Project description

This project aims to characterize the adaptive mechanistic of language acquisition in autistic children.

Autistic children remain non-verbal for a period of several years, and 20% for their entire life. However, many of them present a spontaneous interest in letters and numbers beginning around the second year of life, through their use of web-based information. Our project aims to capitalize on this interest to expose them to letter-sound associations, by developing a personalized video recommendation system based on objects of interest (e.g., letters, numbers, shapes, and animals) to non-verbal autistic children (NVAC).

The principal objective of this research project is to adapt and develop a set of algorithms to identify and diversify the centers of interest of NVAC when watching to videos. This identification and diversification are carried out on videos taken from the listening histories of a group of NVACs.

For the field of autism research, this approach will aid in understanding the mechanisms through which NVAC make choices and learn written language, and how this can or cannot map oral language development. The research will provide a detailed, individual portrait of each child and identify the dimensions triggering autistic children’s interest at the group level. This approach will enable researchers in the field to verify specific hypotheses regarding the evolution of these children’s interests by automatically analyzing the videos that these children listen to a task that cannot be accomplished manually.