Association for Behavior Analysis International

In comprehensive treatment of autism, generative language matrix performance is being developed in a coherent conceptual framework, enabling the organizational management of productive treatment planning, trouble-shooting, and program evaluation. A four-dimensional matrix of social language skills is used to design an overall generative process of language development. The matrix of skills is addressed across generalization modalities, syntax forms, conditional discriminations, and functional communicative relationships. After generative receptive and expressive skills are developed in single-term modes, recombinative generalization is developed through matrix training; folloed by recombinative generalization in comprehension and creative language production matrix training. The organization of the language curriculum is used to control the pacing of both language and related social skills in a systematic manner, in order to result in optimal acceleration. Three studies will present data obtained from children in EIBI over the entire scope and sequence of the language matrix system.

Practitioners of EIBI; Academic Faculty

In intensive early intervention, a basic programming question often is: “how many exemplars should be taught in each program?” In the overall course of comprehensive treatment for autism, behavior therapy is initiated by establishing expressive and receptive single-term labels/tacts. Treatment is continued until the various terms of a sentence become generative response classes. Then matrix training of the individual terms is continued until recombinative multiple-term conditional discriminations are established. Recombinative matrix generalization is a special form of generative response classes. Finally, more complex and abstract comprehension modes are established and training continued until the comprehension forms become recombinative and generalized. Thus the answer to the initial question is that all programs are taught until the exemplars become generative. It is suggested that behavior therapy may more readily progress through higher levels of complexity when thee lower levels are taught until they meet generative criteria. In this investigation, clinical data on two young children’s performances with language matrix programming are presented. A multiple baseline within-subject design including systematic language matrix teaching and probes for generalization was used to document the development of generative response classes and recombinative multiple-term conditional discriminations. The progress of each child was individualized according to their baseline levels, and rates of acquisition. The systematic matrix training resulted in development of generative single-term response classes, recombinative multiple-term conditional discriminations. Generalization was programmed across different stimulus and response modes (e.g., receptive, expressive, written, comprehension) and taught across different sentence terms (e.g. subjects, actions, adjectives, prepositions). Individual programs were taught until generalization occurred to the first presentation of a novel recombination of exemplars embedded with novel distractors.

In the overall course of comprehensive treatment for autism, behavior therapy is initiated by establishing expressive and receptive single-term labels/tacts. Treatment is continued until the various terms of a sentence become generative response classes. Then matrix training of the individual terms is continued until recombinative multiple-term conditional discriminations are established. Recombinative matrix generalization is a special form of generative response classes. Finally, more complex and abstract comprehension modes are established and training continued until the comprehension forms become recombinative and generalized. It is suggested that behavior therapy may more readily progress through higher levels of complexity when thee lower levels are taught until they meet generative criteria. In this investigation, clinical data on a young child’s performances with language matrix programming are presented. A multiple baseline within-subject design including systematic language matrix teaching and probes for generalization was used to document the development of generative response classes and recombinative multiple-term conditional discriminations. The systematic matrix training resulted in development of generative single-term response classes, recombinative multiple-term conditional discriminations. Generalization was programmed across different stimulus and response modes (e.g., receptive, expressive, written, comprehension) and taught across different sentence terms (e.g. subjects, actions, adjectives, prepositions). Individual programs were taught until generalization occurred to the first presentation of a novel recombination of exemplars embedded with novel distractors.

Keene & Larsson (2013) presented a study of training of social comprehension that utilized multiple exemplars to develop generative social comprehension in children with autism. This extension of that study provides further data to show how the generative social comprehension followed prerequisite recombinative generalization through matrix training, and then itself consisted of recombinative generalization. A multiple probe design across five common childhood social concepts (e.g., sharing) was employed for each of the three children who participated in this study. Probes were conducted on the first presentation of novel children’s books as stimuli and the proportion of correct responses to the questions was measured. Training on each social concept continued until a generative mastery criterion was met in which the child responded correctly to at least 14 out of 16 questions on three consecutive novel books. The results showed that all of the children were able to answer an increasing proportion of the questions correctly to novel children’s books as stimuli. Generalization probes across untrained in-vivo social scenarios were also assessed. The children responded to a high percentage of novel questions regarding the social scenarios. In this present extension, data will be presented which shows the sequence of matrix training that preceded the implementation of social comprehension programming, and conclusions will be offered on the appropriate scope and sequence of matrix training curriculums.

Luckey, Pelletier, Miller, & Larsson (2013) presented the methods and outcomes of the implementation of systematic dynamic programming in a EIBI setting. The present study will present a replication and extension of that study, by demonstrating the use of dynamic programming to manage the implementation of matrix training in a comprehensive treatment program for autism. The use of organizational behavior management is critical to ensuring that all children are receiving the most effective matrix programming, and that treatment is optimally accelerated. Overall program evaluation data will be presented on 54 children undergoing generative language matrix programming in EIBI for autism. In addition, specific within-subject controlled studies of the treatment team performance with two children will show the effects of a system for management of clinical outcomes. During baseline, common non-dynamic management systems were in place to manage the children's language matrix programs. The clinical management system, known as Dynamic Programming was introduced via a multiple baseline design across children. Dynamic Programming is an intervention package that includes: (a) therapist self-monitoring while teaching new program exemplars, (b) therapist public posting of child mastery, (c) probes of child behavior to confirm generative mastery of matrix training, (d) dynamic adjustment of daily treatment targets based upon performance, and (e) dynamic adjustment of monthly objectives criteria based upon performance data. Results suggest that the children's rate of acquisition of generative language was accelerated through the implementation of Dynamic Programming.