Word Learning in Children With Autism

Overview

The project highlights one of the primary areas of research within the KIDDRC- language and communication. The focus is on language and communication in children with autism and minimal verbal skills (less than 20 spoken words). Remaining nonverbal past the age of 5 years has been considered a poor prognostic indicator for future language developments, yet few interventions have been developed to address this problem. The Specific Aims for this project are (1) to further investigate a multimodal intervention for school-age children with minimal verbal skills-defined as less than 20 words spontaneously spoken, signed, or selected via graphic symbol selection-and (2) to identify significant covariates associated with differential responding to the intervention. The research addresses an unmet need to promote spoken word production in children who remain essentially nonverbal well past the ages associated with speech acquisition. The project is also innovative because: a) it investigates a multimodal intervention based on principles of phonotactic probability and neighborhood density in combination with augmentative and alternative communication (AAC), and b) it investigates novel predictors of treatment response that are obtained through cutting-edge technologies. This intervention will have better success than past interventions because the intervention will provide increased input through speech, digitized speech and visual images and additional speech sound practice for words that are comprised of high frequency sounds in the child's repertoire. Extant speech sounds in each participant's repertoire will be identified using LENA™ digitized recordings. Vocabulary words will then be selected based on a child's speech sound repertoire and principles of word learning-words with high probability speech sound sequences will be selected and taught with either multimodal intervention or a treatment as usual condition. Responses to these interventions will be evaluated using a Sequential Multiple Assignment Randomized Trials (SMART) design. Different outcomes may be associated with individual and environmental predictors identified in our previous research. Individual predictors include verbal comprehension, imitation skills, adaptive behavior, nonverbal speech sound repertoire, and communication complexity. Communication complexity will be measured with the Communication Complexity Scale (CCS), developed by the Principal Investigator. Environmental predictors include language input to the child as measured with LENA™ recording devices. Results will determine if the multimodal intervention is more successful than treatment as usual for teaching word productions.

Communication is an essential aspect of life for all individuals including children with autism. Many children with autism are severely impaired in their verbal communication development and there is a need for additional research demonstrating the effectiveness of interventions that improve communication for participants with autism and severe communication impairments. The proposed study directly follows from a recently published pilot study demonstrating both the feasibility and importance of an intervention designed for children with autism and minimal verbal skills (less than 20 words pre-intervention) (Brady et al., 2015). The project will follow up this pilot study with a two stage SMART (Sequential Multiple Assignment Randomized Trial) design that will allow comparison of results from an experimental intervention to a treatment as usual condition, and compare two intensities of multimodal intervention. Further, the project will investigate how proposed variables measured at the outset of intervention predict responsiveness to the treatment conditions. Aim 1. Investigate the effectiveness of a multimodal intervention aimed at teaching students with autism and minimal verbal skills to produce new words with speech and augmentative communication (AAC). Briefly, the multimodal approach combines joint book reading, computerized instruction, and AAC. The AAC component teaches participants to select target words on speech generating devices in communication routines. All components will focus on teaching a set of vocabulary selected for each participant based on results from digitized recordings of sounds produced over a 12-hour time period. Target vocabulary will be one-syllable words that consist of sounds in each child's phonetic repertoire in sequences that are highly represented in the English language (i.e., high phonotactic probability) and determine if increasing intensity leads to better outcomes for participants who do not respond to the initial intervention. Increased intensity or regular multimodal intervention may be better for those who do not respond to treatment as usual. Primary outcome measures include: * Total number of words each child learns to say and the number of words each child learns to produce with AAC during intervention. * Total number of words each child says in generalization contexts. Generalization of word use to classrooms and at home will be measured with recordings from generalization probes conducted in homes and classrooms. Aim 2. Investigate potential predictor variables for individual differences in learning outcomes measured in Aim 1. Results from Brady et al. (2015) showed 5 of 10 participants were "high responders" to multi-modal intervention - learning 18 or more new words over approximately 2 months of intervention. However, three participants learned to say only a few words and two participants essentially made no progress. The project will examine individual predictors of differential responding and identify variables that will enable clinicians to personalize intervention according to specific environmental and participant characteristics. The proposed predictor variables are: * Environmental variables: Based on previous research linking the amount and type of environmental input to children's word acquisition, vocabulary input from parents and teachers will be measured with automated speech recognition technology. Existing technology available through LENA™ software (Warren et al., 2010) will be used to measure total number of words produced by communicative partners and conversational turns in home and school environments. * Participant characteristics: Based on the pilot investigation as well as previous longitudinal studies, the following specific child variables are proposed to be predictive of responsiveness to one of the experimental interventions: verbal imitation, receptive vocabulary, adaptive behavior, autism severity, and nonverbal communication complexity measured with the Communication Complexity Scale (CCS; Brady et al., 2012). Results will drive the development of more effective adaptive language interventions for minimally verbal children with autism and related intellectual and developmental disabilities.