific child and in what ways. Research that addresses these intersubject differences in outcomes can inform the selection and implementation of interventions for children with autism. Research Questions Research Question 1 Children with autism often rely on idiosyncratic behaviors to mand for preferred items, events, activities, or the removal of aversive stimuli (Sigafoos et al., 2000). FA followed by function-based communication training (i.e., FCT) is an extremely effective procedure for teaching conventional mands, though it is currently used as a reactive approach to decrease problem behavior; an increase in an acceptable alternative mand is a secondary goal of FCT (Carr & Durand, 1985; Tiger et al., 2008). Recent research has shown that the FA methodology frequently used to assess the function of problem behavior can also be used to identify the function of nonproblematic gestural behavior (Ferreri & Plavnick, in press). However, this approach has only been examined with 2 participants and more research is needed to verify the efficacy of a FA procedure to identify the environmental variables that evoke and maintain gestural behavior. Question. What are the functional properties (i.e., covarying environmental antecedents and maintaining consequences) of gestures emitted by a small number of children with autism? Research Question 2 Current mand training procedures seek to contrive or capture MOs and teach new mand topographies through response prompting and reinforcement (Sundberg, 2007). Two potential problems with this approach include the potential for acquisition error that is difficult to 53 determine and prompt dependency. Identifying the function of gestures might allow for the creation of mand training procedures that reduce the potential for acquisition error and function to eliminate some of the reasons for any error that may occur. Prompt dependency may be addressed by employing a VM approach whereby a video clip of the target behavior replaces the use of a prompt within the teaching procedure. This allows for a more natural mand as the interventionist does not have to prompt the child to “say [target mand].” Question. To what extent does the implementation of function-based and nonfunctionbased mand training procedures affect vocal or alternative (e.g., picture exchange) verbal behavior acquired and emitted by a small number of children with autism? Research Question 3 One of the primary challenges for teaching verbal behavior to children with autism is to establish the generality of responding across multiple listeners, environments, discriminative stimuli, and time (Greer & Ross, 2008; Paul, 2008; Sundberg, 2007). Generality is important to ensure a learner is able to use new skills effectively in multiple situations (Bryson et al., 2007). Children that do not generalize verbal behavior may develop problematic behavior to communicate in some environments or may experience limited social interaction with others (Rogers & Vismara, 2008). Given the importance of communication for children with autism, more information is needed about generalized responding as a function of various mand training procedures. Question. To what extent does the implementation of function-based and nonfunctionbased mand training procedures lead to generalized responding? a. To what extent are acquired mands generalized across listener’s following the implementation of function-based and nonfunction-based mand training 54 sessions? b. To what extent are acquired mands generalized across settings following the implementation of function-based and nonfunction-based mand training sessions? c. To what extent are acquired mands generalized across stimulus materials following the implementation of function-based and nonfunction-based mand training sessions? d. To what extent are acquired mands generalized over time (i.e., maintained) following termination of mand training procedures? Research Question 4 Some researchers have suggested listener behaviors (e.g., orienting, following directions) should be taught to children with autism prior to teaching speaker repertoires, including mand training (Greer & Ross, 2008). However, results of empirical research suggest some listener behaviors may be collaterally impacted when children with autism receive mand training (Charlop-Christy et al., 2002; Jennett et al., 2008; Koegel, O’Dell, & Dunlap, 1988; Koegel, Vernon, & Koegel, 2009). More research is needed to identify the potential for mand training to have collateral impacts on other behaviors and to determine the most appropriate sequence for teaching listener repertoires to children with autism within a verbal behavior approach. Question. To what extent does the implementation of function-based and nonfunctionbased mand training procedures targeting vocal or alternative verbal behavior affect listener behaviors correlated to mand acquisition (i.e., orienting to the speaker, following directions)? Research Question 5 Children with autism often emit problematic behaviors to convey basic needs and wants 55 (Tiger et al., 2008; Wacker et al., 2008). When alternative behaviors are taught during FCT, children often demonstrate a decrease in problem behaviors (Dunlap et al., 2006). It is less certain what the direct role of mand training is on problem behavior when that problem behavior is not explicitly targeted. Charlop-Christy and colleagues (2002) provide the only known experimental evaluation of the collateral effects of mand training on problem behavior. More information is needed regarding changes in levels of problem behavior following the implementation of mand training for children with autism. Question. To what extent does the implementation of function-based and nonfunctionbased mand training procedures targeting vocal or alternative verbal behavior affect problem behaviors emitted by participants? 56 CHAPTER 3 METHOD Experiment 1: Assessment The present chapter includes a description of the methodology used to answer the research question identified in Chapter 2. Each of the following components is described in detail: (a) participants, (b) members of the research team, (c) the experimental setting, (d) materials, (e) pre-experimental procedures, (f) operational definitions of the dependent variable and measurement procedures for that variable, (g) the experimental design used to demonstrate control over environmental variables, (h) experimental procedures employed in this investigation, (i) methods for ensuring the reliability and validity of this investigation, and (j) an overview of data analysis methods. Readers should note that data analysis procedures for SSDs were described in greater detail in Chapter 2. Participants After receiving approval from the Institutional Review Board governing research involving human participants, the experimenter distributed information packets to four public elementary and preschools selected based on geography and the fact that the school had a program that provided services to young children with autism. The information packets describing the purpose of the study and the characteristics of participants sought were distributed to building administrators and teachers of classrooms serving students with autism. Teachers who agreed to participate in the research study were asked to refer children no younger than 3 years, 0 months, and no older than 6 years, 11 months who (a) had an autism diagnosis made by a licensed and trained psychologist or psychiatrist and (b) did not spontaneously emit conventional functional communication to initiate interactions with others. Conventional 57 functional communication was defined as vocal speech, sign language, pictorial communication, or using a voice output device to mand or tact. Seven students were initially referred for participation in this investigation. Two of the referred students did not meet eligibility criteria as they frequently initiated interactions with others using vocal mands and tacts. The remaining five students met eligibility criteria, had permission to participate from parents or guardians, and were enrolled for participation in this study. Four of the participants were boys and one was a girl. Prior to the present investigation, each participant was diagnosed with autism by an independent psychologist who used the Gilliam Autism Rating Scale – 2nd Edition (GARS-2; Gilliam, 2006) or the Childhood Autism Rating Scale (CARS; Schopler, Reichler, DeVellis, & Daly, 1980). The GARS-2 is a screening test used to help identify individuals with autism (Gilliam). An autism index is obtained and scores indicate whether the probability of autism is low (69 or less), likely (70-84), or very likely (85 or higher). The GARS-2 was used in combination with direct observation to confirm a diagnosis of autism based on the criteria in the Diagnostic and Statistical Manual of Mental Disorders-IV-TR (American Psychiatric Association, 2000). The CARS is a rating scale used to help diagnose autism in children (Schopler et al.). Scores range from 15 to 60 with a score of 30 or above indicating a diagnosis of autism and a score of 36 or higher indicating severe autism. The Pre-School Language Scale – 4th Edition (PLS-4; Zimmerman, Steiner, & Pond, 2002) is a measure of receptive and expressive language abilities and was administered to each participant prior to the current investigation. Standard scores (SS) on the PLS-4 should be interpreted with a mean of 100 and standard deviation of 15. Severity of language impairment is identified as mild (SS = 78 to 85), moderate (SS = 71 to 77), or severe (SS = 70 or below). 58 Fuller. Fuller was a 4-year, 11-month old male whose score on the GARS-2 was 115, which suggested autism was very likely. Fuller was in his second year in an ECSE classroom and had not previously received any one-on-one services. His auditory comprehension and expressive communication standard scores on the PLS-4 were 50 and 61, respectively. Both scores indicate severe language impairment. Fuller occasionally emitted the vocal responses “flush it” and “T-rex.” These were likely delayed echolalia as they were not emitted in the presence of corresponding stimuli (Schreibman & Lovaas, 1974). Fuller did not attend to others, follow simple directions, or imitate. He occasionally initiated an interaction with a teacher by grabbing her hand or arm. Bailey. Bailey was a 6-year, 6-month old female whose score on the CARS was 55, which corresponds to severe autism. Bailey was in her third year within a public ASD classroom and had previously received some speech and language therapy in small group and occasional one-on-one sessions. Her auditory comprehension and expressive communication standard scores on the PLS-4 were both 50, which indicated severe language impairment. She emitted no other vocal verbal behavior and did not use an alternative communication system. Bailey occasionally attended to others when they spoke to her, imitated others’ behavior if prompted, and followed some simple directions. She did initiate interactions with others by grabbing another person’s hand or taking items out of another person’s hand. Victor. Victor was a 4-year, 7-month old male whose score on the GARS-2 was 115, which suggested autism was very likely. Victor was in his second year in an ECSE classroom and had not previously received any one-on-one services. His auditory comprehension and expressive communication standard scores on the PLS-4 were 57 and 69, respectively. These scores indicated severe language impairment. Victor’s teacher and mother reported that he did 59 not emit vocal verbal behavior at any time. Victor did not attend to others, follow simple directions, or imitate. He frequently initiated interactions with his teacher using gestural behavior. Morris. Morris was a 3-year, 4-month old male whose score on the CARS was 43.5, which corresponds with a diagnosis of severe autism. Morris was in his first year in an ASD classroom and had not previously received any one-on-one services. His auditory comprehension and expressive communication standard scores on the PLS-4 were 74 and 78, respectively. These scores indicated moderate to mild language impairment. Morris’ mother reported that he did emit vocal words when he was 2 years old, though he never had a repertoire larger than two words. Morris did not attend to others, follow simple directions, or imitate. He occasionally initiated interactions with others. Matthew. Matthew was a 4-year, 11-month old male whose score on the GARS-2 was 113, which suggested autism was very likely. Matthew was in his second year in an ECSE classroom and had not previously received any one-on-one services. His auditory comprehension and expressive communication standard scores on the PLS-4 were 50 and 61, respectively. These scores indicated severe language impairment. Matthew’s teacher and mother reported that he did not emit vocal verbal behavior at any time. Matthew did attend to others when they spoke to him, though he did not follow simple directions or imitate. Matthew occasionally initiated interactions with his others by moving and looking toward them. A summary of participant characteristics can be found in Table 3.1. Demographic information, diagnostic information, and standardized language scores are included for each participant. Diagnostic information includes the instrument used, the score, and the rating that corresponds with the score. Language scores include the auditory comprehension and expressive 60 communication scores, severity rating, and age equivalencies from the PLS-4. Experimenter & Observers The experimenter was a third year doctoral student in Special Education at Michigan State University. He had previously completed a research apprenticeship examining the efficacy of a methodology to identify the function of early communicative behaviors emitted by young children with severe developmental disabilities. He was a former teacher of children with autism and also supervised the delivery of ABA home programs for children with autism. All observers were undergraduate or graduate students with coursework in behavioral interventions and had experience collecting observational data in special education classrooms or home ABA programs. The experimenter provided additional data collection training in the form of a 20-min didactic presentation specific to the current investigation followed by practice sessions using video of children similar to the participants in the present investigation (see Appendix A). The presentation included a description of the target behavior and video samples of children with autism or other developmental disabilities emitting gestural behavior. Observers were required to demonstrate 90% agreement with data collected by the experimenter over three consecutive 5-min scoring sessions using the video samples prior to scoring behavior for the current investigation. Agreement was scored by dividing the smaller number of identified behaviors by the larger number and multiplying by 100 to obtain a percentage. 61 62 Materials Materials used with participants during the assessment included several preferred items identified during a paired choice preference assessment (Fisher et al., 1992) for each participant; classroom materials such as books, puzzles, and sorting activities; and a series of novel items such as dancing figures, spinning tops, and light up toys. The student and experimenter sat in child-sized chairs at tables raised approximately 75 cm from the ground.!The experimenters used a Canon! Vixia HG20 video recorder placed on a tripod to record sessions for later data collection and a MotivAider® vibrating timer to signal the end of predetermined intervals. Pre-Experimental Procedure Functional assessment inventory. Following referral of a student, the teacher and speech and language pathologist were asked to independently complete the Inventory of Potential Communication Acts (IPCA; Sigafoos et al., 2000), which can be found in Appendix B. The IPCA is an inventory that can be completed by parents or teachers to indicate caregiver perceptions of the behaviors children with severe language impairment use to communicate (Keen et al., 2001). The test-retest reliability of the IPCA is high (Sigafoos, Arthur-Kelly, & Butterfield 2006) though inter-rater reliability and validity have not been formally established. The IPCA was selected over other instruments because it is a brief inventory that teacher’s would likely be willing to complete and has previously been used to identify idiosyncratic behaviors that may serve a verbal function for children with severe disabilities (Sigafoos et al., 2000; Keen et al., 2001, Keen, Woodyatt, & Sigafoos, 2002). Any behaviors identified by educational professionals were included amongst a list of potential communicative behaviors assessed during pre-experimental observations. Preference assessment. A paired choice preference assessment (Fisher et al., 1992) was 63 conducted with each participant following the completion of the functional assessment inventory (see Appendix C for protocol and data collection instrument). This procedure has high concurrent validity for selecting items that function as potent reinforcers for an individual (Fisher et al.). Eight stimuli were selected based on parent and teacher recommendations and offered to each participant in a 2-item array. Each assessment involved 56 trials, which allowed each stimulus to be paired with all other stimuli in both the left and right presentation positions. A trial started when two stimuli were placed on a tray in front of a student and the student was told to “pick one.” The participant was given 20-s access to the selected stimulus, which was then gently removed by the experimenter. A 7-s intertrial interval separated the completion of one trial and initiation of the next trial. Each preference assessment took approximately 33 min to complete. Preferences were rank ordered for each participant based on the number of times the participant selected each stimuli. The three most frequently selected stimuli were identified as high preference, the next two most frequently selected stimuli were identified as medium preference, and the final three stimuli were identified as low preference (see Table 3.2). 64 Table 3.2 Rankings of preferences for each participant Participant Fuller HP Small wind up toys Balloon inflator Pinwheel Preferred Stimuli MP Helicopter Crackers LP Blocks Fruit snacks Music toys Bailey Spinning tops Polly Pocket Doll Tinkerbell movie Skittles Play-doh Cars movie Small wind up toys Blocks Victor Thomas the Train toy Balloon inflator Gummi worms Little People play set Spinning tops Light-up ball Toy car Blocks Morris Light up ball Balloon inflator Spinning tops Light up globe Croaking frog Koosh-ball Fritos Thomas the Train movie Matthew Koosh-ball Raisins Small wind up toys Starburst candy Light up globe Toy car Rubber band Crackers Note. HP = high preference; MP = medium preference; LP = low preference Functional assessment observations. The experimenter conducted two 20-min observations occurring on different days for each referred student. The goal of the observations was to first identify gestural behaviors emitted by each participant and second to determine which of those behaviors was frequently followed by a consequence mediated by another person. Consistent with previous research examining prelinguistic behavior or potential communicative acts, the length, location, and time of observations were based on the teacher’s report of a time and place where the student was likely to engage in potential communicative behavior (Keen et al., 2001; Meadan, Halle, Ostrosky, & DeStefano, 2008; Warren, Yoder, Gazdag, Kim, & Jones, 1993). Behavioral research methods including operationally defining target behaviors and brief 65 time sampling were used to develop procedures for both observations (Cooper et al., 2007). Observation 1. During the first observation, the experimenter recorded the occurrence of behaviors identified by Sigafoos and colleagues (2000) as potential communicative acts and any additional behaviors listed by each participant’s teacher on the IPCA (see IPCA observation checklist; Appendix D). Examples of behaviors measured during the first observation included pointing, reaching, moving toward others, and alternating gaze between objects and people. Operationally defined target behaviors (Appendix E) were recorded using 30-s partial interval sampling. Observation 2. The experimenter conducted a second observation to identify up to three frequently emitted potential communicative behaviors noted during the first observation and environmental variables correlated to those behaviors. Target behavior selection criteria included (a) the behavior being reported (via inventory) or observed to occur on a daily basis, (b) the behavior being reported or observed to occur primarily in the presence (i.e. within 5 ft) of other people and was directed toward other people, (c) the behavior being initiated by the student (as opposed to a response to the initiation of peer or adult), (d) the classroom teacher reported the behavior as an acceptable, or nonproblematic behavior. Environmental variables identified by educators’ completed IPCAs or those identified in previous research to frequently evoke and maintain verbal behavior (Brady & Halle, 1997; Greer & Ross, 2008; Halle & Meadan, 2007; Sundberg & Michael, 2001) were operationally defined (Appendix F) and measured during the second observation using the Environmental Observation Instrument (EOI; Appendix G). Examples of environmental variables included being alone, a unique or interesting environmental stimulus, preferred items being visible, obtaining a preferred item, obtaining peer or adult attention, and the removal of work or assigned tasks. 66 The EOI was a scatter plot procedure, which is empirically supported as a method for identifying temporal relations between environmental variables and target behaviors (Alberto & Troutman, 2008). The experimenter recorded the frequency of each target behavior and any environmental variables that correlated with an occurrence of a target behavior during a 20-min observation period. At the end of the observation, the experimenter identified gestures that (a) demonstrated a correlation with multiple antecedents and consequences and (b) involved a topography judged by the experimenter as similar to conventional verbal behavior (e.g., pointing was selected over rocking).!These behaviors were then selected as dependent measures for the experimental FA. Definition and Measurement of Dependent Variables Following the second observation, a target gesture(s) was selected for each participant using the target behavior selection criteria. Fuller’s gestures were grasping the experimenter’s hand and reaching. A gesture was scored if Fuller grasped the experimenter’s hand or arm with his own hand or extended his arm away from his body and in the direction of the experimenter. Bailey’s gesture was hand grabbing, defined as placing her hand on (a) the experimenter’s hand or (b) an item in the experimenter’s hand for a minimum of 1 s. The target gesture for Victor and Matthew was gazing toward and approaching the experimenter. A gesture was scored if the participant directed his eyes toward the experimenter and he either walked toward and stopped within 0.5 m of the experimenter or extended his arm toward the experimenter. Morris’ target gesture was moving toward the experimenter, which was defined as walking toward and stopping within 0.5 m of the experimenter or extending his arm toward the experimenter. This could include but was not limited to Morris making physical contact, such as grasping the hand of the 67 experimenter. Specific target behaviors and operational definitions for each participant are summarized in Table 3.3. The primary observer used a handheld counter to tally the frequency of each target behavior during live FA sessions; the secondary observer did the same during video recorded FA sessions. Rate of target behavior was obtained by dividing the total occurrences of a target behavior within a session by the number of minutes for a particular session. Data were directly imported into an excel spreadsheet for graphing and visual analysis by the experimenter on a daily basis. Daily visual analysis allowed for immediate decisions regarding the continuation or termination of the FA for each participant. 68 Table 3.3 Operationally defined target behaviors Participant Fuller Behavior Grasp hand or reaches Definition Participant grasps experimenter’s hand/arm with his hand or extends arm away from body and toward experimenter. Bailey Hand grabbing Participant grabs other person or places hand on other person for a minimum of 1 s. If participant has an item in hand, the item is counted as part of the hand and is therefore scored if it is in contact with experimenter for minimum of 1 s. Victor Eye gaze and approach Participant directs eye gaze toward and approaches experimenter. Approach is defined as walking up to and stopping within 0.5 m of experimenter or extending arm away from body toward experimenter. Morris Move Toward Participant extends arm toward or walks up to and stops within 0.5 m of experimenter. This may include grabbing and holding or pulling experimenter’s hand. Matthew Eye gaze and approach Participant directs eye gaze toward and approaches experimenter. Approach is defined as walking up to and stopping within 0.5 m of experimenter or extending arm away from body toward experimenter. Experimental Design An ATD was used during FAs to compare the occurrence of the target behavior during a series of alternating conditions. Variables within each condition included potential MOs, discriminative stimuli, and reinforcing consequences previously hypothesized to evoke and maintain mands (Carr & Durand, 1985; Dube et al., 2004; Greer & Ross, 2008; Hagopian, Bruzek, Bowman, & Jennett, 2007; Skinner, 1957; Taylor & Hoch, 2008). The action condition involved a visible, though inaccessible high preference stimulus as an antecedent and adult assistance in obtaining or operating the stimulus as a consequence. The escape condition involved the presentation of a potentially aversive stimulus as an antecedent and the contingent 69 removal of the stimulus as a consequence. The attention condition involved brief deprivation of attention as an antecedent and vocal and physical attention contingent on the target behavior as a consequence. The play, or control, condition involved frequent attention, free access to high preference stimuli, and no aversive stimuli to reduce the value of the consequences delivered in each test condition as reinforcers. The names of each condition, corresponding environmental variables manipulated within that condition, and the identified function of behavior that reliably occurs during a specific condition are identified in Table 3.4. The MO and antecedent describe components of the condition administered by the experimenter regardless of participant behavior. The consequence for each test condition occurred contingent on the target response. The identified function indicates the type of mand emitted by the participant if the gesture reliably occurred within that condition and not in the play condition. 