Grounding the computational principles of language in neurobiology requires cross-modal and cross-linguistic data.

Abstract

Murphy's discussion (2025) of his recent ROSE model includes explicit linking hypotheses connecting computational, algorithmic, and implementational levels in the study of language and its neurobiological basis. Here, I argue that establishing the neural basis of the abstract principles underlying natural language syntax will require new data from sign languages, tactile sign languages, as well as typologically diverse spoken languages. The assumption of modality-independent processes for structure building lies at the heart of ROSE, but the proposed correlates for hierarchical and sequential operations must be subjected to empirical test across languages and modalities in the future.