The origin of language syntax remains one of the central open questions in neuroscience. Two competing views dominate the debate: that syntax emerged as a discrete, species-specific novelty (the saltationist view), or that it arose gradually through the repurposing of pre-existing sensorimotor circuits. Here we review evidence in favor of the gradualist hypothesis and propose that three convergent elements, rooted in the sensorimotor brain, provided the evolutionary substrate for language syntax. First, the progressive expansion of direct cortico-motoneuronal (CM) projections from primary motor cortex to spinal motoneurons freed motor control from the rigidity of prewired synergies, enabling flexible, hierarchically nested action sequences. Second, the premotor system — particularly monkey area F5, the homolog of human Broca’s area — implements a goal-directed motor lexicon with mirror properties constituting a precondition for intentional communication. Third, during observation of others’ actions, the mirror system must simultaneously maintain competing goal hypotheses and resolve them incrementally — a process we term “hyperprediction” — that is computationally analogous to syntactic parsing. We further argue that Broca’s area is not a dedicated language hub but one node within a broader cortico-subcortical network: the basal ganglia implement hierarchical chunking and rule learning across motor and linguistic domains, as evidenced by shared neural substrates for tool use and language syntax. Finally, we propose that recursion is not language-specific, and that the co-evolution of hand morphology and cortico-basal ganglia circuits provides a mechanistically grounded account of how syntactic capacities emerged from sensorimotor precursors, with five falsifiable predictions testable by current methods.