70 Table 3.4 Functional analysis conditions with programmed variables. Condition Action Programmed Variables MO Antecedent Consequence Identified Function Deprivation Preferred Adult assists in Behavior is a mand from preferred activity and obtaining/operating for assistance item/activity adult are item visible Escape Presentation of aversive stimulus Task assigned, intrusive person present, or nonpreferred item present Person, task or nonpreferred item removed Behavior is a mand for the removal of person, work demands, or nonpreferred item Attention Deprivation from attention “You play while I get work done” Verbal and physical attention (e.g., “hi, how are you”, tickle, pat on back) Behavior is a mand for attention Play (control) None Access to preferred items & attention None Clear control of behavior has not been identified in other conditions. Ongoing assessment is required to identify function. Note. MO = motivating operation 71 Experimental Procedure Functional analyses were administered following the interview and observations, and were designed to assess for functional relations between a target behavior in the child’s repertoire and the environmental conditions hypothesized to evoke and maintain the target behavior (see Table 3.4). The experimenter implemented all FAs in a small room with a table, two chairs, and a cabinet in which materials were stored. A FA session consisted of the implementation of one of the FA conditions. Action and play (i.e., control) conditions lasted 5 min and attention and escape conditions lasted 10 min. The variability in session length was intentionally programmed to account for differences in MOs for each condition. For example, the MO in the escape condition was the presentation of an aversive stimulus and was likely strengthened as a function of extended exposure to the experimental condition. However, the MO for action is deprivation from the item or event presented during the action condition and is likely weakened as a function of exposure during the FA session (i.e., satiation). A maximum of four sessions (one of each condition) were completed during a single day for each student and were separated by a minimum of 10 min. Functional analyses continued for each participant until the student demonstrated consistent responding (Horner et al., 2005) in the control condition and at least one test condition. During all FA conditions, the experimenter ignored nonproblematic behaviors that were not targeted and redirected any problematic behaviors (e.g., vocal or motor stereotypy, crying, flopping) with light physical guidance. Action. The action condition was used to assess for the behavior as a mand for an adult to assist the child in obtaining or operating a preferred item. To increase the value of items used during the action condition as potential reinforcers (Michael, 2000), participants’ could not 72 access these items outside of the action FA sessions. At the beginning of the action condition, the experimenter guided the child into the assessment room where a preferred item or activity that required adult assistance was set out for the child. For example, the child’s most preferred toy was placed inside a transparent container with a lid that the child could not open. The experimenter instructed the child to play with the preferred item and stepped away from the child. If the child engaged in the target behavior at any time during the action condition, the experimenter assisted in obtaining the item or operating a toy (e.g., opened the container, spun a top) and then stepped away. After consuming an edible item or interacting with the toy for 20 s, the experimenter told the child “all done” and started the sequence from the beginning (i.e., returned the item to the container/table and stepped away). If after 20 s, the child did not engage in the target behavior, the experimenter briefly manipulated the item in the child’s line of sight, then returned the item to the container or placed it back on the table and stepped away. This process continued for the duration of the action condition. Escape. The purpose of the escape condition was to test if the child engaged in gestural behavior to escape or avoid a nonpreferred item (for Fuller), a person (for Bailey and Victor), or a demand to complete a task (for Morris and Matthew). Conditions varied across participants to account for differences in consequences correlated to gestural behavior identified during the second pre-experimental observation. Fuller’s gesture correlated to the removal of musical toys, gestures emitted by Bailey and Victor correlated with an increase in physical space between the participant and others, and gestures emitted by Morris and Matthew correlated to the avoidance or removal of demands to complete a task. To account for intersubject differences, programmed antecedents varied across participants during the escape condition. For Fuller, the antecedent involved presenting him with 73 toys that played music. For Bailey and Victor, the antecedent was the experimenter standing or sitting near the participant and talking while the participant interacted with moderately preferred items. For Morris and Matthew, the antecedent involved a direction to complete nonpreferred tasks such as coloring a picture, sorting blocks, or completing simple puzzles. The consequent variable was always the removal of the nonpreferred antecedent condition for 20 s. Specifically, the experimenter removed the music toy from Fuller’s line of vision, stepped away from Bailey and Victor, or removed the assigned task from Morris and Matthew. Participants were instructed to sit at a table or on the floor and were presented with the antecedent stimuli indicated above to start the escape condition (see Appendix G for a procedural example). For Morris and Matthew, the experimenter used a three-prompt sequence (Marcus, et al., 2001) including (a) the instruction followed by a 5-s pause, (b) the instruction with a model followed by a 5-s pause, and (c) the instruction with manual guidance was used to initiate a task sequence. The child received praise for completing a task unless manual guidance was required to initiate the task. Prompts and praise were not delivered to other participants during the escape condition. Anytime the child engaged in the target behavior, the experimenter delivered the programmed consequence for a period of 20 s while interacting minimally with the child. This sequence (i.e., present antecedent, observe for behavior, administer consequence) continued for the entire 10-min escape condition. Attention. The attention condition was used to assess for a gesture maintained by various forms of attention delivered by the experimenter (e.g., vocal statement, high-five’s, tickles, pat on the back). Participants were directed to sit down in a chair at or near the assessment table and were directed to play with low preference items (e.g., puzzle, Mr. Potato Head"). The experimenter turned away and sat or stood at least 1 m away from the participant and pretended 74 to complete paperwork attached to a clipboard. If the child engaged in the target behavior at any time during this condition, the experimenter delivered attention in the form of a vocal statement related to the activity the child was engaged in (e.g., “you found a red hat for Mr. Potato Head"”) and physical stimulation (e.g., tickle, pat on back). The participant was then directed back to the table to play with the toys. This continued for the duration of the attention condition. Play. The play condition was used as a control condition to test for the occurrence of the target behavior when the student received noncontingent access to preferred items and adult attention. During the play condition, the child was instructed to sit at a table in the assessment room and was given unlimited access to his or her most preferred toys or snacks and also received physical and verbal attention from the experimenter every 20 s. Additionally, the experimenter consistently assisted in the operation of any items the participant was not able to operate independently (e.g., wind-up toys). By delivering all potential reinforcers noncontingently, the experimenter attempted to decrease the likelihood of the behavior occurring as the child already had free access to items or events hypothesized to maintain the behavior. High rates of behavior during the play condition indicate that the experimenter did not adequately identify the variable(s) responsible for maintaining the behavior and further assessment would be required. Interobserver Agreement A secondary observer collected data for at least 33% of FA sessions across participants and conditions. Interobserver agreement (IOA) was calculated by comparing the primary observer’s data with the secondary observer’s data using the total agreement method (Cooper et al., 2007). The smaller amount was divided by the larger amount and multiplied by 100 to obtain a percentage of agreement. Mean IOA for Fuller during the action, attention, escape, and play 75 conditions was 93%, 100%, 86%, and 100%, respectively. Ranges are not included for any participants as IOA data were only collected during one or two sessions from each condition for each participant. Mean IOA for Bailey during the action, attention, escape, and play conditions was 93%, 100%, 100%, and 100%, respectively. Mean IOA for Victor during the action, attention, escape, and play conditions was 93%, 100%, 100%, and 100%, respectively. Mean IOA for Morris was 100% across all conditions. Mean IOA for Matthew during the action, attention, escape, and play conditions was 100%, 86%, 100%, and 100%, respectively. Procedural Integrity Procedural integrity is the degree to which research components are implemented as intended (Peterson, Homer, & Wonderlich, 1982). Accurate inferences about the function of communicative behavior can only be made when the procedures are accurately implemented. Therefore, a second year doctoral student in special education assessed the accuracy with which the experimenter implemented FA sessions. The experimenter randomly selected 20% of all FA sessions and provided the assessor with the video of each session and a categorical checklist (Appendix H) to determine whether or not the experimenter implemented the critical components for each condition. Mean percentage of procedural integrity during the action, attention, escape, and play conditions was 98% (range, 94% to 100%), 100%, 95% (91% to 100%), and 100%, respectively. Data Analysis Research Question 1 What are the functional properties (i.e., covarying environmental antecedents and maintaining consequences) of gestures emitted by a small group of children with autism? Visual inspection of the FA data was used to answer research question one for individual 76 participants. Visual inspection is used in single subject research methodology to compare differences in level, trend, and variability of data across experimental conditions (Horner et al., 2005). By analyzing multiple dimensions of behavior (i.e., level, trend, variability), visual analysis is capable of providing a complete description of the effects of the independent variables. A functional relation was identified when a participant consistently demonstrated considerably higher rates of behavior during at least one test condition when compared to the control condition. The terms “consistently” and “considerably” may present some level of uncertainty for readers more familiar with clearly established criteria, such as statistical significance or cutoff scores. When using visual inspection, consistency refers to predictable behavior under specific experimental conditions over a minimum of three sessions (Horner et al., 2005). “Considerably higher rates of behavior” means the experimenter detects visible differences in data series’ when assessed on a graph. A common complaint levied against this form of analysis is that effects are not identified visually (i.e., no functional relation exists) despite a statistical analysis revealing significant differences between baseline and treatment (Busk & Marascuilo, 1992). Though statistically significant differences may reveal some change in behavior, they are less likely than a more conservative visual analysis to reveal clinically significant changes in behavior. Because a purpose of ABA research is to identify how behaviors are changed to improve the day-to-day life of participants, visual analysis was selected for the present investigation. 77 CHAPTER 4 RESULTS Experiment 1: Assessment The results of this experiment are analyzed first in terms of the overall outcomes of the FA. Individual outcomes are then discussed with an emphasis on functional relations that can be derived from the data. Figures 4.1 to 4.5 display results of the FA for each participant. Gestural behavior averaged across participants was highest during the action condition (M = 1.5 per min) and demonstrated an increasing trend across sessions for most participants. Gestures were emitted at a moderate rate during the attention (M = 0.6 per min) condition and demonstrated a decreasing trend across sessions for most participants. Gestures occurred at a low rate during the escape condition (M = 0.2 per min) and remained stable at near zero levels during the play condition (M = 0.05 per min) for all participants. The results indicate that the majority of participants were most likely to emit gestural behavior when they were presented with high preference stimuli they could not access or operate and the gesture was followed by adult assistance. Gestures therefore functioned as mands for adult assistance for most, but not all, participants. Fuller Data for each of Fuller’s FA conditions are shown in Figure 4.1. Fuller demonstrated the highest rate of gestures per min during the action condition (M = 1.6) and demonstrated an increasing trend across sessions. He demonstrated minimal responding and a decreasing trend during the escape (M = 0.3), and near zero stability during the attention (M = 0.1) and play (M = 0.03) conditions. There was no overlap observed between responding during the action condition with responding during any other condition. Finally, effects of the action condition could be 78 considered immediate as relatively high responding (i.e., 1.3 per min) was observed during the first FA session for this condition. These results suggest that Fuller was most likely to emit gestural behavior to initiate an interaction with others when a preferred item he could not access or operate was present and the other person mediated his access to the item. Fuller was not likely to engage in gestural behavior to initiate interactions with others during any other conditions, such as when he sat by himself in a chair or when a nonpreferred stimulus was placed in front of him. Thus, Fuller’s gesture can be interpreted as a mand for assistance. ./01! 233453675! 283675! 9:80;4! Fuller’s rate of gestures '$%! '! &$%! &! #$%! #! ! "#$%! &! '! (! )! %! *! +! ,! -! &#! &&! &'! Sessions Figure 4.1. Fuller’s rate of gestural behavior during each of the FA conditions. The action condition is denoted by the shaded triangles, escape by the open triangles, attention by the open circles, and play by the shaded circles. 79 Bailey Bailey’s responding during each experimental condition is displayed in Figure 4.2. Bailey’s mean rate of responding was highest though somewhat variable during the action condition (M = 1.52). She consistently demonstrated some responding during the attention condition (M = 0.52) though a decreasing trend was also observed. Bailey emitted minimal gestures during the escape condition (M = 0.24) and no responding during the play condition. Bailey’s behavior demonstrated some overlap between the action and attention conditions, though differentiation of these conditions was apparent during the final two sessions for each condition. She demonstrated a similar level of responding during the initial FA session for all conditions suggesting a slight latency to the observed effects of the experimental conditions. Overall, Bailey engaged in gestural behavior when she could not access or operate preferred items and the experimenter mediated her access to the item. Though gestures occurred during early FA sessions for both the escape and attention conditions, these responses were not maintained over time. Taken collectively, the results indicate the function of Bailey’s gesture was a mand for adult assistance. 80 (! 233453675! ./01! 9:80;4! 283675! Bailey’s rate of gestures '$%! '! &$%! &! #$%! #! ! &! "#$%! '! (! )! %! *! +! ,! -! &#! &&! &'! &(! &)! &%! &*! &+! &,! &-! '#! Sessions Figure 4.2. Bailey’s rate of gestural behavior during each of the FA conditions. The action condition is denoted by the shaded triangles, escape by the open triangles, attention by the open circles, and play by the shaded circles. 81 Victor The results of Victor’s FA are displayed in Figure 4.3. Victor engaged in the highest rate and an increasing trend of gestural behavior during the action condition (M = 1.9). Gestural behavior demonstrated a decreasing trend and low rate during the attention condition (M = 0.5) and was near zero (M = .01) during the escape and play conditions. There was no overlap in responding during the action condition with any of the other conditions and responding during the action condition occurred immediately. Similar to Fuller and Bailey, Victor engaged in gestures when presented with an item he could not access or operate and the experimenter mediated his access to the item following gestural behavior. Victor’s gestures could therefore be interpreted as mands for assistance. 82 233453675! 9:80;4! 283675! ./01! '$,! Victor’s rate of gestures '$(! &$,! &$(! #$,! #$(! "#$'! ! &! '! (! )! %! *! +! ,! -! &#! &&! &'! Sessions Figure 4.3. Victor’s rate of gestural behavior during each of the FA conditions. The action condition is denoted by the shaded triangles, escape by the open triangles, attention by the open circles, and play by the shaded circles. 83 Morris The rate per min of Morris’ gestural behavior is displayed in Figure 4.4. Morris demonstrated the highest rate of responding in addition to an increasing trend during the action condition (M = 1.33). He demonstrated minimal responding during the attention (M = 0.13) and play (M = 0.07) conditions and no responding during the escape condition. There was no overlap in responding during the action condition with responding during any other condition and the effects of the action condition could be considered immediate as responding occurred during the first session. Based on the FA, Morris was most likely to emit gestural behavior when presented with preferred items he could not access or operate and the experimenter mediated his access to those items. These results suggest Morris’ gesture functioned as a mand for assistance. 84 '$%! 233453675! ./01! 9:80;4! 283675! Morris’ rate of gestures '! &$%! &! #$%! #! ! &! "#$%! '! (! )! %! *! +! ,! -! &#! &&! &'! Sessions Figure 4.4. Morris’ rate of gestural behavior during each of the FA conditions. The action condition is denoted by the shaded triangles, escape by the open triangles, attention by the open circles, and play by the shaded circles. 85 Matthew Matthew’s responding across all conditions is displayed in Figure 4.5. He engaged in high and stable rates of gestural behavior during the attention condition (M = 1.7 per min). Matthew also engaged in gestures during the action condition (M = 1.0), though responding was consistently lower than the attention condition and was variable across sessions. Matthew demonstrated low and stable responding during the escape (M = 0.2) and play (M = 0.1) conditions. These results indicate Matthew’s gestural behavior most reliably occurred when he was temporarily deprived of attention and the experimenter provided physical and vocal forms of attention contingent on the target response. Matthew sometimes emitted gestural behavior when presented with preferred items he could not access or operate and the experimenter mediated his access to those items. Therefore, his gestural behavior could be interpreted most frequently as a mand for attention from others and sometimes as a mand for assistance. 86 233453675! ./01! 9:80;4! 283675! Matthew’s rate of gestures '$%! '! &$%! &! #$%! #! ! &! "#$%! '! (! )! %! *! +! ,! -! &#! &&! &'! &(! &)! &%! &*! Sessions Figure 4.5. Matthew’s rate of gestural behavior during each of the FA conditions. The action condition is denoted by the shaded triangles, escape by the open triangles, attention by the open circles, and play by the shaded circles. 87 CHAPTER 5 DISCUSSION: ASSESSMENT The results of Experiment 1 in relation to Research Question 1 are discussed in the present chapter. The extent to which the results support and extend previous FA research and how they might inform mand training for children with autism are also discussed. This is followed by a discussion of limitations and suggestions for future research. Research Question 1 What are the functional properties (i.e., covarying environmental antecedents and maintaining consequences) of gestural behavior emitted by a small group of children with autism? The results indicate that four of the five participants emitted gestures when they were presented with a preferred item they could not access or activate and adult assistance was provided to help the child obtain a preferred consequence. The increase in gestural behavior across sessions during the action condition showed that the delivery of adult assistance was a reinforcing consequence. The presentation of high preference inaccessible stimuli was an EO that temporarily increased the value of adult assistance as a reinforcing consequence and therefore increased the probability that gestural behavior would occur. Behavior that is controlled by MOs and maintained by specific reinforcement is, by definition, a mand (Skinner, 1957). Therefore, Fuller, Bailey, Victor, and Morris all engaged in gestural behaviors to mand for adult assistance. Matthew emitted the highest rate of gestural behavior when briefly deprived of attention and attention was delivered contingent upon the occurrence of the target response. The delivery of attention by the experimenter reinforced Matthew’s gestural behavior. Deprivation from 88 attention was an EO that temporarily altered the value of attention as a reinforcer and therefore increased the probability he would engage in behaviors that obtained attention in the past. Although Matthew engaged in some gestural behavior during the action condition, his responding was less reliable across FA sessions. Deprivation from attention was a stronger EO than high preference inaccessible stimuli and contingent attention a more reliable reinforcer than contingent assistance. Unlike the other participants, Matthew emitted gestural behavior to mand for attention. These results support previous research showing that children with autism emit idiosyncratic gestures that function as mands (Ferreri & Plavnick, in press; Langdon et al., 2008; Sigafoos et al., 2000). Importantly, the findings also suggest the specific reinforcers that maintain these mands may vary across individuals. These findings extend the use of pretreatment FA to nonproblematic behavior and suggest subtle variations to mand training may be necessary to produce consistent positive outcomes (Ahearn et al., 2007; Carr & Durand, 1985; Dufrene, Steuert Watson, & Kazmerski, 2008; Hagopian, et al., 2007; Iwata et al., 1994; Marcus et al., 2001). Implications for Functional Analyses In order to understand the contribution of pretreatment FA, it may be helpful to revisit the manner in which this type of assessment informs intervention. FAs enable an interventionist to identify specific variables relevant to an individual’s learning history that do and do not cause problem behavior (Hanley et al., 2003). Once these variables are identified, an interventionist can intervene by removing antecedent stimuli responsible for evoking the behavior or by interrupting the relation between the behavior and the reinforcing consequence. Because the 89 antecedent and consequent stimuli vary across individuals, the manipulation of these variables during treatment must also vary (Iwata et al., 1982/1994). By extending FA to nonproblematic behaviors, interventionists can identify MOs and discriminative stimuli that evoke an approximation of a target response and consequences known to reinforce the behavior. When MOs, discriminative stimuli, and reinforcers are known prior to intervention, an interventionist can increase learning opportunities and shape or prompt new responses that are likely to be maintained in multiple environments. The FA of gestural behavior could therefore inform educational interventions similar to the way in which the FA of problem behavior informs behavioral interventions (Carr & Durand, 1985; Iwata, et al., 1982/1994; Marcus et al., 2001). The extension of FA to gestural behavior might also be able to address some limitations of the FA methodology. The FA of problem behavior has been criticized for intentionally attempting to strengthen a target behavior in order to identify the variables responsible for its occurrence (Neef & Peterson, 2007). For some behaviors such as self-injury, aggression, or property destruction, it may be dangerous or unacceptable to care providers to strengthen a response prior to treatment. However, gestural and problematic behaviors can become members of the same response class (Langdon et al., 2008), which is often demonstrated by the occurrence of specific gestures immediately prior to an occurrence of the problem behavior (Borrero & Borrero, 2008). When this occurs, a FA of the gesture may provide the same information as a FA of the problem behavior without the potential to strengthen a dangerous or undesirable behavior. Another benefit of the FA of gestural behavior is that it may allow for preventive treatment of problem behavior. Based on the laboratory model developed by Shabani and colleagues (2009), it is possible for children with severe disabilities to develop a response class 90 hierarchy whereby problem behavior only occurs when nonproblematic precursers to the behavior, such as gestures, are not reinforced. Busy classrooms are likely to simulate such a scenario, as teachers often do not notice potentially communicative gestures emitted by children with severe autism (Keen, Sigafoos, & Woodyatt, 2005). These children are therefore at great risk for the development of problematic behavior. However, pretreatment FA of gestural behavior followed by a function-based intervention may help a child learn to communicate without relying on problem behavior. Implications for Mand Training Fuller, Bailey, Victor, and Morris engaged in the highest rate of gestural behavior during the action condition. Thus, mand training for those four participants would likely be most effective when they are taught to mand for preferred items or assistance from others. This is similar to many mand training procedures, which involve deprivation from preferred items to evoke an initiating behavior such as eye gaze, pointing, or reaching; following an initiating behavior, an interventionist can immediately prompt or shape a conventional mand (Bondy & Frost, 1994; Drash et al., 1999; Hall & Sundberg, 1987; Halle et al., 1979; Jennett et al., 2008). The results for the four participants are therefore not surprising in that they align with previous research showing that children with autism are most likely to acquire mands if an EO that restricts access to high preference items is contrived and the item is delivered contingent on the target response (Sundberg, 2007; Sundberg & Michael, 2001). The results for the four participants whose gestures functioned as mands for assistance may clarify a reason why mand training is often effective. Children with autism are frequently taught to mand for preferred items, which are identified by preference assessments and are often effective as reinforcers (Jennett et al., 2008; Sundberg, 2007). By manipulating access to high 91 preference items, interventionists may be able to contrive an EO that evokes gestural behavior (Bondy & Frost, 1994; Sundberg & Michael, 2001) in a manner similar to the action condition used in the present investigation. If a gesture consistently occurs under such conditions, an interventionist can assume an EO has been contrived and that contingent access to the item is reinforcing the gesture. Any conventional mand topography that is then trained as a target response would be functionally equivalent to the gesture. Assuming the child is able to emit the selected response, they are likely to acquire the new mand topography. The potential limitation of mand training in the manner just described is that the strength of the response is limited by the restriction of preferences selected by the experimenter or interventionist. Such a restriction could mean that a child capable of vocal speech does not learn vocal mands or that a child does not acquire mands as rapidly as possible because the stimuli used to evoke and maintain the response are not as potent as they could be. It is possible that consequences outside of those involved in a preference assessment, such as attention or the removal of aversive stimuli, could evoke and maintain mands with greater reliability than items involved in the assessment and often used during mand training (Sundberg, 2007). Therefore, the FA used to identify classes of reinforcers and related MOs with the greatest potency prior to training could increase the likelihood of optimal training procedures. Matthew’s results exemplify the limitation to mand training outlined above, as he was more likely to emit gestures during the attention condition. Therefore, it is likely that mand training would be more effective when it replicates the attention, as opposed to the action, condition. Attention can function as a reinforcer though it may not be selected as a high preference consequence because deprivation from attention is not typically used to create the EO for selecting attention over other stimuli during a preference assessment (Cote, Thompson, 92 Hanley, & McKerchar, 2007; Fisher et al., 1992). Attention could therefore be overlooked as a variable that can be manipulated to evoke and maintain mands for an individual child. By conducting a FA of gestures prior to mand training, an interventionist can develop procedures that are sensitive to variations in function across participants. This may improve the individualization of educational interventions, which is necessary to teach language or alternative forms of verbal behavior to children with severe autism (NSR, 2009). Alternative Communication Systems Many children with autism do not acquire vocal speech and need an alternative communication system, such as the PECS or sign language, to mand for preferred items, actions, or events. The extension of the FA methodology to gestural behaviors may inform the process by which an alternative communication system is used to teach verbal behavior to a child with autism. Professional judgment is the current standard for deciding if and when a child requires a communicative topography other than vocal speech (Schlosser & Wendt, 2008). Aside from recommendations that professionals should first attempt to teach vocal speech to children with autism, there are not clear guidelines regarding when an alternative system should be used (Schlosser & Wendt; Sundberg & Michael, 2001). Given the absence of empirical guidelines for alternative communication systems, interventionists could prematurely begin teaching a child to communicate using an alternative system or continue attempting to teach speech to a child who is not likely to acquire such a capacity. For example, none of the participants in the present investigation demonstrated a vocal mand repertoire despite several attempts to evoke speech by classroom teachers. Perhaps this was because participants were not going to acquire a vocal repertoire and therefore required an alternative communication system, though it was also possible that the environmental conditions 93 created by the teachers were not sufficient to evoke the target response. A FA prior to mand training might ensure that optimal conditions for mand training had been identified and the absence of vocal mands following such training would indicate an alternative system may be necessary. Limitations & Future Research Although the FA methodology identified functional relations between gestures and controlling variables, some limitations of the present procedures could be addressed by future research. First, the action condition combined assistance and tangible items as consequent variables. For example, a preferred item was placed in a tightly sealed container or on a shelf where the child could not access it. Participants may have been more likely to respond when they possessed the item (i.e., they were given the jar) than when they did not (i.e., item on shelf). This could occur because an item in their possession is more evocative than an item several feet away. This might explain the variability in responding demonstrated by Bailey during the action condition. Future research should assess for these controlling variables under different experimental conditions in order to increase the sensitivity to idiosyncratic responding across participants. A second limitation was that the attention condition might not have replicated naturally occurring events responsible for controlling gestural behavior for all participants. This could occur because attention may only function as a reinforcer when it is delivered in a certain way or by a specific person, which could explain the pattern of responding during the attention condition displayed by Bailey and Victor. A moderate or high level of gestural behavior during early attention sessions rapidly decreased to low levels during later sessions. Deprivation from attention may have evoked gestural behavior though the delivery of attention from the 94 experimenter, who was a relative stranger, did not maintain the response. A teacher who was familiar to the participants could have been a conditioned reinforcer or could have delivered attention in a certain way that was not identified by the procedures used in the present investigation. This limitation is somewhat mitigated by the fact that Bailey and Victor engaged in reliable gestural behavior during the action condition. Alternatively, the attention condition could, at best, produce responding that was dependent on a specific listener. An interventionist would likely select conditions that produce reliable behavior across numerous settings rather than conditions that produce behavior that is limited to a specific person. The FA was therefore useful in that it provided information that could be used to inform mand training. Nevertheless, future research should examine manipulations to antecedents such as the person delivering attention or the type of attention delivered while holding the consequence constant during an extended attention condition. A third limitation was that the stimuli used during the escape condition may not have been aversive. The purpose of the escape condition in a FA is to present an aversive stimulus that evokes problem behavior maintained by the removal of the aversive stimulus (Iwata et al., 1982/1994). Stimuli were selected for the condition based on teacher reports of items participants did not like or correlations with the removal of specific stimuli during pre-experimental observations. The aversiveness of the stimuli was not empirically assessed, thus it is possible participants did not engage in gestures during the escape condition because the stimuli presented were not actually aversive. It is also possible that the stimuli were initially aversive but extended exposure reduced this effect across escape sessions. The latter is especially relevant to children with autism, for whom novel stimuli may be temporarily aversive (Luiselli, 2008). This would 95 explain gradually decreasing levels of gestural behavior demonstrated by four out of the five participants during the escape condition. A final possibility is that the removal of stimuli noted during pre-experimental observations was correlated but not functionally related to gestural behavior, which would indicate the FA accurately captured variables responsible for producing gestural behavior. Ultimately, the purpose of the FA was to reliably identify conditions under which a child was likely to acquire new mand topographies. The results therefore provide valuable information even if the stimuli were not aversive or the aversive properties decreased over time. First, if aversive stimuli occurring in the everyday environment cannot be easily identified by classroom teachers or during brief observation, it may be extremely difficult and possibly unethical to create conditions that are mildly aversive without inadvertently producing problem behavior. Second, if stimuli lose aversive qualities over time, any trained response that is maintained by the removal of the stimulus would become less reliable. As indicated by the FA, participants in the present investigation would therefore be more likely to acquire mands that are maintained by action or attention across a variety of environments. Several additional considerations of the present investigation will be important for future research. First, though pretreatment FAs may improve mand training protocols, additional research is needed to verify this conclusion. Intervention studies need to be conducted whereby the functional relation identified during FA is used to successfully teach novel mands to participants. Additionally, a control condition within such an investigation will allow for an examination of the effectiveness of environmental variables that were not functionally related to the gestural behavior. 