Sequence structure in children's speech reveals non-linear development of relations between syntactic and semantic categories

Why do children learn some words earlier than others? Can children's speech patterns reveal how their evolving representations of language determine what they learn? This study presents a systemic analysis of children's speech and uses low-dimensional embeddings to examine how the contextual knowledge reflected in their utterances reorganizes as linguistic experience increases. We show that the order and position of words in sequences produced by children from different age groups reflect changes in the way they represent categories of words. Rather than being ungrammatical, children's utterances seem to be structured by temporary grammars, which optimize the distribution of information in sequences. We demonstrate how semantic networks reorganize to support learning and suggest that English provides functionally ambiguous (multipurpose) categories to aid the reorganization of semantic spaces. These findings are intriguing and counterintuitive as they suggest that not knowing the exact meaning of words facilitates both learning and communication.