So-called "gesture-first" accounts of language origins suggest that human language began with gestural communication, including movements of the hand and face. However, a prevailing problem for such theories has been to explain how language moved "from hand to mouth"-that is, how vocal communication may have gradually replaced gestural communication, to become the predominant mode of language. Here, we present a mechanism by which this transition may have been possible. We specify several intrinsically iconic aspects of vocal behavior, related to embodied cognitive processes. We argue that, making use of pre-existing relationships between sound and non-sound (including physical movement and internal physiological states), vocal iconicity may have paved the way for a shift from gesturing together with a relatively inflexible vocal systems to flexible, open-ended spoken language with gestural complement.