96 Second, the inverse relation between problem behavior and mand acquisition has been reliably demonstrated by researchers attempting to suppress problem behavior (Carr & Durand, 1985; Durand & Carr, 1991; Langdon et al., 2008; Marcus et al., 2001; Wacker et al., 1990). It therefore seems possible that mand training could have a collateral or preventive effect on problem behavior. However, very little research has empirically demonstrated such a functional relation (Charlop-Christy et al., 2002). The educational, behavioral analytic, and autism research communities stand to benefit from investigations that experimentally reveal the relation between early mand acquisition and problem behavior. Third, the empirical evidence for transporting FAs to applied settings is mixed (Neef & Peterson, 2007). Though Iwata and colleagues (2000) trained undergraduate psychology students to accurately implement FA procedures, efficient methods for teaching educators to plan, implement, and analyze experimental FAs are not yet known (Wallace, Doney, Mintz-Resudek, & Tarbox, 2004). Given the long term benefits of identifying effective educational and behavioral interventions at an early age, future research should continue to evaluate training procedures that make the comprehensive administration of FAs feasible for school personnel. The present investigation showed that the FA methodology previously used for problem behavior can be extended to identify the functional properties of nonproblematic verbal behavior. This might have important implications for teaching verbal repertoires to individuals with autism. However, these implications are uncertain until interventions based on the outcomes of FA are empirically examined. Therefore, Experiment 2 examines the effects of mand training based on the functional relations identified during Experiment 1. 97 CHAPTER 6 METHOD Experiment 2: Intervention The present chapter describes the methodology used to answer Research Questions 2-5 identified in Chapter 2. The following components of the experiment are described in detail: (a) the participants and setting, (b) operational definitions of all dependent variables and measurement procedures, (c) the experimental procedures, (d) the experimental design, (e) methods for ensuring the reliability and validity of this investigation based on the common standards in ABA research, and (f) an overview of data analysis procedures. Readers are referred to Chapter 2 for an in-depth discussion of data analysis methods in single subject research. Participants and Setting All participants who participated in the first experiment also participated in the second experiment. Participants were 3 to 6 years old, attended public preschool programs, and demonstrated severe autistic symptoms including speech and language impairment. All participants attended an eclectic 2.5 hr program for five days each week except Bailey, who attended an eclectic 5.5 hr program for five days each week. None of the participants engaged in functional communication using vocal or alternative communication systems. Intersubject variability in listener and speaker repertoires required variations to experimental procedures for some participants. The listener and speaker repertoires of each participant are described in the present section and all procedural modifications are described in the representative sections below. Fuller. Fuller had a small vocal repertoire consisting of the phrases “flush it” and “Trex.” Each phrase was consistent with Schreibman and Lovaas’ (1974) definition of delayed 98 echolalia as stimuli related to the phrase were not present when the vocal responses were emitted. The experimenter asked Fuller’s teacher to ensure he was given an opportunity to use the bathroom any time he said “flush it” though the teacher reported no correlation between the response and eliminating in the bathroom or his diaper. Fuller did not imitate others though he did follow several simple instructions and he reliably attended to video and computer screens when movies or cartoons were displayed. Bailey. Bailey emitted the response “it’s okay” during and following tantrum behavior, which involved yelling, crying, and flopping on the floor. She emitted no other vocal verbal behavior prior to the present investigation. Bailey did not imitate others and did not follow simple instructions. She did attend to video and computer screens when movies or cartoons were displayed. Bailey’s teacher had attempted to teach her to exchange a picture for preferred food items at snack time though Bailey was dependent on adult prompts to emit the response. Victor. Victor demonstrated no vocal verbal behavior prior to the present investigation though he did emit loud vocalizations in the form of “eee” or “aaahhh” during tantrums, which involved yelling and throwing items. Victor could imitate others, though he did so unreliably. He did not follow simple directions and did not reliably attend to video and computer screens when movies or cartoons were displayed. Morris. Morris demonstrated no vocal verbal behavior prior to the present investigation. He did emit the sound “aaa” when playing independently but would not echo the sound following an adult model. Morris did not demonstrate an imitative repertoire or follow simple directions. He did attend to video and computer screens when movies or cartoons were displayed. Matthew. Matthew did not demonstrate vocal verbal behavior prior to the present 99 investigation. He loudly yelled the sounds “eee” and “aaa” during a variety of classroom conditions including, being alone, being with others, working, playing, and eating. He would not emit these sounds as an echoic response to an adult’s model. The results of a FA of these vocalizations revealed they were automatically reinforced and were not likely to serve a social function. Matthew did not imitate, follow simple directions, or attend to video or computer screens prior to the present investigation. Video models. Four typically developing children ranging from 3 to 9 years of age participated as video models in the present investigation. The experimenter recruited models by sending a letter of informed consent to parents of children within the participants’ classrooms. The letter explained the purpose of the investigation and identified the need for typically developing children to participate as video models. Within the letter, the experimenter asked parents if they had a typically developing child who would be willing to participate in the investigation as a video model. Assent to participate was obtained from all models and permission to record video of models was obtained from parents prior to recording any videos. The experimenter’s daughter and three children who were siblings of the participants’ classmates volunteered to be video models. Jane, the experimenter’s daughter, was a 3-year-old female peer who modeled responses for Bailey, Victor, and Matthew. Mackenzie was a 9-yearold female peer who modeled responses for Fuller and Victor. Sebastian was a 5-year-old male peer who modeled responses for Bailey. And Henry was a 7-year-old male peer who modeled responses for Morris. Video models volunteered to create video clips at recess, during lunch, or after school. Recording sessions never exceeded 10 min. Setting. All baseline and training sessions were conducted within each participant’s school building and in the same private rooms used during the FA. Generalization sessions were 100 conducted in each participant’s classroom. Two of the classrooms were Early Childhood Special Education classrooms located in the same elementary school. Each classroom included a teacher, two paraprofessionals, and 10 or 11 students with various disabilities. The third classroom was a public school ASD program administered at the county level. A teacher, three paraprofessionals, and seven children with autism between 2 and 6 years of age were in the ASD classroom. Materials Several items were used to evoke verbal behavior. Spinning tops, wind-up toys, balloon inflators, balls that lit up when bounced, cars that made noise when pushed, Polly Pocket# dolls, Thomas the Train# toys, and bubbles are examples of some of these items. Preferred items were placed in containers with lids that small children could not open. Edible items such as skittles, fruit snacks, starbursts, and potato chips were used with some participants. Puzzles, sorting games, building blocks, and crayons and paper were used during some intervention sessions. When pictorial exchange was selected as an alternative mode of verbal behavior, the experimenter used laminated 4.5 cm x 4.5 cm picture cards of preferred items as symbols for the actual items. Each picture card displayed a word printed on the top of the card and below that, a picture that symbolized the word. Materials related to technology included a Canon! Vixia HG20 placed on a tripod and used to record video and audio during all intervention sessions. The video camera was also used to record video models engaging in the target behaviors. Video clips were streamed into an MP4 format using iSkysoft software and a MacBook computer. The video clips were then loaded onto the experimenter’s iPhone 3G!, which was used to show the video clips to participants. MotivAider® vibrating timers were used to signal researchers at the end of predetermined intervals during baseline and intervention conditions. 101 Video clips. During intervention, participants viewed 15- to 27-s video clips of a typically developing peer emitting the target response. All video clips started with a close-up view of a specific setting condition such as a peer model sitting alone or an interesting item inside a transparent container. The camera zoomed out to capture a scene including a peer model, an adult listener, and any stimulus materials involved in the original setting condition. The model then looked at the adult and emitted the target response. The video clip ended with the adult providing the consequence specified by the model’s mand. Function-based video clips depicted a peer model manding for consequences that were functionally related to the consequence identified as a reinforcer for gestural behavior during Experiment 1. For example, if a participant’s gesture functioned as a mand for attention, a video clip started by showing a child sitting alone. The child then stood up, walked across the room to an adult, and emitted a mand for attention (e.g., high-five, play chase). The adult responded by extending her hand to give the model five or engaged in a game of chase with the model. Nonfunction-based video clips showed a peer model manding for a class of consequences that were not functionally related to the consequence identified as a reinforcer during Experiment 1. For example, if a participant’s gesture functioned as a mand for the removal of an aversive stimulus, the video depicted a teacher delivering a task demand to a peer model, such as instructing the model to complete a puzzle. After placing one piece of the puzzle in the correct location, the model manded for a break and the teacher responded by saying “okay, take a break,” picking up the puzzle, and walking away from the child. Definition and Measurement of Dependent Variables Primary dependent variables included verbal behavior targeted during mand training sessions and generalization probes. Additional dependent variables included listener and problem 102 behaviors. The listener behaviors for each participant were orienting to the speaker and following directions. Problem behaviors varied across participants and included crying, yelling, mild aggression, property destruction, mouthing, and stereotypy. The following sections define each dependent variable and describe measurement procedures. Target behaviors. Target behaviors included six verbal responses for each participant. The topography of responding was vocal for all participants except Matthew. The response topography selected for Matthew was pictorial exchange based on the absence of vocal responding and results of an experimental analysis demonstrating faster acquisition of independent picture exchange behavior than independent sign language behavior under highly controlled instructional conditions. Table 6.1 presents specific target behaviors for each participant during both conditions and aligns those behaviors with the function of each participant’s gestural behavior identified during experiment one. Target behaviors were taught as word pairs; one word was taught using function-based training procedures and the other using nonfunction-based procedures. Specific words were selected based on several criteria. First, an emphasis was placed on teaching responses that were likely to contact reinforcement in natural environments. For example, instead of teaching participants to ask for a toy specific to training conditions, they were taught to ask for help operating a toy as the mand “help” could be used in a variety of settings. 103 Table 6.1 Target mands and topographies for each participant Target mands Participant Function of Gesture Obtain Assistance Response Topography Vocal Verbal FB NFB Help me Open this I want ____ Come Play Break Look Bailey Obtain Assistance Vocal Verbal Help me Open Movie Come Play Break Look Victor Obtain Assistance Vocal Verbal Approximations Open Help Again Wow Look Break Morris Obtain Assistance Vocal Verbal Approximations Go Top Matthew Obtain Attention Picture Exchange Tickle Chase High 5 Break Ball Music Toy Fuller Note. FB = function-based; NFB = nonfunction-based A second consideration used to select specific target words was that function-based mands had to align with the function of gestures analyzed during Experiment 1. For example, if the function of a participant’s gesture was to obtain attention from adults, the function-based mand could be “look” or “come play” as these responses typically result in attention. Conversely, nonfunction-based mands were selected because they did not align with the function of gestures analyzed during experiment one. For example, if the function of a gesture was to obtain attention, the nonfunction-based mand could be “break” or “all done” because these responses 104 typically result in escape from task demands but not in attention. A third consideration for the selection of target responses was that the function-based target within a word pair could not be easier to pronounce than the nonfunction-based target. This was done to rule out effort as an explanatory factor if function-based were acquired and nonfunction-based mands were not. A licensed speech and language pathologist was consulted to confirm that all function-based mands required the same or more effort than nonfunction-based mands. Each word pair was introduced sequentially, once the student demonstrated mastery of at least one word from the previous pairing. For Fuller and Bailey, each phoneme in the target response needed to be emitted in the correct order and with no pause in between to be scored as a correct response. That is, the word needed to be stated clearly. Correct responses were initially scored for Victor and Morris if any phoneme in the targeted response was emitted. The response requirement was increased when the participant either demonstrated two consecutive sessions with 100% correct responding or emitted a closer approximation to the target word. A correct response was recorded for Matthew if he picked up the picture card and placed the card in the palm of the experimenter’s hand. Target behaviors were recorded correct for all participants if they occurred within 20 s of the experimenter’s presentation of stimulus materials or within 20 s of the participant being exposed to a setting condition designed to evoke the behavior (described in the procedures section). Responses were recorded as incorrect if they did not occur within that same time frame. Percent of trials with an accurate response were measured and recorded live during all baseline, training, and generalization sessions. Generalized responding. Generalized responding included the emission of a target 105 response (Table 6.1) in a new setting or situation and as target response maintenance occurring under post VM follow-up conditions (Cooper et al., 2007). Generalized responding was measured as a percentage of correct responding during probes not preceded by a display of the video model. No prompts were delivered prior to the response and no unrelated reinforcers were delivered following the response. Generalized responding in a new setting or situation was defined as the occurrence of a target response in the students’ classroom and either in the presence of a new listener or when a participant was presented with stimulus materials not used during training. Response maintenance during follow-up conditions was defined as an acquired verbal response emitted after the termination of all mand training procedures. Follow-up observations occurred at 1, 2, 4, and 8 weeks following the termination of mand training to assess for response maintenance. Listener behaviors. Listener behaviors were recorded to analyze the extent to which mand training impacted behaviors that might be correlated with mand acquisition (Koegell, Vernon, & Koegel, 2009). Listener behaviors included orienting to a speaker and following directions. Orienting was defined as turning head toward speaker within 3 s of the participant being spoken to. Following directions was defined as engaging in physical behavior that corresponded to a vocally delivered instruction, such as sitting in a chair when instructed to “sit in the chair.” The participant had to initiate physical movement within 3 s of the experimenter delivering the instruction. Listener behaviors were probed when seven to 10 rapidly administered trials for each response were delivered during baseline and intervention conditions. Orienting trials consisted of the experimenter calling a participant’s name and making a statement when the participant was engaged with another item. Following directions trials consisted of the experimenter delivering 106 an instruction for the participant to engage in a particular behavior. Data were collected from randomly selected video recordings and converted to a percentage of correct responding. The randomization process consisted of creating a numbered list of each session in Microsoft Excel$, randomizing the list, and selecting every other session to record data. Problem behaviors. Problem behaviors were recorded to analyze the extent to which mand training conditions affected problem behavior. Problem behaviors included any of the following behaviors listed and defined in Appendix E: bolting, flopping on floor, echolalia, motor stereotypy, vocal stereotypy, whimpering, whining, crying, yelling, mouthing, property destruction, aggression, and self-injury. These behaviors were probed using a 30-s partialinterval procedure and recorded as percentage of intervals with problem behavior during all conditions. Data were collected from the same video recording used to collect listener behaviors. Procedure In order to clarify the procedure section, a brief overview of events will be explained prior to a technological description of the procedures. Broadly speaking, the experimental procedures involved repeated measurement of the dependent variables for each participant during baseline and intervention conditions. The baseline condition was an interactive play session involving the experimenter and an individual participant. The intervention condition involved VM procedures to teach the target behaviors. The intervention was divided into two conditions in order to compare the effects of function-based and nonfunction-based VM for each participant. If participants did not acquire target behaviors during the initial VM condition, additional VM phases were introduced with slight modifications to the procedures. Modifications varied across participants to account for differences in pre-experimental verbal repertoires. Figure 6.1 visually illustrates the sequence of 107 phases and the decision-making process used in the present experiment, which is described in the text below. Table 6.2 depicts a timeline of procedures for each participant, starting with the preexperimental procedures implemented during experiment one. The table indicates the research activities a particular participant was involved in at any given time. Procedures started in January and continued until the middle of June when the school year ended for all participants. 108 109 Yes Yes Yes VM Phase 3: Experimenter assigns quick task as a consequence for an incorrect response during third phase. VM Phase 2: Increase response requirement to shape target mand. Add word pairs as participant demonstrates acquisition! VM Phase 2: Experimenter initiates interaction and participant responds to mand for completion of event during second phase. Mands are unreliable VM Phase 1: Reinforce approximation until participant demonstrates 100% accuracy across 2 consecutive sessions Continue with Phase 1 procedures Figure 6.1. Sequence of video modeling (VM) conditions and phases. Participants’ exposure to phases varied based on response to procedures during each phase. Select alternative communication system to teach mands. Repeat entire process starting with baseline. No! Participant has a reported history of emitting vocal verbal behavior No! Participant emits approximation of target mand during either VM condition No! VM Phase 1: Implement functionbased and nonfunction-based VM conditions. Participant emits target mands during either VM condition Conduct baseline with all participants 110 Baseline. The purpose of the baseline condition was to identify the extent to which each participant engaged in each dependent measure when presented with an antecedent event that would typically evoke such behavior. Baseline for each participant consisted of a series of trials administered by the experimenter embedded within a 30-min play session. The experimenter and participant were the only individuals in the room during play sessions. During each play session the experimenter provided opportunities for a child to engage in preselected target mands by creating conditions that had the potential to evoke the mand and in a manner that shared similar features to the videos used during the intervention condition (described below). For example, if a target response was “open” or “open this” the experimenter placed a highly preferred item inside a transparent container that the participant was not able to open. The container was then placed within the reach and visual field of the participant. The trial concluded when the child emitted the target mand or 20 s post stimulus presentation. If a target response was “look”, the experimenter placed an interesting object, such as a toy bear that played music and danced, in the center of the room to increase the likelihood of the participant recruiting the experimenter’s attention upon interacting with the item. Stimuli associated with each of the six target mands were presented one at a time and in a randomized order. Matthew required a baseline for both vocal verbal and picture exchange topographies. During the picture exchange baseline, the experimenter held a picture icon symbolizing the target response in front of Matthew’s field of vision and then set the icon on the table prior to presenting the stimulus as described above. All other aspects of the picture exchange baseline were identical to the vocal verbal baseline procedures. The experimenter provided five response opportunities (i.e., trials) for each target behavior during baseline conditions. If the participant emitted the target response within 20 s, the 111 experimenter engaged in the behavior that was indicated by the child’s mand (e.g., attention was delivered if the child said “look”). If the child engaged in gestures identified in experiment one but not the target response, the experimenter distracted the child by activating another toy or manually guiding the child to interact with another item. All other behaviors such as yelling, crying, flopping, throwing items, and vocal or motor stereotypy were ignored or lightly redirected with minimal interaction between the experimenter and participant. A 10-s intertrial interval followed the termination of prescribed consequences. Following each baseline play session, participants were observed for 20 min during naturally occurring classroom conditions to assess for targeted verbal behavior occurring in the classroom environment. Stimuli were presented in the same manner as the analogue setting to contrive opportunities for responses to occur within the classroom setting. Correlated behaviors were not examined in the classroom setting. The baseline condition lasted until a participant demonstrated a steady state of targeted verbal behavior across a minimum of three experimental sessions. At that time, the initial word pair was targeted for the VM intervention and all other word pairs were probed under baseline conditions; baseline probes continued until each word pair was sequentially targeted for acquisition in VM conditions. Video modeling: phase 1. Following baseline, a VM procedure was used to teach language or alternative communication to each participant. The intervention consisted of two conditions: function-based VM and nonfunction-based VM. Function-based conditions were designed to replicate the functional relation identified during experiment one whereas nonfunction-based conditions were designed to teach a verbal response that was unrelated to conditions identified during experiment one. Table 6.3 displays the antecedent and consequence 112 variables that were manipulated for each target response during the function-based and nonfunction-based conditions. Additional information pertaining to implementation procedures can be found in Appendix J. Antecedent variables were potential MOs and discriminative stimuli designed to evoke the target response. These are referred to as evocative events. Consequence variables were potential reinforcers that a verbal community typically delivers to individuals following the corresponding target response (Skinner, 1957). For example, as the function of Matthew’s gesture was to obtain attention, a functionbased response was a mand for a high five. After showing him the video of a child asking for a high five via picture exchange, the experimenter moved away from Matthew to leave him alone as an evocative event. If Matthew placed the high five picture card in the experimenter’s palm, the experimenter delivered an energetic high five as a consequence for the target mand. A nonfunction-based response for Matthew was to request a break from working on puzzles. After showing him a video of a peer model asking for a break from completing a puzzle, the experimenter placed a puzzle in front of Matthew and instructed him to complete the puzzle as an evocative event. If Matthew placed the break picture card in the experimenter’s palm, the experimenter picked up the puzzle and walked to the other side of the room for 20 s as a consequence. 