Exploring Early Word Learning: The Shape Bias in Autistic and Non-Autistic Children

Abstract

Language development is a central part of early childhood, and differences in how children learn words can have lasting effects on communication. Autism is often associated with delays in language acquisition, yet the specific processes underlying these differences are not always well understood. One mechanism that supports early word learning is the shape bias, where children extend new object labels based on shared shape rather than other features such as color or texture. While this strategy has been well documented in non-autistic children, prior research has reported mixed findings regarding its presence in autistic children. The present study examined shape-bias performance in non-autistic children and compared their results with a previously tested sample of autistic children using the same remotely administered tasks. Participants completed both a noun and adjective task to assess whether they appropriately generalized labels based on shape versus other features. Results indicated that both groups demonstrated clear evidence of the shape bias. Non-autistic toddlers reliably prioritized shape, and autistic children performed at comparable levels, with slightly higher accuracy on noun trials. Notably, non-autistic children as young as two years old performed similarly to autistic children aged five to six when matched on vocabulary. These findings suggest that the shape bias is present in autistic children and is closely tied to vocabulary development rather than diagnostic status. When language ability is taken into account, autistic children exhibit word-learning patterns similar to those of younger, vocabulary-matched non-autistic peers. This study highlights the importance of early vocabulary growth and methodological considerations in understanding how children learn words. Keywords: autism, shape bias, language, generalizations