An iterative probabilistic model of speech-related vocalization rate growth due to child-caregiver interaction

摘要

Over the course of the first four years of life, the proportion of children's vocalizations that are speech-related increases steadily. The rate of this growth is reduced for children with autism spectrum disorder (ASD) and for children from households with relatively lower socioeconomic status (SES). The present study attempts to model this set of findings, treating adult responses as reinforcers of child behavior. The model starts with a 50% chance of producing a speech-related vocalization and gradually increases this probability by updating its speechrelated vocalization and not-speech-related vocalization probabilities each time a response is received. Numbers of vocalizations per day and rates of adult responding to the two vocalization types are drawn from human data to create high SES typically developing (TD), high SES ASD, low SES TD, and low SES ASD versions. The model shows growth in speech-related vocalizations that matches well to that observed for the human children and that matches the differences observed across clinical and SES groups. Some aspects of speech-related vocalization development are not well accounted for by the model; possible explanations and extensions are proposed.