113 114 Condition Function-based Fuller Matthew I want [name of item] Tickle Matthew Fuller, Bailey, Victor, Morris Open or open this Matthew Bailey Victor Morris Chase Students Fuller, Bailey, Victor Target Response Help or help me High 5 Movie Again Go Evocative Event HP item that required assistance to operate Wind top or blow up balloon and say “ready, set” then pause Provide brief access to preferred activity Provided movie player but did not turn on movie Brief ignore Brief ignore Brief ignore HP item on high shelf HP item in opaque container with tightened lid Evocative events and consequences delivered for each target behavior Table 6.3 Eject top so that it spins or let go of balloon Provide additional access to preferred activity Turn on movie Give student a high 5 Chase student around room and then catch him Tickle student Obtain item from shelf and give to participant Open container Contingent Consequence Operate item 115 Condition Nonfunction-based Table 6.3 (cont’d). Students Victor, Morris Fuller, Bailey, Victor Break Look Morris Fuller, Bailey, Victor, Matthew Come play Matthew Matthew Top Fuller, Bailey Target Response Wow Ball Music Toy Hold music toy just out of reach Hold ball just out of reach Spin top and ask “what’s that?” Novel toy that did something unique Work tasks assigned LP toys on floor and playing alone Evocative Event Provide extended access to novel and exciting event (e.g., spinning a top that lit up Give student music toy Give student ball Provide physical and vocal attention Enhance activity by playing in fun way Remove assigned task for 20 s Enhance activity by playing in fun way Contingent Consequence Provide physical and vocal attention The experimenter started all VM sessions by instructing the participant to sit down and view a video clip that corresponded with the target behavior identified for a specific session. After showing the participant a video on an iPhone 3G!, the experimenter created an evocative event, paused 20 s to allow the participant to emit the target response, and provided the programmed consequence contingent on the participant emitting the target response. The experimenter ignored or lightly redirected all other responses and allowed the 20-s response period to elapse if the participant did not emit the target response. Each trial was followed by a 10-s intertrial interval and another trial was initiated until five trials were conducted for a particular condition. All VM training sessions involved five trials of either the function-based or nonfunctionbased condition, a 3-min break, and five trials of the second condition for a particular word pair. Function-based trials were administered first for the majority of sessions. However, one of every four sessions was randomly selected for nonfunction-based trials to be administered prior to function-based trials. The purpose of administering trials in the order described above was to address threats to internal validity in two ways. First, function-based trials were typically administered prior to nonfunction-based trials to control for sequence effects. By implementing function-based conditions first, any acquisition of function-based mands was less likely a result of exposure to nonfunction-based training sessions. Second, nonfunction-based trials were randomly implemented first during one out of every four VM sessions to rule out training fatigue as an explanation if function-based mands were acquired but nonfunction-based mands were not. A slight procedural variation was included for Matthew to account for a variation in his response topography. During picture exchange training for Matthew, the experimenter held a 116 picture icon of the target mand in Matthew’s field of vision and placed the picture on the table within Matthew’s reach immediately following the evocative event. Consistent with the PECS procedures (Frost & Bondy, 2002), the experimenter sat within 0.5 m of Matthew during the 20-s response period to allow for easier acquisition of the exchange. No prompts aside from the presentation of the video were provided during training sessions. Additionally, no reinforcers aside from those specified by participants’ target mands were delivered for accurate responding. However, because Victor and Matthew did not follow simple directions or attend to videos prior to the VM intervention, the experimenter gave them small pieces of candy for sitting down when instructed and directing eye gaze toward the video screen during all VM conditions. Once a participant demonstrated 80% accurate responding across 3 consecutive sessions for one of the targets in a word pair, the word was considered “acquired” and an additional word pair was incorporated into training sessions. Training for acquired words involved the presentation of stimulus materials prior to the video model and a progressive time delay starting at 5 s to allow for transfer of stimulus control from the video to the materials (i.e., the participant had an opportunity to respond to the conditions prior to viewing the video). Training for target responses not yet acquired continued to include video modeling prior to presentation of stimulus materials. As depicted in figure 6.1, procedural variations were developed for participants who did not emit the target response during either VM condition. If a participant did not emit the full target response but emitted either a partial vocal response or caregivers reported previously hearing the child emit vocal words (i.e., Victor and Morris, respectively), additional phases were included in the experimental conditions. The additional phases involved shaping target 117 responses, adult initiated mand sequences, or delivery of task demands following incorrect responses. If the participant did not emit a partial response after six consecutive VM sessions and caregivers had not previously reported hearing the participant emit a vocal word (i.e., Matthew), an alternative topography was selected and phase one started over. Video modeling: phase 2. An additional phase was included for Victor and Matthew to account for shaping responses from initial approximations. Victor’s initial target response was a vocal approximation that included any phoneme within the complete word. Once Victor demonstrated 100% accurate responding across two consecutive sessions for the initial response requirement or began emitting additional sounds within the target response, his response requirement was increased for that word only. For example, the requirement for “open” went from “o” to “o-en.” Victor had to emit a response with at least 75% of the correct phonemes in the correct order at 80% accuracy for three consecutive sessions before word pairs were added to training sessions. For Matthew, the second phase was consistent with the PECS procedures and involved increasing the distance between himself and the experimenter once he acquired three target responses from one of the VM conditions (Frost & Bondy, 2002). Matthew’s original response requirement was to place a picture card in the experimenter’s hand, which was 0.5 m away. During the second phase, Matthew had to travel approximately 1.5 m to complete the picture exchange, which more closely approximated typically occurring classroom conditions. Additionally, instead of conducting five consecutive trials for each mand, the six target mands were randomly assigned to each of 10 trials. The reason for randomly assigning target mands to trials for Matthew during the second phase was to attempt to support accurate manding in the future. As Matthew’s response 118 topography was always picking up a picture card and placing it in an adult’s hand, it was important to begin teaching him that consequences varied depending on the picture that was on the card (Frost & Bondy, 2002). Randomly assigning mands to trials allowed for exposure to variations in mands and consequences during training sessions. It was hoped that this exposure would increase the effectiveness of future mand training for Matthew. A second VM phase was also included for Morris as he did not emit mands during the first phase though his teacher and parents reported hearing him emit vocal words. During the second phase, the experimenter showed Morris a video of an adult initiating an interaction with a peer model. The adult in the video showed the peer a toy and said “ready, set...” while looking expectantly toward the peer (function-based), or asked the peer “what’s that?” after handing an item to the peer (nonfunction-based). The peer then responded by saying “go” or labeling the item. The adult in the video complied with the mand during function-based sessions or provided praise and physical attention for accurately labeling items during nonfunction-based sessions. Following the video, the experimenter engaged in the same behavior that was modeled and then looked expectantly toward Morris. If Morris emitted the target response, the prescribed consequence was delivered (e.g., spinning the top for “go”, or responding with praise and physical attention for correctly answered questions). Video modeling: phase 3. A third phase of VM was included for Morris after it was discovered that viewing video clips was itself a high preference activity for Morris. A preference assessment was conducted following VM session 33 and showed that viewing video clips of the models was Morris’ most preferred activity or item. Thus, viewing videos may have functioned as noncontingent reinforcement for some unidentified behavior(s) during the first and second phases. 119 During the third phase, Morris was instructed to complete a simple task such as a puzzle following incorrect responses but not correct responses. The purpose of this manipulation was to create a delay between an incorrect response and the opportunity to view a video clip at the beginning of the following trial. If viewing the video clips functioned as a reinforcer, it was assumed that immediate access to the video would increase the likelihood of the target mand compared to delayed access to the video. If Morris did not emit the target response after 20 s, the experimenter presented a five-piece wooden puzzle that Morris could complete in approximately 30 s. The experimenter provided an oral instruction to complete the puzzle, followed by a model and physical prompt if Morris did not comply with the instruction. Once the puzzle was completed, another VM trial was initiated. Training for generalization. The VM procedures were designed to promote generalization from training sessions to natural environments using several tactics described in a seminal article by Stokes and Baer (1977). First, target responses were selected, in part, because of the likelihood that the response would be followed by reinforcement in natural environments. Second, multiple stimulus exemplars were used for each target response. For example, when teaching Fuller to emit “help me” the experimenter presented him with a wind-up toy for one trial, a spinning top that had to be twisted together and released for a second trial, and a balloon with an inflator for another trial. Each video clip varied in correspondence with the stimuli used during each trial, though the target response was always the same. Additionally, the adult who responded to the peer model in the video also varied across trials. A third component of the procedure used to program for generalization involved the inclusion of each participants’ teacher as an adult listener within the video clips. The teachers as listeners in the video clips functioned as common stimuli for the generalization setting. These procedures were incorporated into all 120 phases of both VM conditions. Generalization probes. Throughout all experimental conditions, the experimenter probed for generalization of the target response to a new setting or situation by contriving conditions in the participants’ classroom that were similar to the training sessions. For example, to evoke the mand “open this,” a participant was given a bag of potato chips that he could not open himself for snack. To evoke the mand “help me,” the experimenter asked the teacher to direct the student to complete a fine motor task she was not able to complete independently. All generalization probes were conducted in the students’ classroom and involved either novel stimuli, a listener not involved in mand training, or a combination of the two. Similar to baseline, listeners complied with target mands emitted by participants and ignored or redirected other behavior. No videos were presented prior to the three to five probes conducted during a generalization session and no additional reinforcers were provided for correct responding. Generalization probes could not be conducted for Victor due to scheduling difficulties with the classroom teacher. Follow-up. A series of follow up sessions occurred 1, 2, 4, and 8 weeks following the termination of VM for all target behaviors for Fuller and at 1, 2, and 4 weeks post VM for Bailey. Follow-up sessions were identical to baseline conditions. The purpose of these sessions was to assess for response maintenance of acquired skills after termination of verbal behavior training conducted by the experimenter. Follow-up sessions were not conducted for Victor, Morris, and Matthew because the VM conditions for these participants were administered through the end of the school year in mid-June; the participants were not available for additional sessions beyond that point. Experimental Design 121 An ATD or ATD within a MPD across behaviors was used to examine the differential effects of function-based and nonfunction-based interventions on the targeted and related behaviors emitted by each participant. The ATD was used to demonstrate the effects of each intervention protocol administered to a participant on an alternating schedule. The MPD was used to attempt two controlled replications of the experiment whenever possible. The MPD is a variant of the MBD and is used when baseline behavior is extremely stable and intermittent measures are capable of demonstrating any change in behavior prior to implementation of the independent variable (Cooper et al., 2007). In the present experiment, the ATD was used to identify functional relations between dependent measures and one of the two intervention conditions. By sequentially applying the interventions to additional targets using a MPD, the strength of an observed functional relation between each intervention and target behaviors could be enhanced. This combined design allows for a comparison of one intervention against another intervention (ATD) as well as the identification of the individual effect of each intervention (MPD). Sequential application of the intervention was not possible for Morris or Matthew due to the alterations to experimental procedures. However, the ATD allowed for a comparison between the two intervention conditions. The only other single subject design that could compare the effects of two independent variables would have been a traditional reversal design. The ATD was selected over a reversal to decrease the likelihood of sequence effects impacting behavior and because it allowed for a more rapid assessment of differential effects of the experimental conditions. Reliability and Validity Measurement reliability. All research team members participating in data collection were trained to collect data in the same manner described for experiment one and prior to 122 participating in any research activities. Observers were required to demonstrate 90% agreement ratings across two consecutive practice sessions with the experimenter prior to collecting participant data. A secondary observer independently collected data during 25% of the sessions for all target behaviors across all conditions, participants, and experimental phases. Agreement was calculated by comparing the primary observer’s data with the secondary observer’s data using a point-by-point reliability calculation (Cooper et al., 2007). That is, each trial was scored as an agreement or disagreement; total agreements were divided by the sum of agreements and disagreements and multiplied by 100 to obtain a percentage. The mean and range of IOA across all conditions and target responses for each participant is presented in table 6.4. 123 124 Internal validity. The experimental design controlled for threats to internal validity by eliminating factors such as maturation, historical effects, testing, or selection bias (Kazdin, 1982). Additionally, an implementation protocol and checklist (see Appendices J and K) with a behavioral description of each procedural component was developed to ensure accurate implementation of the procedures. A secondary observer recorded the occurrence or nonoccurrence of each component during 20% of all intervention sessions across participants, conditions, and experimental phases. Procedural integrity was 96% (range, 91% to 100%). External validity. To address issues of external validity pertaining to this investigation, only participants with autism who demonstrated similar levels of communicative competence prior to research activities were included. Additionally, repeated demonstrations of functional relations between the dependent variable and the independent variable across several participants provides a strong indication of the overall effectiveness or ineffectiveness of a particular procedure (Sidman, 1960). The characteristics of participants and number of replications within the study increased the ability to generalize identified functional relations to a particular subgroup of young children with autism (Horner et al., 2005). Social validity. Social validity offers a subjective analysis of the degree to which an intervention targets socially significant behavior, uses acceptable procedures, and demonstrates clinically significant outcomes (Wolf, 1978). Social validity is a critical component of ABA interventions and research (Cooper et al., 2007). An assessment of social validity must consider the overall importance of the goals for a participant, the extent to which the procedures are acceptable to consumers (e.g., participant, family members, teachers), and the extent to which outcomes allow the participant to access reinforcers. Social validity was assessed in the present investigation by asking parents and teachers of 125 participants to view and evaluate a 5-min video clip of the participant during each of the baseline, nonfunction-based, and function-based conditions. Video clips were presented in random order and raters were unaware as to the experimental conditions for each video. Following each video clip, consumers were asked to indicate their agreement or disagreement with four statements pertaining to the goals, procedures, and outcomes of the intervention (see Appendix L for full questionnaire). Statement one was “the student is able to communicate basic needs and wants in a way that people who do not know him/her are likely to understand.” Statement two was “the student appears to be having a good time.” Statement three was “the student is learning skills in addition to communication.” Statement four was “this is a good use of the student’s time.” A 10-point rating scale was used with a score of 10 indicating “total agreement” and a score of 1 indicating “total disagreement.” Table 6.5 displays the social validity ratings provided by caregivers after viewing the video clips. Table 6.5 Social validity ratings for all experimental conditions Question Other listeners would understand this response Baseline 3.5 (2-5) FB 9.25 (8-10) NFB 5.5 (1-10) Student is having a good time 5.5 (4-7) 10 7.25 (5-10) Student is learning skills in addition to communication 5 (2-10) 9.5 (8-10) 7.5 (6-10) Good use of student’s time 8.25 (5-10) 9.75 (9-10) Note. FB = function-based; NFB = nonfunction-based 8.5 (6-10) 126 In addition to rating video clips, educators were asked to complete a brief survey pertaining to the overall project after the completion of all experimental procedures. Specifically, educators were asked to evaluate the importance of the project goals, the likelihood that they would continue using the intervention with the participant, the likelihood that they would use the intervention with other students, and the likelihood that they would recommend the intervention to other educators or to parents. The same rating scale used for video clips was used for the overall satisfaction questions. The mean rating for each of the four questions was 10. Data Analysis Visual analysis was used to answer Research Questions 2-5. Visual analysis was selected for reasons discussed in Chapter 2 (pp. 24-27) and in the data analysis section of Experiment 1 (pp. 81-82). Question 2. To what extent does the implementation of function-based and nonfunctionbased mand training procedures affect vocal or alternative (e.g., picture exchange) verbal behavior acquired and emitted by a small group of children with autism? Visual inspection (Horner et al., 2005) was used to identify relations between a participant’s acquisition and emission of vocal or alternative verbal behavior and the application of each type of mand training procedure. Visual inspection is a type of data analysis used to identify socially significant changes in a dependent variable as a function of some manipulation to the independent variable (Horner et al.). Visual inspection is not sensitive to small changes in behavior and is therefore ideal for identifying meaningful changes in participant behavior that can be attributed to an educational or behavioral intervention (Cooper et al., 2007). Level, trend, variability, and latency were assessed following the application and removal 127 of each training procedure to targeted dependent variables. The MPD across behaviors assessed for functional relations between the application of each type of communication training procedure and the emission of targeted verbal operants. The ATD within the MPD assessed for any differential effects of the two intervention procedures (function-based versus nonfunctionbased). Simply speaking, the effects of both mand training procedures on target responding could be evaluated independently and comparatively with one another. Question 3. To what extent does the implementation of function-based and nonfunctionbased mand training procedures lead to generalized responding? a. To what extent are acquired mands generalized across listener’s following the implementation of function-based and nonfunction-based mand training sessions? b. To what extent are acquired mands generalized across settings following the implementation of function-based and nonfunction-based mand training sessions? c. To what extent are acquired mands generalized across stimulus materials following the implementation of function-based and nonfunction-based mand training sessions? d. To what extent are acquired mands generalized over time (i.e., maintained) following termination of mand training procedures? Generalized responding was assessed using visual inspection of data collected during baseline, generalization probes within the classroom during VM conditions, and the follow-up condition (i.e., observations at 1, 2, 4, and 8 weeks following the termination of training). Subquestions ‘a’ ‘b’ and ‘c’ were answered by analyzing the level, trend, and variability of 128 generalized responding for target behaviors acquired during each VM condition. Subquestion ‘d’ was answered by analyzing the level of responding during follow-up conditions compared the level of responding over the final three sessions of VM conditions. Question 4. To what extent does the implementation of function-based and nonfunctionbased mand training procedures targeting vocal or alternative verbal behavior affect listener behaviors correlated to mand acquisition (i.e., orienting to the speaker, following directions)? Visual analysis of listener behaviors during baseline and VM conditions allowed for closer examination of a potential relation between the acquisition of a mand repertoire and listener behaviors. Responding during baseline conditions was compared to responding during both VM conditions and each VM condition was compared to the other VM condition. It was possible that the acquisition of a mand repertoire would lead to no or minimal changes in listener behavior, which would suggest no relation between these variables. It was also possible, though unlikely, that listener behaviors would change in the same way and to the same degree during both VM conditions, which would indicate a correlation between experimental conditions and the change in listener behavior. Finally, it was possible that the VM conditions would have a differential effect on the occurrence of listener behavior, which would indicate a functional relation between one of the VM conditions and an increase or decrease in listener behavior. Question 5. To what extent does the implementation of function-based and nonfunctionbased mand training procedures targeting vocal or alternative verbal behavior affect problem behaviors emitted by participants? Visual analysis of problem behaviors during baseline and VM conditions allowed for closer examination of a potential relation between the acquisition of a mand repertoire and problem behavior. Responding during baseline conditions was compared to responding during 129 both VM conditions and each VM condition was compared to the other VM condition. It was possible that the acquisition of a mand repertoire would lead to no or minimal changes in problem behavior, which would suggest no relation between these variables. It was also possible, though unlikely, that problem behavior would change in the same way and to the same degree during both VM conditions, which would indicate a correlation between experimental conditions and the change in problem behavior. Finally, it was possible that the VM conditions would differentially affect problem behavior, which would indicate a functional relation between one of the VM conditions and an increase or decrease in problem behavior. 130 CHAPTER 7 RESULTS Experiment 2: Intervention Table 7.1 displays the mean percentage of responding for all participants for each dependent measure across all experimental conditions. The mean percentage of trials each participant emitted accurate mands during baseline, function-based, and nonfunction-based conditions is displayed in the left-most column. The mean percentage of trials each participant engaged in orienting and direction following behavior for each condition is then displayed. Finally, the right-most column displays the mean percentage of intervals each participant engaged in problem behavior across conditions. Figures 7.1 through 7.15 display results for each of the five students who participated in Experiment 2. The first figure for each participant displays the percentage of trials a participant emitted the target mand during baseline, VM, generalization, and follow-up conditions. Shaded circles represent function-based responding and shaded triangles represent nonfunction-based responding during training sessions. Open circles represent generalized function-based responding and open triangles indicate generalized nonfunction-based responding during all conditions. The second figure for each participant displays the percentage of trials a participant engaged in orienting and following directions during baseline and VM conditions. The third figure for each participant displays the percentage of intervals a participant demonstrated problem behavior during baseline and VM conditions. Shaded circles represent behavior during the function-based condition and open circles represent behavior during the nonfunction-based condition. 131 132 28% 28% 17% 43% 59% 41% 41% 33% 38% 23% 31% 57% 72% 83% 80% 63% 68% 37% 63% 40% 41% 17% 17% Matthew 0% 80% 41% 91% 87% 71% 68% 71% 90% Note: BL = baseline; FB = function-based condition; NFB = nonfunction-based condition. ! 0% Morris 86% 2% 37% 0% Victor 81% 20% 26% 8% 16% 30% 34% 43% 40% Problem Behavior BL FB NFB 48% 10% 48% 75% 0% Bailey Table 7.1 Mean responding for each participant across all dependent measures and conditions ! Participant Target Mands Orienting Following Directions BL FB NFB BL FB NFB BL FB NFB Fuller 0% 74% 4% 16% 54% 22% 30% 79% 60% Fuller Targeted and generalized responding. Results of VM on targeted and generalized mands for Fuller are displayed in Figure 7.1. Fuller demonstrated no responding during baseline across all word pairs. During the function-based VM condition for the first word pair, Fuller immediately acquired the target mand and his mean percentage of responding was 78% (range, 20% to 100%). During nonfunction-based VM, Fuller did not acquire the target mand and his mean percentage of responding was 5% (range, 0% to 20%). Fuller generalized the functionbased response to the classroom and to novel listeners or stimulus materials with 100% accuracy across all sessions. He also maintained the behavior with 100% accurate responding when presented with an evocative event in the classroom at 1, 2, 4, and 8 weeks post VM. Similar changes in responding occurred for the second word pair; Fuller’s mean percentage of responding was 88% (range, 60% to 100%) and 0% during function-based and nonfunction-based conditions, respectively. Fuller generalized the function-based response with 100% accuracy across all sessions. He also maintained the behavior with 100% accurate responding when presented with an evocative event in the classroom at 1, 2, 4, and 8 weeks post VM. 133 Percent of trials Fuller emitted the target mand ! ! &##! -#! ,#! +#! *#! %#! )#! (#! '#! &#! #! "&#! Baseline Video Modeling Follow-up Generalization Probes Function based Nonfunction based Word Pair 1 “Help me/ Come play” ! ! &##! -#! ,#! +#! *#! %#! )#! (#! '#! &#! #! "&#! Word Pair 2 “Open this/ Break” ! ! &##! -#! ,#! +#! *#! %#! Word Pair )#! “I want/ (#! look” '#! &#! #! "&#! &! )! +! &#! &(! &*! &-! ''! '%! ',! (&! ()! (+! )#! )(! )*! )-! %'! %%! %,! *&! ! 3 ! ! Sessions Figure 7.1. The percent of trials Fuller emitted the target mand during all conditions and across targeted word pairs. Shaded data indicate VM training trials; open data indicate generalization probes. Follow up consists of generalization probes at 1, 2, 4, and 8 weeks post-training. 134 The function-based response took longer to emerge for the third word pair, though levels eventually increased in a manner similar to the first word pairs. Fuller’s mean percentage of responding was 63% (range, 0% to 100%) and 4% (range, 0% to 20%) during the function-based and nonfunction-based conditions, respectively. The function-based response was generalized to the classroom and to novel listeners or stimulus materials at a mean percentage of 83% (range, 50% to 100%). Fuller also maintained the behavior with 100%, 100%, 100%, and 80% accurate responding when presented with an evocative event in the classroom at 1, 2, 4, and 8 weeks post VM, respectively. These results demonstrate a clear functional relation between the functionbased VM condition and mand acquisition for Fuller. Listener behaviors. Figure 7.2 shows the percent of trials Fuller oriented to the speaker and followed directions during baseline and VM. During baseline, Fuller’s mean percentage of orienting was 16% (range, 0% to 30%). During VM, Fuller’s mean percentage of orienting was 54% (range, 22% to 80%) and 22% (range of 14% to 44%) under function-based and nonfunction-based conditions, respectively. The differentiation in responding across conditions suggests a functional relation between the function-based VM condition and orienting behavior for Fuller. Fuller’s mean percentage of following directions during baseline was 30% (range, 10% to 40%). He demonstrated an increased level of following directions during both VM conditions, though responding under function-based conditions (M = 79% [range, 60% to 100%]) was higher than responding under nonfunction-based conditions (M = 54% [range, 22% to 80%]). Despite some differentiation, the variability in following directions and decrease in responding during later sessions of the function-based condition does not allow for clear conclusions about the relation between either VM condition and following directions. 135 Video Modeling Percent of trials Fuller oriented to speaker Baseline <=583675">0:4?! &##! @75A=583675">0:4?! -#! ,#! +#! *#! %#! )#! (#! '#! &#! Orienting #! ! "&#! &! (! %! +! -! &&! &(! &%! &+! &-! '&! '(! '%! '+! Percent of trials Fuller followed directions &##! -#! ,#! +#! *#! %#! )#! (#! '#! Following Directions &#! #! ! "&#! &! (! %! +! -! &&! &(! &%! &+! &-! '&! '(! '%! '+! Sessions Figure 7.2. The percent of trials Fuller oriented to the speaker (top panel) and followed directions (bottom panel) during baseline and both VM conditions. Shaded data indicate responding during function-based conditions and open data indicate nonfunction-based conditions. 136 Problem behavior. Figure 7.3 displays the percent of intervals Fuller engaged in problematic behaviors during baseline and both VM conditions. Fuller’s primary problem behavior was crying though he occasionally engaged in vocal and motor stereotypy, head banging, yelling, and mouthing. During baseline, Fuller’s mean percentage of problem behavior was 48% (range, 30% to 70%). During VM conditions, Fuller’s mean percentage of problem behavior was 18% (range, 0% to 40%) and 48% (range, 30% to 80%) under function-based and nonfunction-based conditions, respectively. The differentiation in level of problem behavior during the two VM conditions demonstrates a probable functional relation between the functionbased VM condition and a reduction in Fuller’s problem behavior. 137 138 Percent of intervals Fuller engaged in problem behavior "$ %$ &$ Baseline '$ ($ )$ *$ +$ Nonfunction-based Sessions ,$ "#$ ""$ "%$ "&$ "'$ "($ ")$ "*$ "+$ ",$ %#$ %"$ %%$ %&$ %'$ %($ %)$ %*$ Function-based Video Modeling Figure 7.3. The percent of intervals Fuller engaged in problem behavior during baseline and both VM conditions. Shaded data indicate responding during function-based conditions and open data indicate nonfunction-based conditions. !"