Uncontroversial evidence of vocal production learning, the capacity to modify vocal output on the basis of experience, is sparsely distributed in the animal kingdom. We suggest that this is in large part due to a trade-off between vocal learning complexity and a much more widely distributed trait—non-vocal dexterity. We argue that given some generally required neural and anatomical conditions for vocal production learning, species lacking both the (manual) appendages fine-tuned for grasping, manipulation, etc. and the neural control of those structures are more likely to display complex vocal learning. Conversely, the presence and control of these (manual) structures relegates the vocal apparatus to simpler vocal, feeding, and manipulation behaviors in other species. In other words, vocal learners tend to be flyers or swimmers. We also address the obvious exception to this generalization: humans are both highly dexterous and complex vocal learners. We hypothesize that the degree of bipedalism in land species and its connection with locomotion and breathing control is also a factor that helps shape the distribution of the vocal learning phenotype.