#$ $ #$ "#$ %#$ &#$ '#$ (#$ )#$ *#$ +#$ ,#$ "##$ Bailey Targeted and generalized responding. Results of VM on targeted and generalized mands for Bailey are displayed in Figure 7.4. Bailey demonstrated no target responding during baseline across all word pairs. When VM was implemented for the first word pair, Bailey’s mean percentage of responding was 97% (range, 80% to 100%) and 0% during function-based and nonfunction-based conditions, respectively. The function-based response was generalized to the classroom and used with novel listeners or stimulus materials with a mean percentage of 95% (range, 86% to 100%). She also maintained the behavior with 100% accurate responding when presented with an evocative event in the classroom at 2 and 4 weeks post VM. Bailey demonstrated similar changes in responding for the second word pair; her mean percentage of responding was 88% (range, 60% to 100%) and 0% during function-based and nonfunction-based conditions, respectively. The function-based response was generalized with a mean percentage of 92% (range, 75% to 100%). When presented with an evocative event in the classroom, Bailey emitted the function-based response with 100%, 100%, and 66% accuracy at 1, 2, and 4 weeks post VM, respectively. The third function-based response took longer to emerge, though levels increased similar to the first word pairs. Mean percentage of responding for the third pair was 60% (range, 0% to 100%) and 5% (range, 0% to 20%) during the function-based and nonfunction-based conditions, respectively. The function-based response was generalized with 100% accuracy across all sessions and maintained at 100% accuracy at 1, 2, and 4 weeks post VM. These results demonstrate a clear functional relation between function-based VM and mand acquisition for Bailey. 139 Percent of trials Bailey emitted the target mand ! &##! -#! ,#! +#! *#! %#! )#! (#! '#! &#! #! "&#! Baseline Video Modeling Follow-up Function based Generalization Word Pair 1 “Help me/ Come play” Nonfunction based ! ! &##! -#! ,#! +#! *#! %#! )#! (#! '#! &#! #! "&#! Word Pair 2 “Open/ Break” ! ! &##! -#! ,#! +#! *#! %#! )#! Word Pair (#! “Movie/ '#! Look” &#! #! "&#! &! (! %! +! -! &&! &(! &%! &+! &-! '&! '(! '%! '+! '-! (&! ((! (%! (+! (-! )&! ! 3 ! ! Sessions Figure 7.4. The percent of trials Bailey emitted the target mand during all conditions and across targeted word pairs. Shaded data indicate VM training trials; open data indicate generalization probes. Follow up consists of generalization probes at 1, 2, and 4 weeks post training (2 and 4 weeks for word pair 1). 140 Listener behaviors. Figure 7.5 shows the percent of trials Bailey oriented to the speaker and followed directions during baseline and both VM conditions. During baseline, Bailey’s mean percentage of orienting to the speaker was 17% (range, 10% to 40%). During VM, Bailey’s mean percentage of orienting was 41% (range, 20% to 60%) and 38% (range, 0% to 67%) under function-based and nonfunction-based conditions, respectively. Orienting was slightly more stable during the function-based condition though the results do not make it possible to identify a clear relation between orienting and either VM condition. Bailey’s mean percentage of followed directions was 57% (range, 14% to 90%) during baseline conditions. During VM, Bailey’s mean percentage of following directions was 80% (range, 57% to 100%) and 68% (range, 40% to 100%) during function-based and nonfunctionbased conditions, respectively. The increase in following directions identified under both VM conditions suggests a correlation between the intervention and the target response. A functional relation between one of the VM conditions and following directions cannot be identified due to high variability and the corresponding overlap between the two VM conditions. 141 Percent of trials Bailey oriented to speaker Baseline Video Modeling &##! <=583675">0:4?! @75A=583675">0:4?! -#! ,#! +#! *#! %#! )#! (#! '#! Orienting &#! #! !"&#! &! '! (! )! %! *! +! ,! -! &#! &&! &'! &(! &)! &%! &*! &+! &,! Percent of trials Bailey followed directions &##! -#! ,#! +#! *#! %#! )#! (#! '#! &#! Following Directions #! !"&#! &! '! (! )! %! *! +! ,! -! &#! &&! &'! &(! &)! &%! &*! &+! &,! Sessions Figure 7.5. The percent of trials Bailey oriented to the speaker (top panel) and followed directions (bottom panel) during baseline and both VM conditions. Shaded data indicate responding during function-based conditions and open data indicate nonfunction-based conditions. 142 Problem behavior. Figure 7.6 displays the percent of intervals Bailey engaged in problematic behaviors across all conditions. Her primary problem behavior was crying though she occasionally engaged in motor stereotypy and flopping. During baseline, Bailey’s mean percentage of problem behavior was 63% (range, 40% to 90%). Problem behavior decreased during both VM conditions, though the function-based condition was more effective in suppressing problem behavior (M = 16% [range, 10% to 30%]) than nonfunction-based conditions (M = 40% [range, 10% to 100%]). The immediate decrease in problem behavior during the function-based condition along with differentiation of responding across conditions during the final nine sessions indicates a probable functional relation between the function-based VM condition and a reduction in problematic behavior. 143 144 ! Percent of intervals Bailey engaged in problem behavior #! $! %! &! '! (! )! *! ,-./012.345678! #"! Sessions +! ##! #$! 92.:-./012.345678! Video Modeling #%! #&! #'! #(! #)! #*! Figure 7.6. The percentage of intervals Bailey engaged in problem behavior during baseline and both VM conditions. Shaded data indicates function-based VM conditions and open data indicates nonfunction-based conditions. "! #"! $"! %"! &"! '"! ("! )"! *"! +"! #""! Baseline Victor Targeted behavior. Figure 7.7 displays results of targeted mands for Victor, though generalized responding was not obtained due to scheduling difficulties with the classroom teacher. Victor demonstrated no target responding during baseline across all word pairs. When VM was implemented for the first word pair Victor demonstrated an increase in function-based responding and minimal change in nonfunction-based responding. During phase one, the mean percentage of responding for function-based and nonfunction-based conditions was 100% and 10% (range, 0% to 20%), respectively. During phase two, Victor’s mean percentage of responding for function-based and nonfunction-based conditions was 86% (range, 60% to 100%) and 8% (range, 0% to 40%), respectively. Vocal mands were successfully shaped into closer approximations of the target response during the function-based condition. A different pattern of responding emerged for the second word pair. During phase one, Victor’s mean percentage of responding for function-based and nonfunction-based conditions was 67% (range, 40% to 80%) and 47% (range, 0% to 80%), respectively. Victor emitted a closer approximation to the complete function-based response, which signaled an increase in the response requirement while the nonfunction-based response requirement remained the same. During phase two, Victor’s mean percentage of responding was 96% (range, 80% to 100%) and 10% (range, 0% to 20%) under function-based and nonfunction-based conditions, respectively. Though the pattern was slightly different than the first word pair, the outcome was the same. Victor’s function-based mand had been shaped and the nonfunction-based mand was not. Victor immediately emitted the function-based response “/gen/” but did not emit any approximation of the nonfunction-based response “break” for the third word pair. Responding for this word pairing was 80% and 0% during every session of function-based and nonfunction- 145 based VM, respectively. A second phase was not included for the third target as the school year ended before this could occur. ! Percent of trials Victor emitted the target mand Baseline “O”/”ow” &##! -#! ,#! +#! *#! %#! )#! (#! '#! &#! #! "&#! ! &##! -#! ,#! +#! *#! %#! )#! (#! '#! &#! #! "&#! Video Modeling “O-en” / “ow” Function-based Nonfunction-based Word Pair 1 “open/wow” “eh” / ”oo” “elp” / ”oo” ! Word Pair 2 “help/look” ! ! &##! “gain” / ”b” -#! ,#! +#! *#! %#! )#! (#! Word Pair 2 '#! “again/break” &#! #! "&#! &! (! %! +! -! &&! &(! &%! &+! &-! '&! '(! '%! '+! '-! (&! ((! (%! (+! (-! )&! ! Sessions ! ! Figure 7.7. The percent of trials Victor emitted an approximation of the target mand during all conditions and across word pairs. 146 Listener behaviors. Figure 7.8 shows the percent of trials Victor engaged in listener behaviors during baseline and both VM conditions. During baseline, Victor’s mean percentage of orienting to the speaker was 17% (range, 10% to 30%). Mean percentage of orienting increased to 59% (range, 33% to 80%) and 32% (range, 0% to 50%) during the function-based and nonfunction-based conditions, respectively. Despite slight overlap, the overall differentiation in results suggests a probable functional relation between the function-based VM condition and orienting behavior for Victor. During baseline, Victor’s mean percentage of following directions was 31% (range, 20% to 43%). During VM, the mean percentage of following directions was 83% (range, 57% to 100%) and 63% (range, 20% to 100%) under function-based and nonfunction-based conditions, respectively. No clear relation between the VM conditions and following directions could be identified due to high variability during the nonfunction-based condition. Problem behavior. Figure 7.9 displays the percent of intervals Victor engaged in problematic behaviors during baseline and both VM conditions. His primary problem behavior was crying and yelling, though Victor also engaged in flopping, vocal and motor stereotypy, and property destruction. During baseline, Victor’s mean percentage of problem behavior was 37% (range, 30% to 50%). During VM, Victor’s mean percentage of problem behavior was 8% (range, 0% to 20%) and 43% (range, 20% to 70%) under the function-based and nonfunctionbased conditions, respectively. The differentiation in responding across conditions indicates a probable functional relation between function-based VM conditions and a reduction in problem behavior for Victor. 147 Percent of trials Victor oriented to speaker Baseline &##! Video Modeling <=583675">0:4?! @75A=583675">0:4?! -#! ,#! +#! *#! %#! )#! (#! '#! &#! Orienting #! ! "&#! &! '! (! )! %! *! +! ,! -! &#! &&! &'! &(! &)! &%! &*! &+! &,! &-! '#! '&! Percent of trials Victor followed directions &##! -#! ,#! +#! *#! %#! )#! (#! '#! &#! Following Directions #! ! "&#! &! '! (! )! %! *! +! ,! -! &#! &&! &'! &(! &)! &%! &*! &+! &,! &-! '#! '&! Sessions Figure 7.8. The percent of trials Victor oriented to the speaker (top panel) and followed directions (bottom panel) during baseline and both VM conditions. Shaded data indicate responding during function-based conditions and open data indicate nonfunction-based conditions. 148 149 Percent of intervals Victor engaged in problem behavior "$ %$ &$ '$ ($ )$ *$ +$ ,$ "#$ ""$ -./0123/!45678$ "%$ Video Modeling "&$ "'$ "($ ")$ 93/:./0123/!45678$ "*$ "+$ ",$ %#$ %"$ %%$ %&$ Sessions Figure 7.9. The percentage of intervals Victor engaged in problem behavior during baseline and both VM conditions. Shaded data indicates function-based VM conditions and open data indicates nonfunction-based conditions. !"#$ $ #$ "#$ %#$ &#$ '#$ (#$ )#$ *#$ +#$ ,#$ "##$ Baseline Morris Targeted and generalized responding. Results of VM on targeted mands for Morris are displayed in Figure 7.10. Morris demonstrated no responding during the baseline condition for either response target. During VM phase 1, Morris did not emit either the function-based or nonfunction-based response (M = 0%). During VM phase two, Morris’ mean percentage of responding was 42% (range, 0% to 80%) and 29% (range, 0% to 80%) for the function-based and nonfunction-based conditions, respectively. However, Morris’ responding was variable and he did not meet criterion to introduce additional word pairs. During VM phase 3, Morris’ mean percentage of responding was 40% (range, 0% to 80%) and 20% (range, 0% to 40%) during function-based and nonfunction-based conditions, respectively. Due to variability across both VM conditions, a functional relation between mand acquisition and VM was not identified for Morris. Follow-up data was not collected due to the completion of the school year. Listener behaviors. Figure 7.11 shows the percent of trials Morris engaged in listener behaviors during baseline and both VM conditions within the second phase. Listener behaviors were not recorded during the first VM phase as the video recordings from those sessions could not be recovered when the experimenter’s hard drive crashed. Morris’ mean percentage of orienting to the speaker was 28% (range, 13% to 50%) during baseline. During VM, Morris’ mean percentage of orienting was 43% (range, 25% to 60%) and 41% (range, 0% to 60%) during function-based and nonfunction-based conditions, respectively. Based on the similarity in means and high overlap between data series from each condition, no clear relation was identified between orienting and either VM condition for Morris. 150 151 Percent of trials Morris emitted target mand $ Video Modeling Phase 1 Nonfunction-based Function-based Video Modeling Phase 2 Video Modeling Phase 3 Sessions "$ %$ &$ '$ ($ )$ *$ +$ ,$ "#$""$"%$"&$"'$"($")$"*$"+$",$%#$%"$%%$%&$%'$%($%)$%*$%+$%,$&#$&"$&%$&&$&'$&($&)$&*$&+$&,$ Baseline Figure 7.10. The percent of trials Morris emitted target mands across all conditions and phases. Circles indicate function-based conditions and triangles indicate nonfunction-based conditions. !"#$ #$ "#$ %#$ &#$ '#$ (#$ )#$ *#$ +#$ ,#$ "##$ Percent of trials Morris oriented to speaker Baseline &##! -#! Video Modeling - Phase 2 <=583675">0:4?! @75A=583675">0:4?! ,#! +#! *#! %#! )#! (#! '#! Orienting &#! #! !"&#! &! '! (! )! %! *! +! ,! -! &#! &&! &'! &(! &)! &%! &*! &+! &,! &-! '#! '&! ''! '(! Percent of trials Morris followed directions &##! -#! ,#! +#! *#! %#! )#! (#! '#! &#! Following Directions #! "&#! ! &! '! (! )! %! *! +! ,! -! &#! &&! &'! &(! &)! &%! &*! &+! &,! &-! '#! '&! ''! '(! Sessions Figure 7.11. The percent of trials Morris oriented to the speaker (top panel) and followed directions (bottom panel) during baseline and both VM conditions within phase two. Shaded data indicate responding during function-based conditions and open data indicate nonfunction-based conditions. 152 During baseline, Morris’ mean percentage of following directions was 23% (range, 0% to 50%). During VM, mean percentage of following directions was 72% (range, 40% to 100%) and 75% (range, 50% to 100%) during function-based and nonfunction-based conditions, respectively. The increase in following directions when VM was implemented suggests a correlation between the intervention and the response. However, due to high data overlap, it is not clear whether a single condition or a combination of both conditions was responsible for the change in behavior. Problem behavior. Figure 7.12 displays the percent of intervals Morris engaged in problematic behaviors during baseline and both VM conditions within phase two. Problem behavior was not recorded due to the loss of video recordings. Morris’ primary problem behaviors were mouthing and vocal stereotypy, though he also occasionally engaged in flopping and motor stereotypy. During baseline, Morris’ mean percentage of problem behavior was 37% (range, 10% to 50%). During VM, Morris’ mean percentage of problem behavior was 26% (range, 10% to 90%) and 34% (range, 0% to 20%) during function-based and nonfunction-based conditions, respectively. Despite some variability, problem behavior decreased in a similar pattern across both VM conditions. No clear relation can be identified between the reduction in problem behavior from baseline levels and either of the VM conditions for Morris. 153 154 Percent of intervals Morris engaged in problem behavior "$ %$ &$ '$ ($ )$ *$ +$ ,$ "#$ "%$ Sessions ""$ -./0123/!45678$ "&$ "'$ "($ ")$ 93/:./0123/!45678$ "*$ "+$ ",$ %#$ %"$ %%$ %&$ Figure 7.12. The percentage of intervals Morris engaged in problem behavior during baseline and both VM conditions within phase two. Shaded data indicates function-based VM conditions and open data indicates nonfunction-based conditions. !"#$ $ #$ "#$ %#$ &#$ '#$ (#$ )#$ *#$ +#$ ,#$ "##$ Matthew Targeted and generalized responding. Matthew’s original intervention was adjusted to target a picture exchange topography after he did not demonstrate any vocal verbal responding during the initial intervention sessions. Figure 7.13 displays Matthew’s independent picture exchanges during baseline and both VM conditions. Matthew demonstrated no responding during the picture exchange baseline. During VM for the first picture exchange pair, Matthew’s mean percentage of picture exchange behavior was 73% (range, 20% to 100%) and 10% (range, 0% to 20%) under function-based and nonfunctionbased conditions, respectively. When the second word pair was introduced, Matthew’s mean percentage of picture exchange behavior was 85% (range. 60% to 100%) and 50% (range, 20% to 80%) for the function-based and nonfunction-based conditions, respectively. For the third word pair, Matthew’s mean picture exchange behavior was 93% (range, 80% to 100%) and 87% (range, 80% to 100%) for the function-based and nonfunction-based conditions, respectively. When the experimenter increased the distance Matthew had to travel to mand (i.e., phase two), mean percentage of picture exchange behavior was 73% (range, 40% to 100%) and 47% (range, 20% to 100%) during the function-based and nonfunction-based conditions, respectively. The results demonstrate a functional relation between the function-based VM condition and a rapid acquisition of target mands. 155 156 Percent of trials Matthew emitted target mand $ "$ &$ ($ *$ ,$ ""$ "&$ "($ "*$ Nonfunction-based Function-based ",$ %&$ Sessions %"$ Video Modeling - Phase 1 %($ %*$ %,$ &"$ &&$ &($ Generalization probes &*$ &,$ '"$ '&$ Video Modeling Phase 2 Figure 7.13. The percentage of trials Matthew emitted mands across all conditions and phases. Circles represent functionbased VM and triangles represent nonfunction-based VM. Open data represents generalization probes for the corresponding condition. Separated data series indicate training for a new word. !"#$ #$ "#$ %#$ &#$ '#$ (#$ )#$ *#$ +#$ ,#$ "##$ Baseline Listener behaviors. Figure 7.14 shows the percent of trials Matthew engaged in listener behaviors during baseline and both VM conditions. During baseline, Matthew’s mean percentage of orienting to the speaker was 91% (range, 63% to 100%). During VM, Matthew’s mean percentage of orienting was 87% (range, 57% to 100%) and 71% (range, 50% to 83%) during function-based and nonfunction-based conditions, respectively. Because Matthew demonstrated a high level of orienting during baseline and relatively consistent levels of responding across conditions, no relation between orienting and either VM condition was identified. Matthew’s mean percentage of following directions was 68% (range, 50% to 80%) during baseline conditions. During VM, Matthew’s mean percentage of responding was 90% (range, 75% to 100%) and 71% (range, 44% to 100%) under the function-based and nonfunctionbased conditions, respectively. Despite some overlap between the two conditions, the overall differentiation in responding indicates a probable functional relation between the function-based VM condition and following directions for Matthew. Problem behaviors. Figure 7.15 displays the percent of intervals Matthew engaged in problematic behaviors during baseline and both VM conditions. Matthew’s primary problem behavior was vocal stereotypy though he occasionally engaged in aggression and motor stereotypy. During baseline, Matthew’s mean percentage of problem behavior was 40% (range, 30% to 50%). During VM, Matthew’s mean percentage of problem behavior was 20% (range, 0% to 30%) and 30% (range, 0% to 50%) under the function-based and nonfunction-based conditions, respectively. No clear relation could be identified between the reduction in problem behavior and either VM condition due to high data overlap between the two conditions. 157 Video Modeling Baseline Percent of trials Matthew oriented to speaker <=583675">0:4?! &##! -#! ,#! +#! *#! %#! )#! (#! '#! Orienting &#! #! !"&#! Percent of trials Matthew followed directions @75A=583675">0:4?! &! '! (! )! %! *! +! ,! -! &#! &&! &'! &(! &)! &%! &*! &+! &,! &-! '#! &##! -#! ,#! +#! *#! %#! )#! (#! '#! &#! Following Directions #! !"&#! &! '! (! )! %! *! +! ,! -! &#! &&! &'! &(! &)! &%! &*! &+! &,! &-! '#! Sessions Figure 7.14. The percent of trials Matthew oriented to the speaker (top panel) and followed directions (bottom panel) during baseline and both VM conditions. Shaded data indicate responding during function-based conditions and open data indicate nonfunctionbased conditions. 158 159 Percent of intervals Matthew engaged in problem behavior $ "$ %$ &$ '$ ($ )$ *$ +$ ,$ -./0123/!45678$ ""$ "%$ "&$ "'$ 93/:./0123/!45678$ Sessions "#$ Video Modeling "($ ")$ "*$ "+$ ",$ %#$ Figure 7.15. The percentage of intervals Matthew engaged in problem behavior during baseline and both VM conditions. Shaded data indicates function-based VM conditions and open data indicates nonfunction-based conditions. !"#$ #$ "#$ %#$ &#$ '#$ (#$ )#$ *#$ +#$ ,#$ "##$ Baseline CHAPTER 8 DISCUSSION The primary purpose of Experiment 2 was to identify the effects of function-based and nonfunction-based VM on mand acquisition and generalization for young children with autism. An additional purpose was to examine collateral effects of the VM conditions on behaviors correlated with mand acquisition including orienting to the speaker, following directions, and problem behaviors. The present chapter discusses the results of the experiment within the context of each experimental question as well as the previous research on video modeling. Implications for practice, limitations of the experiment, and suggestions for future research are also discussed. Intervention Research Questions Question 2 To what extent does the implementation of function-based and nonfunction-based mand training procedures affect vocal or alternative (e.g., picture exchange) verbal behavior acquired and emitted by a small group of children with autism? Prior to the experimental conditions, participants were not observed or reported to emit any mands using a conventional communication system. Baseline measures confirmed the absence of targeted mand topographies or approximations for all participants. When VM was introduced, four out of five participants acquired three novel mand topographies during the function-based condition, when the environmental variables replicated the assessment conditions most likely to produce gestural behavior. However, none of the participants acquired mands during nonfunction-based conditions, when the programmed variables were not consistent with the results of the FA. These results are similar to Carr and Durand’s (1985) original examination of FCT as a 160 treatment to reduce disruptive behavior; the functionally equivalent response led to mand acquisition while the unrelated response did not. Though this phenomenon has been extensively replicated across participant populations (Durand & Kishi, 1987; Dunlap et al., 2006; Frea & Hughes, 1997), types of problem behaviors (Fyffe, Kahng, Fittro, & Russell, 2004; Mace & Lalli, 1991; Wacker et al., 1990), and types of research settings (Hagopian, Fisher, Sullivan, Acquisto, & LeBlanc, 1998; Kemp & Carr, 1995; Lalli, Casey, & Kates, 1995), the present investigation extended the principle of functional equivalence to mand training without an emphasis on problematic behavior. This suggests FCT can be delivered proactively to children who demonstrate deficits in verbal behavior but who do not engage in severe or frequent problem behavior. Fuller and Bailey. As Fuller and Bailey demonstrated the capacity to emit words during pre-experimental observations, it was not surprising that they could emit vocal mands. However, the differential responding between function-based and nonfunction-based conditions may demonstrate the benefit of a pretreatment FA prior to mand training. As the intervention components were identical across conditions, the absence of mands during the nonfunction-based condition was attributable to EOs rarely being captured or contrived and the inability of the programmed consequences to reinforce any responding that did occur. These errors are avoidable when the EOs and reinforcers that evoke and maintain gestures are identified prior to mand training. Interestingly, both participants demonstrated rapid acquisition of the first and second function-based mands yet demonstrated a delay to acquisition of the third response. The reason for this delay was not clear and could have occurred for several reasons. First, it was possible that the pattern was a coincidence and occurred due to MOs not known by the experimenter. This 161 seems unlikely as the procedures for contriving the EO were not changed and they appeared to be extremely effective once the participants started emitting the target response. Though an AO prior to the experimental session could explain this pattern, teachers of both participants confirmed that items used during intervention sessions were not accessed in the classroom; satiation to the experimental stimuli was therefore unlikely. Another explanation was that the third function-based mand was more difficult to pronounce than the first two. Typically, an utterance with more sounds requires more effort to emit than an utterance with fewer sounds (Tager-Flusberg et al., 2009). Thus, it is possible that the increase in response requirement from two to three words (i.e., “I want [insert specific item].”) for Fuller’s third mand required additional training sessions before he could emit the response. However, Bailey’s third mand was a single word (“movie”) and she had previously demonstrated the ability to emit a two-word utterance (“help me”). Therefore, it seemed unlikely that the third word was more difficult to pronounce than previous words. A third possible explanation was that the participants required additional trials to discriminate the stimulus conditions as the size of their mand repertoire increased. Such an explanation would be consistent with previous hypotheses that children with autism may struggle to emit words consistently as their vocabulary increases (Jennett et al., 2008; Tager Flusberg et al., 2009). This explanation seems likely as both participants rapidly acquired the target response once they came in contact with the reinforcers. This suggests the evocative event was an effective EO and that participants were able to emit all of the sounds within the new responses. The patterns demonstrated by Fuller and Bailey represent optimal outcomes for participants in the current investigation. They learned to emit three mands in a variety of situations that others could understand. As videos and prompts were not necessary to evoke the 162 response, the mand was under pure control of the EO. This is an extremely important component of mand training as it allows an individual to mand in a variety of environments (Sundberg & Michael, 2001). Victor and Matthew. Unlike Fuller and Bailey, Victor and Matthew provided no previous evidence that they would be able to emit clear vocal words during experimental sessions. Both participants emitted vocalizations (e.g., “aahhh” “eeehhh”) during preexperimental observation but clear words were not observed or reported by teachers or parents. Because of this uncertainty, identifying items or events that functioned as reinforcers for gestures offered valuable information to the researchers when evaluating the effects of the intervention for each participant. In Victor’s case, the reinforcers and EOs functionally related to gestural behavior were essential for teaching vocal mands. The reliability of the EO to evoke an approximation of the target response provided a clue as to the likelihood Victor would be able to emit a more complex response. The experimenter could assume that differential reinforcement of successive approximations of the target response would be successful (Sundberg, 2007). Additionally, immediate success during early training sessions is more likely to coincide with ongoing compliance during future sessions (Koegel et al., 2009; Sundberg & Michael, 2001). Perhaps most importantly, reinforcing any approximations Victor made was critical as his behavior during the nonfunction-based condition showed that vocalizations would vary and could disappear if not continuously reinforced. For Matthew, the absence of vocal responding during all VM conditions indicated a change in procedures would be necessary. Although it was possible that EOs were not contrived during VM sessions, this conclusion seemed unlikely as deprivation from attention was an 163 effective EO during FA. Other explanations for the absence of vocal responding were that Matthew could not produce vocal speech or that VM was not an effective intervention for him. As care providers reported no previous language and Matthew attended to the videos when they were shown during the intervention, the experimenter assumed VM could be effective and that Matthew was likely unable to emit vocal speech at the time of treatment. Therefore, picture exchange was selected as an alternative topography for mand training. Matthew’s immediate acquisition of the function-based response when picture exchange was implemented supported the experimenter’s assumption. The results showed that deprivation from attention could evoke mands for attention and that using VM to teach a conventional mand topography was an effective intervention for Matthew. Therefore, the absence of vocal mands during phase one was most likely a result of Matthew not being able to speak. The procedures for Victor and Matthew demonstrate the benefit of pretreatment FA for making data-based decisions before and during treatment. The experimenter could rely on previous evidence that the antecedent and consequent manipulations for each participant were more likely than other variables to evoke and maintain the target response. When Victor demonstrated partial rather than full vocal responses during early VM sessions, the results of the FA indicated that the absence of the full response was more likely due to an emerging capacity for speech and not inadequate consequences for the approximations. Similarly, when Matthew did not emit vocalizations during VM, the results of the FA helped suggest that an alternative topography was likely necessary. The FA therefore reduced the chance that an intervention would be abandoned too early or sustained when it was not likely to be effective. The results for Victor and Matthew offer two systematic replications of the functionbased VM intervention originally applied to Fuller and Bailey. In Victor’s case, the results show 164 that VM can be extended to teach vocal speech to individuals with autism who have not previously demonstrated this capacity. Matthew’s results demonstrate the extension of VM to teach pictorial exchange as an alternative communication system. Picture exchange is traditionally taught using one adult to prompt the target response and another adult as the communicative partner (Frost & Bondy, 2002). However, Matthew was able to rapidly acquire mands after observing the model and a second adult was not needed to teach this skill. Eliminating the second adult is important from a practical perspective as many public education programs for children with autism do not have the resources to dedicate two educators to a single child (Stahmer, 2007). Morris. A decision making process was also necessary for Morris because he did not emit mands when VM was implemented. Unlike Matthew, Morris’ teacher and parents reported that he had emitted clear vocal words in the past. The reported capacity for speech suggested he was capable of emitting the target response and that several attempts to teach vocal speech should be made before selecting another topography (Greer & Ross, 2008; Sundberg & Michael, 2001). The FA demonstrated that inaccessible stimuli and adult assistance could evoke and maintain mands in the form of gestural behavior. Therefore, the absence of vocal responding during the first phase of VM suggested antecedent stimuli did not sufficiently signal the availability of reinforcement for the target response or the target response was not explicitly modeled for Morris. The second VM phase for Morris was constructed to address the assumption that antecedent teaching procedures needed to be adjusted in order to establish a verbal repertoire. Evidence from pre-experimental classroom observations suggested Morris did not initiate interactive episodes with others but he did engage reciprocally once another person attempted to 165 engage him. For example, when a teacher tickled Morris and then stepped away, Morris would approach and look toward the teacher while pulling his arms and elbows in close to his body as if he was expecting to be tickled again. This pattern was consistent with developmental and behavioral research suggesting many children with autism respond to initiations of others but do not reliably initiate interactions (Schertz & Odom, 2007; Taylor & Hoch, 2008). Therefore, the experimenter showed Morris a video and then initiated an interaction, rather than waiting for Morris to initiate the interaction. A second aspect of the decision making process for Morris was based on Skinner’s (1957) theory that a specific verbal response was more likely to occur when under the control of multiple antecedents. During phase one, any response emitted during VM would be part mand as it was controlled by deprivation. A response would also be part echoic as the target response was the same as the word emitted by the model in the video clip. The experimenter therefore included an intraverbal prompt in the second phase to increase the probability of Morris emitting the target response. As Morris emitted approximations and complete vocal verbal targets during the second phase, the inclusion of an intraverbal prompt delivered by the experimenter appeared to be an effective manipulation. However, responding occurred during both VM conditions and it was highly variable thereby limiting any firm conclusions that could be drawn about the phase two conditions. When it was determined that viewing the video clips was Morris’ most preferred activity, a delay between an incorrect response and the next trial was implemented during a third VM phase to increase the probability of a correct response. If Morris was behaving to obtain access to video clips as quickly as possible, the delay would mean that correct responses provided more immediate access to videos than incorrect responses. However, no change in 166 Morris’ pattern of responding was observed after this manipulation. Though several extraneous variables such as Morris’ young age or frequent absences from school might explain his behavior, it seemed most likely that Morris’ preference for viewing videos of peer models made him a poor candidate for a VM intervention. He could access his most preferred activity noncontingent on emitting the target response. Had time allowed, an ideal manipulation would have involved an additional phase with live modeling to teach Morris to mand for access to video clips. That way, he would only obtain a high preference activity after emitting the target response. However, previous phases had been administered through the end of the school year in mid-June and Morris did not attend the summer program; he was therefore not available for additional experimentation. Question 3 To what extent does the implementation of function-based and nonfunction-based mand training procedures lead to generalized responding? The generalization demonstrated by Fuller, Bailey, and Matthew is a very important contribution of the present investigation as children with autism often fail to generalize trained skills to novel settings (Bodfish, 2004; Greer & Ross, 2008; Halle & Meadan, 2007; Tager Flusberg et al., 2009). These participants emitted mands when presented with evocative events in their classrooms, where a mand repertoire is critical for an individual to control delivery of reinforcers (Sundberg, 2007). The three participants could mand for assistance or attention from a variety of adult listeners and Fuller asked peers for help operating preferred toys in the classroom. In some cases, mands were generalized to conditions that were very dissimilar to training sessions. For example, Fuller’s teacher reported that he approached an adult after coming out of the bathroom while holding the unbuttoned flaps of his pants and said “help me.” 167 Fuller and Bailey both demonstrated a high level of response maintenance across all follow-up sessions. Fuller’s only error occurred when a preferred item was placed on a high shelf and he manded for the experimenter’s help rather than for the specific item. Though this was scored as an error, it is important to note that Fuller generalized the response “help” to contextually appropriate conditions. Bailey’s only error during follow-up conditions occurred when the experimenter gave her a preferred stimuli inside of a sealed container and she emitted the response “movie” multiple times instead of emitting the response “open.” Though this was scored an error, it seemed likely Bailey was manding to watch a movie and was therefore less likely to mand for access to stimuli in the container. The results for Fuller and Bailey align with previous VM research and collectively demonstrate long response maintenance following VM (MacDonald, Sacramone, Mansfield, Wiltz, & Ahearn, 2009; Nikopoulos, Canavan, & Nikopoulou-Smyrni, 2009; Nikopoulos & Keenan, 2007). Victor may have demonstrated a different type of generalization based on his pattern of responding during nonfunction-based mand training for the second word pair. As Victor initially emitted but did not reliably maintain the nonfunction-based mand, he may have generalized imitation of the peer model but his imitative behavior was not maintained, probably due to the absence of reinforcement for the vocal response (Cooper et al., 2007; Koegel & Rincover, 1977). Generalized imitation may have also occurred in the classroom for Victor as he began echoing the verbal behavior of his teacher and classmates including saying “hi Josh” and “bye Josh” to the experimenter when he entered and exited the classroom. If Victor did learn to imitate during the function-based condition, this has important implications for VM. At present, it is assumed that an imitative repertoire is a likely prerequisite to learning through VM rather than a skill that can be acquired during VM (McCoy & Hermansen, 2007). 168 The absence of generalized responding for Morris should be considered in light of the fact that he did not meet acquisition criteria during training. The video clip was therefore not faded from Morris’ training sessions, which would make any generalized responding quite surprising. Ultimately, a response that does not reliably occur during a highly controlled training session is not likely to occur in a different setting (Cooper et al., 2007). Overall, several types of generalization were observed in the present investigation and each type was likely the result of a different feature of the experimental procedures. Response maintenance was likely a result of selecting behaviors that would come into contact with reinforcing consequences in the natural environment. Most listeners within a verbal community reinforce mands by delivering the specified consequence (Skinner, 1957). Therefore, mands can be maintained in a number of contexts, which also increases the likelihood of setting generalization (Stokes & Baer, 1977). Another feature of the present investigation that likely affected setting generalization was using common stimuli during training sessions (Stokes & Baer, 1977). Each participant’s teacher agreed to be the adult listener in at least one of the video clips. The teachers would be the common stimuli as they could also function as a listener for the student in the classroom environment. In addition to setting generalization, Fuller, Bailey, and Matthew demonstrated situation generalization by emitting targeted mands when presented with evocative events and stimuli that were not used during training. It is likely that using multiple exemplars during training sessions supported this type of generalized responding in the classroom (Stokes & Baer, 1977). By learning to mand under a variety of evocative conditions, participants were more likely to learn that a mand, such as “help,” produces adult assistance in a variety of situations. 169 Another potential reason for generalization in the present investigation is the sequence of antecedent variables that occurs when teaching mands using VM. Most mand training procedures include capturing or contriving a MO followed by a response prompt to evoke the target behavior (e.g., Hall & Sundberg, 1987; Jennett et al., 2008). In observational teaching interventions, including VM, the modeled sequence can eliminate the need to deliver a response prompt. As a result, the target behavior may come under direct control of the MO during observational teaching procedures whereas it comes under multiple control of the MO and prompt during typical instructional procedures. This difference may impact the likelihood that the MO evokes the behavior in untrained situations and after explicit instruction is terminated. Question 4 To what extent does the implementation of mand training procedures targeting vocal or alternative verbal behavior affect listener behaviors correlated to the acquisition of targeted behavior (i.e., orienting to the speaker and following directions)? Consistent with previous research, listener repertoires were slightly improved as participants acquired mands (Koegel et al., 1988). However, it is important to note that all participants would require additional listener training in order to ensure these behaviors reliably occurred. These outcomes show that it is not necessary for children to reliably demonstrate skills such as orienting and following directions prior to learning language or alternative forms of verbal behavior. This means that mands can be taught prior to other verbal responses, which may increase a child’s overall compliance with language training and may ultimately serve to make it easier to teach listener skills (Sundberg & Michael, 2001). At the same time, mastery of listener responses is necessary for a comprehensive verbal repertoire (Greer & Ross, 2008) and the results of the present investigation suggest these skills may need to be explicitly targeted. 170 Question 5 To what extent does the implementation of function-based and nonfunction-based mand training procedures targeting vocal or alternative verbal behavior affect problem behaviors emitted by participants? For Fuller, Bailey, and Victor, the lower levels of problem behavior during the functionbased condition was consistent with previous research showing that problem behavior can be reduced by teaching children functionally equivalent mands (Dunlap et al., 2006; Kelley et al., 2002; Lalli et al., 1995). However, an FA of problem behavior was not conducted in the present investigation and the focus of the intervention was not on teaching mands that could function as alternatives to problem behavior. Instead, the results show that the acquisition of mands had a collateral impact on problem behavior. It has long been assumed that communication training affects untargeted problem behavior, though few investigations have demonstrated a functional relation between these variables (Charlop-Christy et al., 2002). As Morris did not acquire mands during either condition, it was not surprising that his problem behaviors were not differentiated across conditions. In fact, his pattern of responding provides additional support for the theory that mand acquisition can have collateral effects on problem behavior. Since the general components of the two conditions were the same, problem behavior would be expected to occur at a similar rate during both conditions. However, when one condition involves motivative variables that lead to mand acquisition, problem behavior should reliably decrease if mand acquisition has a collateral impact on problem behavior. If Morris demonstrated differential levels of problem behavior across the two conditions despite not acquiring mands, a variable other than mand acquisition would have to be responsible for the change in problem behavior. 171 For Matthew, it is possible that his problem behavior might have served a different function than the other participants and was therefore not changed by the acquisition of mands. The primary problem behaviors for Fuller, Bailey, and Victor were crying and yelling. These behaviors are often used to control the delivery of reinforcement mediated by others when children are not able to emit conventional mands (Sundberg, 2007). Thus, vocal mand acquisition was likely to impact these behaviors. However, the primary problem behavior for Matthew was vocal stereotypy. This behavior is often automatically reinforced by the sensory stimulation it provides (Ahearn et al., 2007; Bodfish, 2004). Thus, the acquisition of mands was not likely to alter Matthew’s problem behavior. The reduction of problem behavior when mands are acquired likely works in one of two ways. Similar to FCT, it is possible that targeted mands are functionally equivalent to problem behavior, which becomes unnecessary once the mand is acquired (Tiger et al., 2008). If this were the case, decreases in problem behavior would be limited to situations where the selected mand happens to be functionally equivalent to the problem behavior. Another possibility is that the reinforcers delivered during mand training are of higher quality or occur with greater frequency than reinforcers delivered for problem behavior. Learning a skill that obtains high quality or frequent reinforcement could function as an AO that reduces the value of other reinforcers (Michael, 2007). For example, attention could maintain problematic behavior in an environment that has minimal stimuli. However, once a child can ask for a toy to play with, the environment has changed and the value of attention as a reinforcer could decrease (Michael, 2007). This explanation would suggest a more global change in behavior that occurs as a result of a child controlling the delivery of various reinforcers (Sundberg, 2007). Unfortunately, the present investigation sought to examine the collateral effects of mand 172 training on problem behavior rather than the variables responsible for a reduction in problem behavior as a result of mand training. Nevertheless, mand training can clearly reduce problem behavior (Carr & Durand, 1985; Charlop-Christy et al., 2002; Dunlap et al., 2006). These results therefore support the suggestion that successful mand training is a critical component of early intervention for children with autism (Greer & Ross, 2008; Sundberg, 2007; Sundberg & Michael, 2001). Video Modeling The results of the function-based VM intervention were consistent with several previous applications of VM for children with autism (Charlop-Christy et al., 2000; MacDonald, Sacramone, Mansfield, Wiltz, & Ahearn, 2009; Sherer et al., 2001). Participants rapidly acquired targeted skills and generalized those skills to additional settings and situations. The results of nonfunction-based VM are less consistent with the majority of previous research as participants did not acquire target responses. These differentiated results have important implications for the process of observational learning at a theoretical level. The differences in outcomes between VM conditions demonstrate a potential mechanism by which observational learning – and VM as an observational learning methodology – leads to the acquisition of new behaviors for some children. Previous hypotheses suggest similarities in physical characteristics between the observer and the model impact observational learning outcomes (Bandura, 1977; Buggey, 2005; Darden-Brunson et al., 2009). However, despite using the same model across word pairs, participants in the present investigation acquired functionbased mands and did not acquire nonfunction-based mands. This suggests that observational learning was affected by factors other than physical characteristics. The primary difference between function-based and nonfunction-based video clips was 173 the inclusion of consequences known to reinforce gestural behavior within the function-based video clip. The fact that participants emitted function-based but not nonfunction-based mands supports a less commonly referenced tenant of social learning theory: observation of a target behavior and the consequence for that behavior impact observational learning (Bandura, 1977). Though previous VM research has emphasized structural variables (e.g., age, race, and gender of the model or point of view), the effect of observed reinforcement contingencies on response acquisition is a critical area for further investigation. An additional finding that has important implications for VM research was that participants acquired function-based mands despite no previous evidence of imitating the behavior of others. Some suggest imitation is a prerequisite to VM training though this has been a source of uncertainty in the VM research (McCoy & Hermanson, 2007; Rayner et al., 2009). This discrepancy may be related to the inclusion of observed consequences that are likely to function as reinforcers in the present study and the fact that the same variable has not been considered in previous investigations. When preferred consequences are not included in the modeled sequence, imitative repertoires could be a prerequisite for VM as the participant would be unlikely to emit the modeled behavior if imitation had not been previously conditioned. However, the present investigation shows that children who do not imitate can learn through VM, though observed consequences become a critical component of the intervention. Implications for Practice The results of the present investigations have important implications for practice in educational settings. First, these results should be considered in light of the recent emphasis on EBPs for children with autism (NSR, 2009). Though VM has been identified as an empirically validated practice, the present investigation demonstrated conditions under which this practice 174 could be both effective and ineffective for an individual participant. This difference is an important reminder that EBP for children with autism consists of selecting a practice with empirical support and using professional judgment to individualize the practice so that it works for a specific child (Detrich, 2008; NSR). At present, educational research and policy has focused on identifying practices that are research validated in order to support the selection process (NSR). Additional support will be needed for the individualization process as educators begin to reliably select EBPs for children with autism. Pretreatment FA, as demonstrated in experiment one, is a methodology that can be used to empirically inform the individualization of educational interventions. A second implication is the general utility of VM as an instructional methodology. Once a video clip was created, it could be used for any participant who could emit the modeled topography and for whom the modeled consequence was likely to function as a reinforcer. Though creating a video may require more up front effort than providing a prompt or live model, the potential for long-term use may be very beneficial to educators. Additionally, all participants appeared to enjoy viewing video clips as they willingly attended to the iPhone! and often smiled or laughed while watching videos. This may improve participants’ ability to focus on salient features of the modeled sequence and may have increased overall compliance with training sessions. A related consideration for practice is the emergence of technology that can enhance instruction for children with autism. Numerous applications recently developed for smartphones or minicomputers allow educators to create augmentative communication systems, picture activity schedules, and reinforcement systems on devices such as the iPhone!. These devices can also be used to access the internet, take pictures, record video, and monitor progress: all of 175 which are convenient if not necessary components for delivering educational interventions to children with autism. Though future research is needed to determine how best to use smartphones and minicomputers in educational settings, it is apparent these tools can be powerful devices for delivery of instruction (as in the present investigation) and can likely improve the organization of educational programming for students with intensive needs. Limitations and Future Research There are several limitations of the present investigation that could be addressed by future research. First, function-based mands were often compared to mands for which the specified consequence was not likely to have a reinforcing effect (e.g., mands for attention though attention did not reinforce gestures during the FA). When the specified consequence of a mand is not likely to be reinforcing for an individual child, interventionists typically develop procedures to increase the value of consequences such as providing unrelated rewards (Taylor & Hoch, 2008). However, in the present investigation the consequence specified by the child’s mand was the only consequence delivered in both VM conditions. Thus, if a participant emitted a mand during many of the nonfunction-based conditions, it was unlikely to be maintained. In order to approximate more typical mand training procedures, future research should compare function-based and nonfunction-based mand training when unrelated rewards are paired with the specified consequence for any mands that occur during nonfunction-based sessions. Such a comparison would require the fading of unrelated rewards to ensure the nonfunction-based mand was eventually under the control of a MO and specific reinforcement. A second limitation was the absence of experimental control over Morris’ responding during VM conditions. As a result, it is impossible to determine why a validated intervention did not lead to reliable response acquisition. It is possible that Morris was not a good candidate for 176 VM as observing the video was a highly preferred consequence. Rather than functioning as a discriminative stimulus, viewing video clips may have inadvertently functioned as a reinforcer for a behavior or series of behaviors that occurred prior to the video display. Another possibility is that Morris would have demonstrated consistent response acquisition with more consistent school attendance and exposure to the intervention conditions. Though the results for Morris do not negate the clear functional relation identified for other participants, a potential boundary of that functional relation cannot be known as the source of Morris’ variability was not determined. A third limitation is that it is impossible to know which experimental variables were responsible for changes to listener and problem behaviors. Because conditions were designed to isolate components of VM responsible for mand acquisition (i.e., function-based versus nonfunction-based antecedents and consequences) changes in listener or problem behaviors could only be attributed to the broad conditions rather than to a specific component that might influence those responses. This limits the ability to determine why listener behaviors changed for some participants but not others or why a problem behavior decreased but was not be completely suppressed. Future researchers might develop methods to carefully examine the collateral effects of mand training on other behaviors. It could be that mand acquisition leads to changes in those behaviors or that the conditions affect other behaviors regardless of mand acquisition. An important area for future research is to determine how public school personnel can best utilize the procedures applied in the present investigations. It is premature to claim that the function-based approach can be broadly applied as the experimenter, who had extensive training in special education and ABA, implemented the interventions in the present investigation. Some aspects of the intervention may need to be modified for implementation by school personnel. However, the modification of procedures could alter the effectiveness of the approach. Thus, 177 future research will need to identify an implementation process before the function-based approach to behavioral acquisition can be adopted within public school settings. Two additional areas for future research include the potential of VM interventions to help children with autism learn to imitate or acquire an observational learning repertoire (Greer, Singer-Dudek, & Geautreaux, 2006). For example, it is possible Victor’s acquisition of an echoic repertoire was a result of learning to imitate during VM. After learning to reproduce modeled behavior to obtain a specific consequence, Victor might have generalized imitation to situations in which reinforcement was not observed or directly delivered for each response. Future research could examine the potential of VM for teaching children with autism to imitate rather than imitation being a prerequisite to learn from VM (McCoy & Hermansen, 2007). Similarly, Matthew’s teacher reported that he demonstrated increased attending to peers during group sessions and engaged in the same behaviors as his peers such as walking to the front of the classroom when his name was called and selecting a leisure activity. Prior to the VM intervention, Matthew was reported to never observe his peers and waited for direct instruction from adults before selecting a leisure activity. These outcomes suggest it might be possible for VM to not only teach the targeted skill, but also lead to the acquisition of an observational learning repertoire for children with autism. If so, additional research is needed to understand how VM instruction can consistently produce these important outcomes. The present investigation extends the use of function-based interventions to promote the acquisition of a mand repertoire for children with autism. Instead of waiting until problem behavior becomes frequent or severe, intervention agents can apply functional assessment tactics to behaviors that children often use as verbal operants and teach new response topographies based on the results of such an assessment. This approach may improve the efficiency of mand 178 training for children with autism and could prevent the development or worsening of problem behavior that children often use to mand for preferred consequences. The present investigation also offers an illumination of the components that might make VM an effective intervention. Specifically, children may learn to emit new behaviors based on the observed responsereinforcer relation. As demonstrated in the present investigation, such a phenomenon would suggest VM may be ideal for teaching mand training to early language learners. Further, mand training using VM could have long-term effects on a child’s ability to learn through observation. Most importantly, the investigations described in this dissertation directly improved socially significant behaviors for the participants. Children who one day had no method of communication to ask for the things they wanted were speaking the very next day. Through vocal speech or picture exchange, participants learned to ask numerous adults, and an occasional peer, for assistance or attention under a variety of situations. For three participants, the acquisition of a vocal verbal repertoire also led to a collateral decrease in problematic behaviors. If these results can be replicated across settings, intervention agents, and participants, the assessment and intervention methods used in the present investigation may be powerful tools for teaching new skills to children with severe disabilities. 179 APPENDIX A OBSERVER TRAINING PROCEDURES 180 Training Observers to Collect FA Data 1. Materials: a. Tally counter b. Clipboard c. Pencil & Data collection sheet with section for each condition 2. Didactic Training Procedures: a. Review materials b. Describe conditions observers will be observing c. Describe target behaviors d. Describe coding procedure for video file names (i.e., condition and TB codes) e. Show a 1-m video sample for each condition f. Show participants what data looks like once recorded 3. Hands-on Training Procedures: a. Complete all preliminary fields on data sheet (student name, date, target behavior, and FA condition) b. Set tally counter to “0000” c. Observe one session from each condition and have trainee say “stop” any time a TB occurs. Discuss any disagreements between experimenter and observer during this process. d. Trainees observe for entire session and record frequency of target behaviors within the condition e. Compare results to primary experimenter f. Repeat for all conditions until 90% agreement is demonstrated across all conditions for 3 consecutive sessions. 4. Sample of FA data collection sheet Student Name: ________________ Target Behavior: _______________________________________ B034!C! D4::675! &! '! (! )! E75?$!C! F65! 2335$I! ./01I! B4J$I! 283$I! G73! GH! ! ! ! ! B034!C! D4::675! %! *! +! ,! E75?$!C! F65! 2335$I! ./01I! B4J$I! 283$I! G73! GH! ! ! ! ! B034!C! D4::675! -! &#! &&! &'! E75?$!C! F65! 2335$I! ./01I! B4J$I! 283$I! Function(s) of behavior: ____________________________________ 181 G73! GH! ! ! ! ! Training Observers to Collect VM Data 1. Materials: a. Clipboard b. Pencil & Data collection sheet 2. Didactic Training Procedures: a. Review materials b. Describe VM conditions observers will be observing c. Describe target behaviors d. Describe coding procedure for video file names (i.e., condition and TB codes) e. Show a 3-m video sample for each condition (vocally identify start of trial, student response, end of trial) f. Show participants what data looks like once recorded 3. Hands-on Training Procedures: a. Complete all preliminary fields on data sheet (student name, observer name, date, target behavior, and VM condition) b. Observe one session from each condition and have trainee say “stop” when the trial begins and ends and also if a TB occurs. Discuss any disagreements between experimenter and observer during this process. c. Trainees observe for entire session and record correct or incorrect responses for each trial as well as any responses occurring prior to the experimenter showing video. d. Compare results to primary experimenter e. Repeat for all conditions until 90% agreement is demonstrated across all conditions for 2 consecutive sessions. 4. Sample of data collection sheet TB1 Function-based _______________ Nonfunction-based ______________ TB2 Function-based _______________ Nonfunction-based ______________ TB3 Function-based _______________ Nonfunction-based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`! ?4/6V4N1!7A!875:4a=4584!A7N!30N`43! >4U0V67N! ! ! [4:!^^^^^^^!!!!!!!@7!^^^^^^! ! [4:!^^^^^^^!!!!!!!@7!^^^^^^! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! 9X;4N6J4534N!J7V4:!08N7::!N77J!AN7J! 8U6/?!0A34N!J0]65`!:3034J453! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! S`57N4:!0//!57530N`434?!>4U0V67N:!7N! /6`U3/1!N4?6N483:!?05`4N7=:!;N7>/4J! >4U0V67N! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! F0]4:!V780/!87JJ453!05?!/6`U3! ;U1:680/!6534N083675!0A34N!30N`43! >4U0V67N! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! G4NJ65034:!:4::675!0A34N!&#!J65! ! [4:!^^^^^^^!!!!!!!@7!^^^^^^! 215 Procedural Integrity Checklist for Escape Condition Date Observer Participant Experimenter Josh Indicate “Yes” or “No” for behaviors that occur one time For behaviors that can occur multiple times, mark “+” if experimenter behavior is observed for each opportunity and “-“ if not observed for each opportunity -./"0($%&+',#,&+$ S5:3N=83:!:3=?453!37!:63!75!A/77N!7N!65! 8U06N!65!AN753!7A!30>/4! B4/6V4N:!;N7`N0JJ4?!053484?453! :36J=/=:!M6$4$L!575;N4A4NN4?!371L! 875:3053!0335L!7N!30:]!?4J05?O!37!:30N3! 05?!0A34N!>N40]:! ! ! [4:!^^^^^^^!!!!!!!@7!^^^^^^! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! S`57N4:!0//!57530N`434?!>4U0V67N:!7N! /6`U3/1!N4?6N483:!?05`4N7=:!;N7>/4J! >4U0V67N! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! E7J;/434:!N4:;75:4!;N7J;3!U64N0N8U1! 6A!5484::0N1!MA7N!F7NN6:!C!F033U4T! 75/1O! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! b4:;75?:!37!30N`43!>4U0V67N!>1! N4J7V65`!053484?453!:36J=/=:!A7N!'#!:! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! S5:3N=83:!:3=?453!37!N43=N5!37!7N6`650/! /7803675!6A!5484::0N1!A7//7T65`!>N40]! ! G4NJ65034:!:4::675!0A34N!&#!J65! ! [4:!^^^^^^^!!!!!!!@7!^^^^^^! 216 Procedural Integrity Checklist for Play Condition B034! W>:4NV4N! .0N3686;053! 9X;4N6J4534N! ! ! ! Y7:U! S5?68034!Z[4:\!7N!Z@7\!A7N!>4U0V67N:!3U03! 788=N!754!36J4! <7N!>4U0V67N:!3U03!805!788=N!J=/36;/4!36J4:L! J0N]!ZR\!6A!4X;4N6J4534N!>4U0V67N!6:! 7>:4NV4?!A7N!408U!7;;7N3=5631!05?!Z"Z!6A!573! 7>:4NV4?!A7N!408U!7;;7N3=5631! 12"3$%&+',#,&+$ S5:3N=83:!:3=?453!37!:63!?7T5!75!A/77N! 7N!65!8U06N!65!AN753!7A!30>4/! c0N6431!7A!;N4A4NN4?!634J:!T63U65! :3=?453!N408U! d6V4:!8U6/?!033453675!M/0>4/!634J!C! ;N06:4O!4V4N1!'#:!=5/4::!30N`43!7N! ;N7>/4J!>4U0V67N!e=:3!788=NN4?! ! ! [4:!^^^^^^^!!!!!!!@7!^^^^^^! ! [4:!^^^^^^^!!!!!!!@7!^^^^^^! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! f063:!37!?4/6V4N!033453675!A7N!%!:! A7//7T65`!30N`43!7N!;N7>/4J!>4U0V67N! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! K6`U3/1!N4?6N483:!?05`4N7=:!;N7>/4J! >4U0V67N! ! G4NJ65034:!:4::675!0A34N!%!J65! [4:!^^^^^^^!!!!!!!@7!^^^^^^! 217 APPENDIX J EXAMPLE OF VIDEO MODELING PROCEDURES 218 Student: Fuller Classroom: JR, PM Target Behavior: Function-based mands: 1. Help me, 2. Open this, 3. I want _______ Nonfunction-based mands: 1. Come play, 2. Break, 3. Look Purpose of Program: Teach vocal mands. Conditions: (5 trials per session) FB – Function-based Video Modeling NFB – Nonfunction-based Video Modeling Session Sequence: Complete conditions in order listed. Session One – FB / NFB Session Two –FB / NFB Session Three – FB / NFB Session Four – NFB / FB Materials: Data Collection: Record whether participant emits target behavior for each trial If correct, record + If incorrect, record % If correct w/o video, record & • • • • iPhone or iPod Touch High or Low Preference stimuli Clipboard with data sheet Work tasks (puzzle, sorting, stackers, etc) Procedure: Experimenter Behavior Procedure: Student Behavior High Pref: Croaking frog, tops, balloon inflater Medium Pref: Helicopter Low Pref: Suspected disliked: Music toys, streamers Level 1: Immediate Video • Instruct student to sit at table. • • • Comply with student mand Student emits target mand Student does not emit mand or After clip is completed, show student highly preferred item that they cannot access (function-based) or low preference item and instruct student to interact with item (nonfunction-based) then step back and wait 20 s • Student watches movie • • • • Show student iPhone and tell student to “watch the movie” Show movie clip that corresponds with target response Student sits at table 219 • Wait 10 s and start another trial (i.e., instruct student to watch the movie) • following access to reinforcer Complete 5 trials Level 2: Video on 5-s delay • Instruct student to sit at table. • • • • Show student iPhone and tell student to “watch the movie” Show movie clip that corresponds with target response • Comply with student mand Student emits target mand • Student does not emit mand Student emits target mand • Student does not emit mand or following access to reinforcer After clip is completed, show student highly preferred item that they cannot access (function-based) or low preference item and instruct student to interact with item (nonfunction-based) then step back and wait 20 s • Student sits at table • Comply with mand • • Show student highly preferred item that they cannot access (function-based) or low preference item and instruct student to interact with item (nonfunction-based) then step back and wait 5 s • Wait 10 s and start another trial but back up to show movie prior to trial (i.e., no 5 s delay • Increase delay by 2 s every time student emits correct response w/o video. Maintain delay if student emits correct response after video Complete 5 trials • • 220 APPENDIX K PROCEDURAL INTEGRITY CHECKLIST FOR VIDEO MODELING 221 Observer Participant Experimenter Date Session Condition ! Place a checkmark in the corresponding box (yes or no) for each component during each trial Trials Experimental Component Experimenter shows video to participant or provides delay if indicated Experimenter presents a stimulus that replicates the events in the video Experimenter provides consequence indicated by participant if a target response is emitted within 20 s of stimulus presentation If target response is not emitted within 20 s of stimulus presentation, experimenter does nothing Experimenter includes 10 s ITI following completion of programmed consequence 1 Y 2 N ! 222 Y 3 N Y 4 N Y 5 N Y N APPENDIX L SOCIAL VALIDITY QUESTIONNAIRE 223 Please fill out the questionnaire both honestly and to the best of your ability. Thank you for your cooperation! Directions: Video Clip #____ Please view the video clip and then rate the extent to which you agree with the statements below on a scale of 1 to 10 with 1 being “completely disagree” and 10 being “completely agree.” A. The student is able to communicate basic needs and wants in a way that people who do not know him/her are likely to understand. &! '! (! )! %! *! +! ,! -! &#! ! B. The student appears to be having a good time (e.g., smiling, laughing). &! '! (! )! %! *! +! ,! -! &#! ! C. The student is learning skills in addition to communication. &! '! (! )! %! *! +! ,! -! &#! ! +! ,! -! &#! ! D. This is a good use of the student’s time &! '! (! )! %! *! Comments about any undesirable behaviors observed in the video clip: ______________________________________________________________________________ ______________________________________________________________________________ ______________________________________________________________________________ ______________________________________________________ Comments about the student’s communication in the video clip: ______________________________________________________________________________ ______________________________________________________________________________ ______________________________________________________________________________ ______________________________________________________ Directions: 224 After viewing all video clips, please rate the extent to which you agree with the statements below on a scale of 1 to 10 with 1 being “completely disagree” and 10 being “completely agree.” A. The goal of this project, teaching the child to communicate with others, targeted an important skill for the child to learn &! '! (! )! %! *! +! ,! -! &#! ! B. I would like to use the intervention provided by the experimenter with the child in the video &! '! (! )! %! *! +! ,! -! &#! ! C. I would like to use the intervention provided by the experimenter with other children &! '! (! )! %! *! +! ,! -! &#! ! &#! ! D. I plan to recommend this intervention to other educators/parents &! '! (! )! %! *! +! ,! -! Comments: ________________________________________________________________________ ________________________________________________________________________ ________________________________________________________________________ ________________________________________________________________________ Name (optional): ____________________________________ Relation to the student in the video (optional): _____________________________ 225 APPENDIX M PARENT PERMISSION FOR CHILD TO PARTICIPATE 226 November 1, 2009 Dear Parent or Guardian, We are conducting a research study to identify and expand the communicative behaviors young children with autism use to interact with others. We hope this will be useful to parents, teachers, and others by providing information about how and why a child interacts with others and how they can learn to interact with others more effectively. At this time, we are looking for children to participate in order to help us further develop this procedure. Your child’s teacher and principal have agreed to allow students in your child’s school to participate and your child has been referred for participation by his/her classroom teacher. This letter is to request your permission for your child to be included in the study and for the results to be used as data in our study on the effectiveness of this procedure. By allowing your child to participate in this study, you are giving permission for your child’s teacher to provide information to the researchers regarding the ways your child interacts with others. You are also giving permission for the researchers to conduct two observations of your child in his/her school. Each observation will last 20 minutes and is designed to identify behaviors your child uses to communicate with others and events in the environment that potentially cause those behaviors. Following observations, your child will participate in a one-on-one assessment procedure for at most, 30 minutes each day over a period of 2 to 3 weeks. The purpose of the assessment is to determine the best way to set up a learning environment so that your child can learn to communicate basic needs and wants to others. A teaching procedure will then be devised and implemented as part of the daily classroom schedule to increase the interactions between your child and other people. This procedure is for research purposes and is free of charge and the results will be provided to you upon request. ! The potential benefits to your child for taking part in this study are the potential identification of how your child interacts with others and the situations in which that is likely to occur. Additionally, your child may benefit from receiving individualized communication instruction. This may be helpful in guiding future interventions used by caregivers that interact with your child. During the assessment and intervention periods, your child will have opportunities to communicate with researchers in a fun and safe manner. Additionally, attempts to increase communicative behaviors will be made. Allowing your child to participate will help us to improve this procedure to make it more helpful in the future. The potential risks for your child participating in this study include the possibility that engaging in a new activity within a new environment may be distressing for them. Research activities will be modified immediately if any unexpected concerning behaviors (crying, aggression, property destruction, etc) occur. However, challenging behaviors are not anticipated as the procedure is designed to be enjoyable for your child. There are no other known risks for your child participating in this study. The data for this project will be kept confidential unless there is a danger to yourself, your child, or others. Data will be stored in a locked file cabinet in a locked office or on password protected 227 computers inside protected files. Members of the research team and the MSU Institutional Review Board will be the only people to have access to data. The results of this study may be published or presented at professional meetings but the identities of all research participants will remain anonymous. You can indicate your permission for your child to participate by signing the letter and returning it to your child’s teacher. If, after you sign and return the letter, you change your mind, simply let your child’s teacher know and your child will not be asked to participate. There is no penalty for refusing to participate. With your consent, we also plan to video record the assessment and intervention procedures. All video recordings will be collected and stored in a locking file cabinet in a locked office in the College of Education at MSU or on laptop computers with password access for members of the research team only. If you consent, all assessments and interviews will be recorded and reviewed at a later date by members of the research team only. You can indicate your permission for video recording the assessment and intervention procedures by checking the appropriate boxes following the signature line. There is no penalty for not providing consent for recording procedures and your child can still participate in the study if you do not agree to recording. If you have any questions at any time please feel free to contact me by phone at (734) 395-6285 or by email at plavnick@msu.edu. If you have concerns or questions about this study, such as scientific issues, how to do any part of it, or to report an injury, please contact the supervising faculty, Dr. Summer Ferreri: (517) 432-2013, email sferreri@msu.edu, or regular mail at Counseling, Educational Psychology, and Special Education 340 Erickson Hall Michigan State University East Lansing, MI 48824. If you have any questions or concerns about your role and rights as a research participant, would like to obtain information or offer input, or would like to register a complaint about this study, you may contact, anonymously if you wish, Michigan State University’s Human Research Protection Program at 517-355-2180, Fax 517-432-4503, or e-mail irb@msu.edu or regular mail at 207 Olds Hall, MSU, East Lansing, MI 48824. Sincerely, Joshua B. Plavnick Doctoral Student Counseling, Educational Psychology, and Special Education Michigan State University East Lansing, MI 48824 (734) 395-6285 plavnick@msu.edu 228 PERMISSION FOR CHILD’S PARTICIPATION I consent to the participation of my child in the research project entitled “Functional assessment and function-based communication training for young children with autism.” I have read the attached letter and the project has been thoroughly explained to me by Joshua Plavnick. I acknowledge that I have had the opportunity to obtain additional information regarding the project and that any questions I have raised have been answered to my full satisfaction. Furthermore, I understand that I am free to withdraw my consent at any time and to discontinue participation in the project without prejudice. Finally, I acknowledge that I have read the consent form. I sign it freely and voluntarily. A copy has been given to me. Child’s Name: ____________________________ Age: _________ Relationship to child: _______________________________ Signed: Date: (Parent or guardian) I agree to allow audio recording of the interview procedure: Yes No Initials____________ I agree to allow video recording of the assessment procedure: Yes No Initials____________ 229 APPENDIX N TEACHER CONSENT FOR PARTICIPATION 230 November 1, 2009 Dear Teacher, We are conducting a research study to identify and expand the communicative behaviors young children with autism use to interact with others. We hope this will be useful to parents, teachers, and others by providing information about how and why a child interacts with others and how they can learn to interact with others more effectively. At this time, we are looking for children to participate in order to help us learn about whether and how the procedures can identify behaviors that each child uses for communicative purposes and to identify effective methods for teaching new communicative responses. Your classroom has been referred as a potential location of service for children with autism and severe communication impairments. Therefore, this letter is to request your consent to participate in this research study. ! By agreeing to participate, you agree to refer students to the researchers for participation in this research study. Students that do not currently use a clear system to communicate basic needs and wants (i.e. vocal verbal speech, sign language, or picture exchange) are eligible for referral. Additionally, we will ask you to complete a communication inventory, which will take approximately 20 minutes. Following completion of the inventory, researchers will conduct two, 20-minute observations of the referred student in his/her school environment. The purpose of the observations is to identify behaviors the student uses to communicate with others and events in the environment that potentially cause those behaviors. The observations will take place in the environment in which the student is identified to be the most interactive. Following observations, the student will participate in a one-on-one assessment procedure administered by the researchers. The procedure will take at most, 30 minutes each day over a period of 2 to 3 weeks. The purpose of the assessment is to determine the best way to set up a learning environment so that the student can learn to communicate basic needs and wants to others. Following the assessment, each participant will receive 30-minutes of direct communication instruction in the classroom and delivered by the researchers for three days each week for a minimum of four weeks. You may refer up to two students from your classroom for participation. This procedure is being conducted for research purposes and is free of charge and the results of the assessment will be provided to you upon request. The potential benefits for participating in this study include the researchers identifying conditions that increase the student’s nonverbal communicative behavior and which may be helpful in guiding future interventions used by caregivers that interact with the student. During the assessment and intervention procedures, the student will be encouraged to engage in communicative behaviors that are safe and most easily understood by others. Your participation will help us to improve these procedures to make them more helpful in the future. The potential risks for participating in this study include the possibility that engaging in a new activity within a new environment may be distressing for the student. The research procedures will be immediately modified upon occurrence of any unexpected concerning behaviors (crying, aggression, property destruction, etc). However, as the procedure is designed to be enjoyable for the 231 student, challenging behaviors are not anticipated. There are no other foreseeable risks with this assessment procedure. The data for this project will be kept confidential unless there is a danger to anyone involved. All data will be collected with paper and pencil or laptop computers. Data will be stored in a locked file cabinet in a locked office or on password protected computers inside protected files. Members of the research team or the MSU Institutional Review Board will be the only people to have access to data with identifying information. The results of this study may be published or presented at professional meetings but the identities of all research participants will remain anonymous. You can indicate your consent for participation by signing the letter and returning it to a member of the research team. If, after you sign and return the letter, you change your mind, simply let a member of the research team know and you will not be asked to participate. There is no penalty for refusing to participate. With your consent, we also plan to video record the assessment and intervention procedures. All recordings will be collected and stored in a locking file cabinet in a locked office in the College of Education at MSU or on the researchers laptop computers with password access for members of the research team only. If you consent, all assessments and interventions will be recorded and reviewed at a later date for data analysis. You can indicate your permission for video recording by checking the appropriate boxes following the signature line. There is no penalty for not providing consent for recording procedures and you can still participate in the study if you do not agree to recording. If you have any questions at any time please feel free to contact me by phone at (734) 395-6285 or by email at plavnick@msu.edu. If you have concerns or questions about this study, such as scientific issues, how to do any part of it, or to report an injury, please contact the supervising faculty, Dr. Summer Ferreri: (517) 432-2013, email sferreri@msu.edu, or regular mail at Counseling, Educational Psychology, and Special Education 340 Erickson Hall Michigan State University East Lansing, MI 48824. If you have any questions or concerns about your role and rights as a research participant, would like to obtain information or offer input, or would like to register a complaint about this study, you may contact, anonymously if you wish, Michigan State University’s Human Research Protection Program at 517-355-2180, Fax 517-432-4503, or e-mail irb@msu.edu or regular mail at 207 Olds Hall, MSU, East Lansing, MI 48824. Sincerely, Joshua B. Plavnick Doctoral Student Counseling, Educational Psychology, and Special Education Michigan State University East Lansing, MI 48824 (734) 395-6285 plavnick@msu.edu 232 CONSENT FOR PARTICIPATION I consent to participation in the research project entitled “Functional assessment and function-based communication training for young children with autism.” I have read the attached letter and the project was thoroughly explained to me by Joshua Plavnick. I acknowledge that I have had the opportunity to obtain additional information regarding the project and that any questions I have raised have been answered to my full satisfaction. Furthermore, I understand that I am free to withdraw my consent at any time and to discontinue participation in the project without prejudice. Finally, I acknowledge that I have read the consent form. I sign it freely and voluntarily. A copy has been given to me. Printed Name: ___________________________________ Signed: Date: (Teacher) I agree to allow audio recording of the interview procedure: Yes No Initials____________ I agree to allow video recording of the assessment procedure: Yes No Initials____________ 233 ! REFERENCES 234 REFERENCES Ahearn, W. H., Clark, K. M., MacDonald, R. P., & Chung, B. I. (2007). Assessing and treating vocal stereotypy in children with autism. Journal of Applied Behavior Analysis, 40, 263275. doi:10.1901/jaba.2007.30-06 Alberto, P. A., & Troutman, A. C. (2008). Applied Behavior Analysis for Teachers (8th ed.). Upper Saddle River, NJ: Prentice Hall. American Psychiatric Association. (2000). Diagnostic and statistical manual of mental disorders (4th ed., text revision). Washington, DC: Author. Ayllon, T., & Michael, J. (1959). The psychiatric nurse as a behavioral engineer. Journal of the Experimental Analysis of Behavior, 2, 323-334. doi:10.1901/jeab.1959.2-32 Baer, D. M. (1977). “Perhaps it would be better not to know everything.” Journal of Applied Behavior Analysis, 10, 167-172. doi:10.1901/jaba.1977.10-167 Baer, D. M., Wolf, M. M., & Risley, T. R. (1968). Some current dimensions of applied behavior analysis. Journal of Applied Behavior Analysis, 1, 91-97. doi:10.1901/jaba.1968.1-91 Baer, D. M., & Wolf, M. M. (1987). Some still-current dimensions of applied behavior analysis. Journal of Applied Behavior Analysis, 20, 313-327. doi:10.1901/jaba.1987.20-313 Bandura, A. (1977). Social learning theory. New York, NY: General Learning Press. Bijou, S. W., & Baer, D. M. (1965). Child development: Universal stage of infancy. Englewood Cliffs, NJ: Prentice Hall. Bodfish, J. W. (2004). Treating the core features of autism: Are we there yet? Mental Retardation and Developmental Disabilities Research Reviews, 10, 318-326. doi:10.1002/mrdd.20045 Bondy, A. S., & Frost, L. A. (1994). The picture exchange communication system. Focus on Autism and Other Developmental Disabilities, 9, 1-19. doi:10.1177/108835769400900301 Borrero, C. S. W., & Borrero, J. C. (2008). Descriptive and experimental analyses of potential precursers to problem behavior. Journal of Applied Behavior Analysis, 41, 83-96. doi:10.1901/jaba.2008.41-83 Bourret, J., Vollmer, T. R., & Rapp, J. T. (2004). Evaluation of a vocal mand assessment and vocal mand training procedures. Journal of Applied Behavior Analysis, 37, 129-43. doi:10.1901/jaba.2004.37-129 235 Brady, N. C., & Halle, J. W. (1997). Functional analysis of communicative behaviors. Focus on Autism and Other Developmental Disabilities, 12, 95-104. doi:10.1177/108835769701200205 Bryson, S. E., Koegel, L. K., Koegel, R. L., Openden, D., Smith, I. M., & Nefdt, N. (2007). Large scale dissemination and community implementation of pivotal response treatment: Program description and preliminary data. Research and Practice for Persons with Severe Disabilities, 32, 142-153. Retrieved from http://www.atyponlink.com/TASH/doi/abs/10.5555/rpsd.32.2.142 Buggey, T. (2005). Video self-modeling applications with students with autism spectrum disorder in a small private school setting. Focus on Autism and Other Developmental Disabilities, 20, 52–63. doi:10.1177/10883576050200010501 Busk, P. L., & Marascuilo, L. A. (1992). Statistical analyses in single-case research: Issues, procedures, and recommendations, with applications to multiple behaviors. In T. Kratochwill & J. Levin (Eds.), Single-case research design and analysis (pp. 159-185). Hillsdale, NJ: Lawrence Erlbaum Associates. Caronna, E. B., Milunsky, J. M., & Tager-Flusberg, H. (2008). Autism spectrum disorders: Clinical and research frontiers. Archives of Disease in Childhood, 93, 518-523. doi:10.1136/adc.2006.115337 Carr, E. G., & Durand, V. M. (1985). Reducing behavior problems through functional communication training. Journal of Applied Behavior Analysis, 18, 111-126. doi:10.1901/jaba.1985.18-111! ! Carr, E. G., Newsom, C. D., & Binkoff, J. A. (1976). Stimulus control of self-destructive behavior in a psychotic child. Journal of Abnormal Child Psychology, 4, 139-152. doi:10.1007/BF00916518 Casey, S. D., & Merical, C. L. (2006). The use of functional communication training without additional treatment procedures in an inclusive school setting. Behavioral Disorders, 32, 46-54. Retrieved from http://www.ccbd.net/behavioraldisorders/Journal/Journal.cfm?BDID=14BAEF50-30482906-EAF91D47AFADDA51 Catania, A. C. (1998). Learning (4th ed.). Upper Saddle River, NJ: Prentice Hall. Cederlund, M., Hagber, B., Billstedt, E., Gillberg, C., & Gillberg, C. (2008). Asperger syndrome and autism: a comparative longitudinal follow-up study more than 5 years after original diagnosis. Journal of Autism and Developmental Disorders, 38, 72-85. doi:10.1007/s10803-007-0364-6 236 Chambless, D. L., Sanderson, W. C., Shoham, V., Johnson, S. B., Pope, K. S., Crits-Cristoph, P., . . . McCurry, S. (1996). An update on empirically-validated therapies. The Clinical Psychologist, 49, 5-18. Charlop, M. H., & Milstein, J. P. (1989). Teaching autistic children conversational speech using video modeling. Journal of Applied Behavior Analysis, 22, 275-285. doi:10.1901/jaba.1989.22-275 Charlop, M. H., Schreibman, L., & Thibodeau, M. G. (1985). Increasing spontaneous verbal responding in autistic children using a time delay procedure. Journal of Applied Behavior Analysis, 18, 155-166. doi:10.1901/jaba1985.18-155 Charlop-Christy, M. H., Carpenter, M., Le, L., LeBlanc, L. A., & Kellet, K. (2002). Using the picture exchange communication system (PECS) with children with autism: Assessment of PECS acquisition, speech, social-communicative behavior and problem behavior. Journal of Applied Behavior Analysis, 35, 213-231. doi:10.1901/jaba.2002.35-213 Charlop-Christy, M. H., Le, L., & Freeman, K. A. (2000). A comparison of video modeling with in vivo modeling for teaching children with autism. Journal of Autism and Developmental Disorders, 30, 537-552. doi:10.1023/A:1005635326276 Cooper, J. O., Herron, T. E., & Heward, W. L. (2007). Applied Behavior Analysis (2nd ed.). Upper Saddle River, NJ: Prentice Hall. Cote, C. A., Thompson, R. H., Hanley, G. P., & McKerchar, P. M. (2007). Teacher report and direct assessment of preferences for identifying reinforcers for young children. Journal of Applied Behavior Analysis, 40, 157-66. doi: 10.1901/jaba.2007.177-05. Darden-Brunson, F., Green, A., & Goldstein, H. (2008). Video based instruction for children with autism. In J. Luiselli, D. Russo, W. Christian, & S. Wilczynski (Eds.), Effective practices for children with autism: Educational and behavioral support interventions that work (pp. 241-268). New York, NY: Oxford University Press. Dawson, G. (2008). Early behavioral intervention, brain plasticity, and the prevention of autism spectrum disorder. Developmental Psychopathology, 20, 775-803. doi:10.1017/S0954579408000370 Deguchi, H. (1984). Observational learning from a radical-behavioristic viewpoint. The Behavior Analyst, 7, 83-95. Retrieved from http://www.ncbi.nlm.nih.gov.proxy2.cl.msu.edu/pmc/articles/PMC2741741/ Detrich, R. (2008). Evidence-based, empirically supported, or best practice? A guide for the scientist-practitioner. In J. Luiselli, D. Russo, W. Christian, & S. Wilczynski (Eds.), Effective practices for children with autism: Educational and behavioral support interventions that work (pp. 3-26). New York, NY: Oxford University Press. 237 Drash, P. W., High, R. L., & Tudor, R. M. (1999). Using mand training to establish an echoic repertoire in young children with autism. The Analysis of Verbal Behavior, 16, 29-44. Retrieved from http://www.ncbi.nlm.nih.gov.proxy2.cl.msu.edu/pmc/articles/PMC2748578/ Dube, W. V., MacDonald, R. P. F., Mansfield, R. C., Holcomb, W. L., & Ahearn, W. H. (2004). Toward a behavioral analysis of joint attention. Behavior Analyst, 27, 11. Retrieved from http://www.ncbi.nlm.nih.gov.proxy2.cl.msu.edu/pmc/articles/PMC2755395/ Dufrene, B. A., Steuart Watson, T., & Kazmerski, J. S. (2008). Functional analysis and treatment of nail biting. Behavior Modification, 32, 913-27. doi:10.1177/0145445508319484 Dunlap, G., Ester, T., Langhans, S., & Fox, L. (2006). Functional communication training with toddlers in home environments. Journal of Early Intervention, 28, 81-96. doi:10.1177/105381510602800201 Durand, V. M., & Carr, E. G. (1991). Functional communication training to reduce challenging behavior: maintenance and application in new settings. Journal of Applied Behavior Analysis, 24, 251-264. doi:10.1901/jaba/1991/24-151 Durand, V. M., & Crimmins, D. B. (1987). Assessment and treatment of psychotic speech in an autistic child. Journal of Autism and Developmental Disorders, 17, 17-28. doi:10.1007/BF01487257 Durand, V. M., & Kishi, G. (1987). Reducing severe behavior problems among persons with dual sensory impairments: An evaluation of a technical assistance model. Journal of the Association for Persons with Severe Handicaps, 12, 2-10. Ferreri, S. J., & Plavnick, J. B. (in press). A functional analysis of prelinguistic behaviors emitted by young children with severe developmental disabilities. The Analysis of Verbal Behavior. Ferster, C. B., & Skinner, B. F. (1957). Schedules of reinforcement. New York, NY: AppletonCentury-Crofts. Fisher, W., Piazza, C. C., Bowman, L. G., Hagopian, L. P., Owens, J. C., & Slevin, I. (1992). A comparison of tow approaches for identifying reinforcers for persons with severe and profound disabilities. Journal of Applied Behavior Analysis, 25, 491-498. doi:10.1901/jaba.1992.25-491 Fixsen, D. L., Naoom, S. F., Blase, K. A., Friedman, R. M., & Wallace, F. (2005). Implementation research: a synthesis of the literature. Tampa, FL: National Implementation Research Network. 238 Fombonne, E. (2007). Epidemiological surveys of pervasive developmental disorders. In F. Volkmar (Ed.), Autism and pervasive developmental disorders (2nd ed., pp. 33-68). New York, NY: Cambridge University Press. Franklin, R. D., Allison, D. B., & Gorman, B. S. (1996). Introduction. In R. Franklin, D. Allison, & B. Gorman, (Eds.), Design and analysis of single-case research (pp. 1-12). Erlbaum Hillsdale, NJ. Frea, W. D., & Hughes, C. (1997). Functional analysis and treatment of social-communicative behavior of adolescents with developmental disabilities. Journal of Applied Behavior Analysis, 30, 701-704. doi:10.1901/jaba.1997.30-701 Frost, L. A. & Bondy, A. S. (2002). The Picture Exchange Communication System Training Manual (2nd ed.). Newark, DE: Pyramid Educational Consultants. Fyffe, C. E., Kahng, S. W., Fittro, E., & Russell, D. (2004). Functional analysis and treatment of inappropriate sexual behavior. Journal of Applied Behavior Analysis, 37, 401-404. doi:101901/jaba.2004.37-401 Gena, A., Couloura, S., & Kymissis, E. (2005). Modifying the affective behavior of preschoolers with autism using in-vivo or video modeling and reinforcement contingencies. Journal of Autism and Developmental Disorders, 35, 545-556. doi:10.1007/s10803-005-0014-9 Gilliam, J. E. (2006). The Gilliam autism rating scale (2nd ed.). Austin, TX: Pro-Ed. Goldstein, S., Naglieri, J. A., & Ozonoff, S. (2008). Assessment of Autism Spectrum Disorders. New York, NY: Guilford Press. Gorman, B. S., & Allison, D. B. (1996). Statistical alternatives for single-case designs. In R. Franklin, D. Allison, & B. Gorman, (Eds.), Design and analysis of single-case research (pp. 159-214). Hillsdale, NJ: Erlbaum Greer, R. D., Dudek-Singer, J., & Gautreaux, G. (2006). Observational learning. International Journal of Psychology, 41, 486-499. doi:10.1080/00207590500492435 Greer, R. D., & Ross, D. E. (2008). Verbal behavior analysis: Inducing and expanding complex communication in children with language delays. Boston, MA: Allyn & Bacon. Hagopian, L. P., Bruzek, J. L., Bowman, L. G., & Jennett, H. K. (2007). Assessment and treatment of problem behavior occasioned by interruption of free-operant behavior. Journal of Applied Behavior Analysis, 40, 89-103. doi:10.1901/jaba.2007.63-05 Hagopian, L. P., Fisher, W. W., Sullivan, M. T., Acquisto, J., & LeBlanc, L. A. (1998). Effectiveness of functional communication training with and without extinction and punishment: a summary of 21 inpatient cases. Journal of Applied Behavior Analysis, 31, 211-235. doi:10.1901/jaba.1998.31-211 239 Hall, S. S. (2005). Comparing descriptive, experimental and informant-based assessments of problem behaviors. Research in Developmental Disabilities, 26, 514-526. doi:10.1016/j.ridd.2004.11.004 Hall, G. A., & Sundberg, M. L. (1987). Teaching mands by manipulating conditioned establishing operations. Analysis of Verbal Behavior, 5, 41-53. Retrieved from http://www.ncbi.nlm.nih.gov.proxy2.cl.msu.edu/pmc/articles/PMC2748450/ Halle, J., Brady, N. C., & Drasgow, E. (2004). Enhancing socially adaptive communicative repairs of beginning communicators with disabilities. American Journal of SpeechLanguage Pathology, 13, 43-54. doi:10.1044/1058-0360(2004/006) Halle, J. W., Marshall, A. M., & Spradlin, J. E. (1979). Time delay: A technique to increase language use and facilitate generalization in retarded children. Journal of Applied Behavior Analysis, 12, 431-439. doi:10.1901/jaba.1979.12-431 Halle, J., & Meadan, H. (2007). A protocol for assessing early communication of young children with autism and other developmental disabilities. Topics in Early Childhood Special Education, 27, 49-61. doi:10.1177/02711214070270010401 Hancock, T., & Kaiser, A. (2002). The effects of trainer-implemented enhanced milieu teaching on the social communication of children who have autism. Topics in Early Childhood Education, 22, 39-54. doi:10.1177/027112140202200104 Hanley, G. P., Iwata, B. A., & McCord, B. E. (2003). Functional analysis of problem behavior: A review. Journal of Applied Behavior Analysis, 36, 147-185. doi:10.1901/jaba.2003.36147 Haring, T. G., Kennedy, C. H., Adams, M. J., & Pitts-Conway, V. (1987). Teaching generalization of purchasing skills across community settings to autistic youth using videotape modeling. Journal of Applied Behavior Analysis, 20, 89-96. doi:10.1901/jaba.1987.20-89 Hart, B., & Risley, T. R. (1975). Incidental teaching of language in the preschool. Journal of Applied Behavior Analysis, 8, 411-420. doi:10.1901/jaba.1975.8-411 Hartley, S. L., Sikora, D. M., & McCoy, R. (2008). Prevalence and risk factors of maladaptive behavior in young children with Autistic Disorder. Journal of Intellectual Disability Research, 52, 819-829. doi:10.1111/j.1365-2788.2008.01065.x Herzinger, C. V., & Campbell, J. M. (2007). Comparing functional assessment methodologies: A quantitative synthesis. Journal of Autism and Developmental Disorders, 37, 1430-45. doi:10.1007/s10803-006-0219-6 240 Hewett, F. M. (1965). Teaching speech to an autistic child through operant conditioning. American Journal of Orthopsychiatry, 35, 927-936. doi:10.1111/j.19390025.1965.tb00472.x Hine, J. F., & Wolery, M. (2006). Using point-of-view video modeling to teach play to preschoolers with autism. Topics in Early Childhood Special Education, 26, 83-93. doi:10.1177/02711214060260020301 Horner, R. D., & Baer, D. M. (1978). Multiple-probe technique: a variation on the multiple baseline. Journal of Applied Behavior Analysis, 11, 189-196. doi:10.1901/jaba.1978.11189 Horner, R. H., & Carr, E. G. (1997). Behavioral support for students with severe disabilities: Functional assessment and comprehensive intervention. The Journal of Special Education, 31, 84-104. doi:10.1177/002246699703100108 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, 165-179. Retrieved from http://find.galegroup.com.proxy1.cl.msu.edu.proxy2.cl.msu.edu/gtx/infomark.do?&conte ntSet=IACDocuments&type=retrieve&tabID=T002&prodId=AONE&docId=A126850132&source= gale&srcprod=AONE&userGroupName=msu_main&version=1.0 Howlin, P., Goode, S., Hutton, J., & Rutter, M. (2004). Adult outcome for children with autism. Journal of Child Psychology and Psychiatry, 45, 212-229. doi:10.1111/j.14697610.2004.00215.x Individuals with Disabilities Education Improvement Act, 20 U.S.C. § 1400 et seq. (2004). Ittenbach, R. F. & Lawhead, W. F. (1996). Historical and philosophical foundations of singlecase research. In R. Franklin, D. Allison, & B. Gorman, (Eds.), Design and analysis of single-case research (pp. 13-40). Hillsdale, NJ: Erlbaum Iwata, B. A., Dorsey, M. F., Slifer, K. J., Bauman, K. E., & Richman, G. S. (1994). Toward a functional analysis of self-injury. Journal of Applied Behavior Analysis, 27, 197-209. (Reprinted from Analysis and Intervention in Developmental Disabilities, 2, 3-20, 1982). doi:10.1901/jaba.1994.27-197! ! Iwata, B. A., Pace, G. M., Dorsey, M. F., Zarcone, J. R., Vollmer, T. R., Smith, R. G., . . . Willis, K. D. (1994). The functions of self-injurious behavior: An experimental-epidemiological analysis. Journal of Applied Behavior Analysis, 27, 215-240. doi:10.1901/jaba.1994.27215! ! Iwata, B. A., Smith, R. G., & Michael, J. (2000). Current research on the influence of establishing operations on behavior in applied settings. Journal of Applied Behavior 241 Analysis, 33, 401-410. doi:10.1901/jaba.2000.33-411 Iwata, B. A., Wallace, M. D., Kahng, S. W., Lindberg, J. S., Roscoe, E. M., Conners, J. H., . . . Worsdell, A. S. (2000). Skill acquisition in the implementation of functional analysis methodology. Journal of Applied Behavior Analysis, 33, 181-194. doi:10.1901/jaba.2000.33-181 Jennett, H. K., Harris, S. L., & Delmolino, L. (2008). Discrete trial instruction vs. Mand training for teaching children with autism to make requests. The Analysis of Verbal Behavior, 24, 69-85. Johnston, J. M., & Pennypacker, H. S. (2008). Strategies and tactics for behavioral research (3rd ed.). Hillsdale, NJ: Erlbaum.! ! Kanner, L. (1943). Autistic disturbances of affective contact. Nervous Child, 2, 217–50. Retrieved from http://scholar.google.com/scholar?hl=en&q=Nervous+child&btnG=Search&as_sdt=8000 0000&as_ylo=&as_vis=0 ! Kasari, C., Paparella, T., Freeman, S., & Jahromi, L. B. (2008). Language outcome in autism: Randomized comparison of joint attention and play interventions. Journal of Consulting and Clinical Psychology, 76, 125-37. doi:10.1037/0022-006X.76.1.125 Kazdin, A. E. (1982). Single-case research designs. New York, NY: Oxford. Keen, D. (2005). The use of non-verbal repair strategies by children with autism. Research in Developmental Disabilities, 26, 243-254. doi:10.1007/s10882-005-2198-5 Keen, D., Sigafoos, J., & Woodyatt, G. (2001). Replacing prelinguistic behaviors with functional communication. Journal of Autism and Developmental Disorders, 31, 385-398. doi:10.1023/A:1010612618969 Keen, D., Woodyatt, G., & Sigafoos, J. (2002). Verifying teacher perceptions of the potential communicative acts of children with autism. Communication Disorders Quarterly, 23, 133-142. doi:10.1177/15257401020230030201 Keen, D., Sigafoos, J., & Woodyatt, G. (2005). Teacher responses to the communicative attempts of children with autism. Journal of Autism and Developmental Disorders, 17, 19-33. doi:10.1007/s10882-005-2198-5 Kelley, M. E., Lerman, D. C., & Van Camp, C. M. (2002). The effects of competing reinforcement schedules on the acquisition of functional communication. Journal of Applied Behavior Analysis, 35, 59-63. doi:10.1901/jaba.2002.35-59 242 Kemp, D. C., & Carr, E. G. (1995). Reducation of severe problem behavior in community employment using an hypothesis-driven multicomponent approach. Journal of the Association for Persons with Severe Handicaps, 20, 229-247. Koegel, R. K., O’Dell, M. C., & Koegel, L. K. (1987). A natural language teaching paradigm for nonverbal autistic children. Journal of Autism and Developmental Disorders, 17, 187200. doi:10.1007/BF01495055 Koegell, R. K., O’Dell, M. C., & Dunlap, G. (1988). Producing speech use in nonverbal autistic children by reinforcing attempts. Journal of Autism and Developmental Disorders, 18, 525-538. doi:10.1007/BF02211871 Koegel, R. L., & Rincover, A. (1977). Research on the difference between generalization and maintenance in extra-therapy responding. Journal of Applied Behavior Analysis, 10, 1-12. doi: 10.1901/jaba.1977.10-1 Koegel, R. L., Vernon, T. W., & Koegel, L. K. (2009). Improving social initiations in young children with autism using reinforcers with embedded social interactions. Journal of Autism and Developmental Disorders, 39, 1240-1251. doi:10.1007/s10803-009-0732-5 Kogan, M. D., Blumberg, S. J., Schieve, L. A., Boyle, C. A., Perrin, J. M., Ghandour, R. M., . . . van Dyck, P. C. (2009). Prevalence of parent-reported diagnosis of autism spectrum disorder among children in the US, 2007. Pediatrics. Advance online publication. doi: 10.1542/peds.2009-1522 Kratochwill, T. R., Hitchcock, J., Horner, R. H., Levin, J. R., Odom, S. L., Rindskopf, D. M., & Shadish, W. R. (2010). Single-case designs technical documentation. Retrieved from What Works Clearinghouse website: http://ies.ed.gov/ncee/wwc/pdf/wwc_scd.pdf. Lalli, J. S., Casey, S., & Kates, K. (1995). Reducing escape behavior and increasing task completion with functional communication training, extinction, and response chaining. Journal of Applied Behavior Analysis, 28, 261-268. doi:10.1901/jaba.1995.28-261 Lamarre, J. & Holland, J. G. (1985). The functional independence of mands and tacts. Journal of the experimental analysis of behavior, 43, 5-19. doi:10.1901/jeab.1985.43-5 Langdon, N. A., Carr, E. G., & Owen-Deschryver, J. S. (2008). Functional analysis of precursors for serious problem behavior and related intervention. Behavior Modification, 32, 804827. doi:10.1177/0145445508317943 Laraway, S., Snycerski, S., Michael, J., & Poling, A. (2003). Motivating operations and terms to describe them: Some further refinements. Journal of Applied Behavior Analysis, 36, 407414. doi:10.1901/jaba.2003.36-407 LeBlanc, L. A., Esch, J., Sidener, T. M., & Firth, A. M. (2006). Behavioral language interventions for children with autism: comparing applied verbal behavior and naturalistic 243 teaching approaches. The Analysis of Verbal Behaivor, 22, 49-60. Lerman, D. C., Parten, M., Addison, R. L., Vorndran, M. C., Volkert, M. V., & Kodak, T. (2005). A methodology for assessing the functions of emerging speech in children with developmental disabilities. Journal of Applied Behavior Analysis, 38, 303-316. doi:10.1901/jaba.2005.106-04 Lovaas, O. I. (1977). The autistic child: Language development through behavior modification. New York: Irvington. Lovaas, O. I., Berberich, J. P., Perloff, B. F., & Schaeffer, B. (1966). Acquisition of imitative speech by schizophrenic children. Science, 151, 705-707. doi:10.1126/science.151.3711.705 Lovaas, O. I., Koegel, R., Simmons, J. Q., & Stevens Long, J. (1973). Some generalization and follow up measures on autistic children in behavioral therapy. Journal of Applied Behavior Analysis, 6, 131-165. doi:10.1901/jaba.1973.6-131 Lowy Apple, A., Billingsley, F., & Schwartz, I. S. (2005). Effects of video modeling alone and with self-management on compliment-giving behaviors of children with high-functioning ASD. Journal of Positive Behavior Interventions, 7, 33-46. doi:10.1177/10983007050070010401 Luiselli, J. K. (2008). Antecedent (preventive) intervention. In J. Luiselli, D. Russo, W. Christian, & S. Wilczynski (Eds.), Effective practices for children with autism: Educational and behavioral support interventions that work (pp. 393-412). New York, NY: Oxford University Press. MacDonald, R., Sacramone, S., Mansfield, R., Wiltz, K., & Ahearn, W. H. (2009). Using video modeling to teach reciprocal pretend play to children with autism. Journal of Applied Behavior Analysis, 42, 43-55. doi:10.1901/jaba.2009.42-43 Mace, F. C., Browder, D. M., & Lin, Y. (1987). Analysis of demand conditions associated with stereotypy. Journal of Behavior Therapy and Experimental Psychiatry, 18, 25-31. doi:10.1016/0005-7916(87)90068-1 Mace, F. C., & Lalli, J. S. (1991). Linking descriptive and experimental analyses in the treatment of bizarre speech. Journal of Applied Behavior Analysis, 24, 553-562. doi:10.1901/jaba.1991.24-553 Mailick Seltzer, M., Shattuck, P., Abbeduto, L., & Greenberg, J. S. (2005). Trajectory of development in adolescents and adults with autism. Mental Retardation and Developmental Disabilities Research Reviews, 10, 234-247. doi:10.1002/mrdd.20038 244 Marcus, B. A., Vollmer, T. R., Swanson, V., Roane, H. R., & Ringdahl, J. E. (2001). An experimental analysis of aggression. Behavior Modification, 25, 189-213. doi:10.1177/0145445501252002 McArdle, J. (1996). Preface. In R. Franklin, D. Allison, B. & Gorman, (Eds.), Design and analysis of single-case research (pp. vii-xi). Erlbaum Hillsdale, NJ. McCoy, K., & Hermansen, E. (2007). Video modeling for individuals with autism: A review of model types and effects. Education and Treatment of Children, 30, 183-213. doi:10.1353/etc.2007.0029 McPartland, J., & Klin, A. (2006). Asperger’s syndrome. Adolescent Medicine Clinics, 17, 771788. doi:10.1016/j.admecli.2006.06.010 Meadan, H., Halle, J., Ostrosky, M. M., & DeStefano, L. (2008). Communicative behavior in the natural environment: Case studies of two young children with autism and limited expressive language. Focus on Autism and Other Developmental Disabilities, 23, 37-48. doi: 10.1177/1088357607311444 Meadan, H., Halle, J. W., Watkins, R. V., & Chadsey, J. G. (2006). Examining communication repairs of 2 young children with autism spectrum disorder: the influence of the environment. American Journal of Speech and Language Pathology, 15, 57-71. doi:10.1044/1058-0360(2006/007) Michael, J. (1982). Distinguishing between the discriminative and motivational functions of stimuli. Journal of the Experimental Analysis of Behavior, 37, 149-155. doi:10.1901/jeab.1982.37-149 Michael, J. (1988). Establishing operations and the mand. The Analysis of Verbal Behavior, 6, 39. Retrieved from: http://www.ncbi.nlm.nih.gov.proxy2.cl.msu.edu/pmc/articles/PMC2748485/ Michael, J. (2000). Implications and refinements of the establishing operation concept. Journal of Applied Behavior Analysis, 33, 401-410. doi:10.1901/jaba.2000.33-401! Michael, J. (2004). Concepts and principles of behavior analysis. Kalamazoo, MI: Society for the Advancement of Behavior Analysis. Michael, J. (2007). Motivating operations. In J. Cooper, T. Heron, & W. Heward (Eds.), Applied behavior analysis (2nd ed., pp. 374-391). Upper Saddle River, NJ: Prentice Hall. Miltenberger, R. G., Long, E. S., Rapp, J. T., Lumley, V. & Elliott, A. J. (1998). Evaluation of the function of hair pulling: A preliminary investigation. Behavior Therapy, 29, 211-219. doi:10.1016/S0005-7894(98)80003-8 245 National Autism Center. (2009). National Standard’s Report. Randolph, MA: National Autism Center. National Research Council. (2001). Educating children with autism. Committee on Educational Interventions for Children with Autism, Division of Behavioral and Social Sciences and Education. Washington, DC: National Academy Press. Neef, N. A. 2006. Advances in single subject design. In C. Reynolds, & T. Gutkin, (Eds.), The handbook of school psychology (4th ed., pp. 54-64). New York, NY: Wiley. Neef, N. A., Marckel, J., Ferreri, S. J., Bicard, D. F., Endo, S., Aman, M. G., . . . Armstrong, N. (2005). Behavioral assessment of impulsivity: A comparison of children with and without attention deficit hyperactivity disorder. Journal of Applied Behavior Analysis, 38, 23-37. doi:10.1901/jaba.2005.146-02 Neef, N. A., & Peterson, S. M. (2007). Functional analysis. In J. Cooper, T. Heron, & W. Heward (Eds.), Applied behavior analysis (2nd ed., pp. 499-524). Upper Saddle River, NJ: Prentice Hall. New York State Department of Health Early Intervention Program (1999). Clinical practice guideline: The guideline technical report. Autism/pervasive developmental disorders: Assessment and intervention for young children (age 0-3 years). Albany, NY: New York State Department of Health Early Intervention Program. Nikopoulos, C. K., Canavan, C., & Nikopoulou-Smymi, P. (2009). Generalized effects of video modeling on establishing instructional stimulus control in children with autism. Journal of Positive Behavior Interventions, 11, 198-207. doi:10.1177/1098300708325263 Nikopoulos, C. K., & Keenan, M. (2007). Using video modeling to teach complex social sequences to children with autism. Journal of Autism and Developmental Disorders, 37, 678-693. doi:10.1007/s10803-006-0195-x Odom, S. L., Brantlinger, E., Gersten, R., Horner, R. H., Thompson, B., & Harris, K. (2005). Research in special education: Scientific methods and evidence-based practices. Exceptional Children, 71, 137–148. Retrieved from http://find.galegroup.com.proxy1.cl.msu.edu.proxy2.cl.msu.edu/gtx/infomark.do?&conte ntSet=IACDocuments&type=retrieve&tabID=T002&prodId=AONE&docId=A126850130&source= gale&srcprod=AONE&userGroupName=msu_main&version=1.0 Ospina, M. B., Krebs Seida, J., Clark, B., Karkhaneh, M., Hartling, L., Tjosvold, L., . . . Smith, V. (2008). Behavioural and developmental interventions for autism spectrum disorder: a clinical systematic review. PLOS ONE, 3(11), 1-32. doi:10.1371/journal.pone.0003755 Paterson, C. R., & Arco, L. (2007). Using video modeling for generalizing toy play in children with autism. Behavior Modification, 31, 660-681. doi:10.1177/0145445507301651 246 Paul, R. (2008). Interventions to improve communication in autism. Child and Adolescent Psychiatric Clinics of North America, 17, 835-856. doi:10.1016/j.chc.2008.06.011 Peterson, L. Homer, A. L. & Wonderlich, S. A. (1982). The integrity of independent variables in behavior analysis. Journal of Applied Behavior Analysis, 15, 477-492. doi:10.1901/jaba.1982.15-477 Piazza, C. C., Hanley, G. P., Bowman, L. G., Ruyter, J. M., Lindauer, S. E., & Saiontz, D. M. (1997). Functional analysis and treatment of elopement. Journal of Applied Behavior Analysis, 30, 653–672. doi:10.1901/jaba.1997.30-653! ! Piazza, C. C., Hanley, G. P., & Fisher, W. W. (1996). Functional analysis and treatment of cigarette pica. Journal of Applied Behavior Analysis, 29, 437–450. doi:10.1901/jaba.1996.29-437 Rayner, C., Denholm, C., & Sigafoos, J. (2009). Video-Based intervention for individuals with autism: Key questions that remain unanswered. Research in Autism Spectrum Disorders, 3, 291-303. doi:10.1016/j.rasd.2008.09.001 Reagon, K. A., Higbee, T. S., & Endicott, K. (2006). Teaching pretend play skills to a student with autism using video modeling with a sibling as model and play partner. Education and Treatment of Children, 29, 517-528. Retrieved from http://www.educationandtreatmentofchildren.net/contents/index.html Reimers, T. M., Wacker, D. P., Cooper, L. J., Sasso, G. M., Berg, W. K., & Steege, M. W. (1993). Assessing the functional properties of noncompliant behavior in an outpatient setting. Child and Family Behavior Therapy, 15, 1–15. doi:10.1300/J019v15n03_01 Risley, T. R. (2005). Montrose M. Wolf (1935-2004). Journal of Applied Behavior Analysis, 38, 279-287. Risley, T. R., & Hart, B. (1968). Developing correspondence between the non-verbal and verbal behavior of preschool children. Journal of Applied Behavior Analysis, 1, 267-281. doi:10.1901/jaba.1968.1-267 Rogers, S. J., & Vismara, L. A. (2008). Evidence-Based comprehensive treatments for early autism. Journal of Clinical Child and Adolescent Psychology, 37, 8-38. doi:10.1080/15374410701817808 Romanczyk, R. G., & Gillis, J. M. (2008). Practice guidelines for autism education and intervention: Historical perspective and recent developments. In J. Luiselli, D. Russo, W. Christian, & S. Wilczynski (Eds.), Effective practices for children with autism: Educational and behavioral support interventions that work (pp. 27-36). New York, NY: Oxford University Press. 247 Schertz, H. H., & Odom, S. L. (2007). Promoting joint attention in toddlers with autism: A parent-mediated developmental model. Journal of Autism and Developmental Disorders, 37, 1562-1575. doi:10.1007/s10803-006-0290-z Schlosser, R. W., & Wendt, O. (2008). Augmentative and alternative communication intervention for children with autism. In J. Luiselli, D. Russo, W. Christian, & S. Wilczynski (Eds.), Effective practices for children with autism: Educational and behavioral support interventions that work (pp. 325-389). New York, NY: Oxford University Press.! ! Schopler, E., Reichler, R. J., DeVellis, R. F., & Daly, K. (1980). Toward objective classification of childhood autism: Childhood Autism Rating Scale (CARS). Journal of Autism and Developmental Disorders, 10, 91–103. doi:10.1007/BF02408436 Schreibman, L., & Lovaas, O. I. (1974). Rejoinder to Murray’s article. Journal of Humanistic Psychology, 14, 61-62. doi: 10.1177/002216787401400209 Shabani, D. B., Carr, J. E., & Ingeborg Petursdottir, A. (2009). A labratory model for studying response-class hierarchies. Journal of Applied Behavior Analysis, 42, 105-121. doi:10.1901/jaba.2009.42-105 Shavelson, R. J., & Towne, L. (Eds.). (2002). Scientific research in education. Washington, DC: National Academy Press. Sherer, M., Pierce, K. L., Paredes, S., Kisacky, K. L., Ingersoll, B., & Schreibman, L. (2001). Enhancing conversation skills in children with autism via video technology. Which is better, "self" or "other" as a model? Behavior Modification, 25, 140-158. doi:10.1177/0145445501251008 Sidman, M. (1960) Tactics of Scientific Research. New York: Basic Books. Siegel-Causey, E., & Guess, D. (1989). Enhancing nonsymbolic communication interactions among learners with severe disabilities. Baltimore, MD: Brookes. Sigafoos, J., Arthur-Kelly, M., & Butterfield, N. (2006). Enhancing everyday communication for children with disabilities. Baltimore, MD: Brookes. Sigafoos, J., Woodyatt, G., Keen, D., Tait, K., Tucker, M., Roberts-Pennell, D., & Pittendreigh, N. (2000). Identifying potential communicative acts in children with developmental and physical disabilities. Communication Disorders Quarterly, 21, 77-86. doi:10.1177/152574010002100202 Skinner, B. F. (1938). The behavior of organisms: An experimental analysis. New York, NY: Appleton-Century. Skinner, B. F. (1953). Science and Humand Behavior. New York, NY: Macmillan. 248 Skinner, B. F. (1957). Verbal behavior. New York, NY: Appleton-Century-Crofts. Snow, A. V., Lecavalier, L., & Houts, C. (2009). Structure of the Autism Diagnostic InterviewRevised: Diagnostic and phenotypic implications. Journal of Child Psychology and Psychiatry, 50, 734-742. doi:10.1111/j.1469-7610.2008.02018.x Stahmer, A. C. (2007). The basic structure of community early intervention programs for children with autism: Provider descriptions. Journal of Autism and Developmental Disorders, 37, 1344-1354. doi:10.1007/s10803-006-0284-x Stoiber, K. C., & Kratochwill, T. R. (2000). Empirically supported interventions and school psychology: Rationale and methodological issues. Part I. School Psychology Quarterly, 15, 75-105. doi:10.1037/h0088780 Stokes, T. F., & Baer, D. M. (1977). An implicit technology of generalization. Journal of Applied Behavior Analysis, 10, 349. doi:10.1901/jaba.1977.10-349 Sundberg, M. L. (2007). Verbal behavior. In J. Cooper, T. Herron, & W. Heward (Eds.), Applied behavior analysis (2nd ed. pp., 525-548). Upper Saddle River, NJ: Prentice Hall. Sundberg, M. L., & Michael, J. (2001). The benefits of Skinner’s analysis of verbal behavior for children with autism. Behavior Modification, 25, 698-724. doi:10.1177/0145445501255003 Tager-Flusberg, H., Rogers, S., Cooper, J., Landa, R., Lord, C., Paul, R., . . . Yoder, P. (2009). Defining spoken language benchmarks and selecting measures of expressive language development for young children with autism spectrum disorders. Journal of Speech Language and Hearing Research, 52, 643-652. 10.1044/1092-4388(2009/08-0136) Tarbox, R. S. F., & Najdowski, A. C. (2008). Discrete trial training as a teaching paradigm. In J. Luiselli, D. Russo, W. Christian, & S. Wilczynski (Eds.), Effective practices for children with autism: Educational and behavioral support interventions that work (pp. 181-194). New York, NY: Oxford University Press. Taylor, B. A., & Hoch, H. (2008). Teaching children with autism to respond to and initiate bids for joint attention. Journal of Applied Behavior Analysis, 41, 377-391. doi:10.1901/jaba.2008.41-377 Thompson, R. H., & Iwata, B. A. (2007). A comparison of outcomes from descriptive and functional analyses of problem behavior. Journal of Applied Behavior Analysis, 40, 333338. doi:10.1901/jaba.2007.56-06. Tiger, J. H., Hanley, G. P., & Bruzek, J. (2008). Functional communication training: a review and practical guide. Behavior Analysis in Practice, 1, 16-23. Twyman, J. (1996). The functional independence of impure mands and tacts of abstract stimulus 249 properties. The Analysis of Verbal Behavior, 13, 1-19. Retrieved from http://www.ncbi.nlm.nih.gov.proxy2.cl.msu.edu/pmc/articles/PMC2748504/ Ulman, J. D., & Sulzer-Azaroff, B. (1975). Multielement baseline design in educational research. In E. Ramp & G. Semb (Eds.), Behavior analysis: Areas of research and application. Englewood Cliffs, NewJersey: Prentice-Hall. Volkmar, F., & Lord, C. (2007). Diagnosis and definition of autism and other pervasive developmental disorders. In F. Volkmar (Ed.), Autism and pervasive developmental disorders (2nd ed., pp. 1-31). New York, NY: Cambridge University Press. Vollmer, T. R., Northup, J., Ringdahl, J. E., LeBlanc, L. A., Chauvin, T. M. (1996). Functional analysis of severe tantrums displayed by children with language delays. Behavior Modification, 20, 97-115. doi:10.1177/01454455960201005 Wacker, D. P., Berg, W. K., & Harding, J. W. (2008). Single-case research methodology to inform evidence-based practice. In J. Luiselli, D. Russo, W. Christian, & S. Wilczynski (Eds.), Effective practices for children with autism: Educational and behavioral support interventions that work (pp. 61-79). New York, NY: Oxford University Press. Wacker, D. P., Harding, J. W., & Berg, W. K. (2008). Evaluation of mand-reinforcer relations following long-term functional communication training. Journal of Speech Language Pathology and Applied Behavior Analysis, (2.4-3.1), 25-35. doi:10.1901/jaba.2008.-25 Wacker, D. P., Steege, M. W., Northup, J., Sasso, G., Berg, W., Reimers, T., . . . Donn, L. (1990). A component analysis of functional communication training across three topographies of severe behavior problems. Journal of Applied Behavior Analysis, 23, 417-429. doi:10.1901/jaba.1990.23-417 Wallace, M. D., Doney, J. K., Mintz-Resudek, C. M., & Tarbox, R. S. F. (2004). Training educators to implement functional analyses. Journal of Applied Behavior Analysis, 37, 89-92. doi: 10.1901/jaba.2004.37-89 Warren, S. F., Yoder, P. J., Gazdag, G. E., Kim, K., Jones, H. A. (1993). Facilitating prelinguistic communication skills in young children with developmental delay. Journal of Speech and Hearing Research, 36, 83-97. Watson, J. B. (1913). Psychology as the behaviorist views it. Psychological Review, 20, 158-177. Retrieved from http://chasqueweb.ufrgs.br/~slomp/edu01011/watson-behaviorist.pdf Wert, B. Y., & Neisworth, J. T. (2003). Effects of video self-modeling on spontaneous requesting in children with autism. Journal of Positive Behavior Interventions, 5, 30-35. doi:10.1177/10983007030050010501 Wetherby, A. M., Warren, S. F., & Reichle J. (1998). Introduction to transitions in prelinguistic communication. In A. Wetherby, S Warren, & J Reichle (Eds.), Transitions in 250 prelinguistic communication (pp. 1-11). Baltimore, MD: Brookes. Wilczynski, S. M., & Christian, L. (2008). The National Standards Project: Promoting evidencebased practice in Autism Spectrum Disorders. In J. Luiselli, D. Russo, W. Christian, & S. Wilczynski (Eds.), Effective practices for children with autism: Educational and behavioral support interventions that work (pp. 37-60). New York, NY: Oxford University Press. Wolf, M. M. (1978). Social validity: The case for subjective measurement or how applied behavior analysis is finding its heart. Journal of Applied Behavior Analysis, 11, 203-214. doi:10.1901/jaba.1978.11-203! ! Wolf, M. M., Risley, T. R., & Mees, H. L. (1964). Application of operant conditiong procedures to the behaviour problems of an autistic child. Behavior Research and Therapy, 1, 305312. Yoder, P., & Stone, W. L. (2006). A randomized comparison of the effect of two prelinguistic communication interventions on the acquisition of spoken communication in preschoolers with ASD. Journal of Speech, Language, and Hearing Research, 49, 698711. doi:10.1044/1092-4388(2006/051) Zimmerman, I. L., Steiner, V. G., & Pond, R. E. (2002). Preschool Language Scale (4th ed.). San Antonio, TX: The Psychological Corporation. ! 251