I analyze the biological underpinnings of human speech from a comparative perspective. By first identifying mechanisms that are evolutionarily derived relative to other primates, we obtain members of the faculty of language, derived components (FLD). Understanding when and why these evolved is central to understanding the evolution of speech. There is little evidence for human-specific mechanisms in auditory perception, and the hypothesis that speech perception is "special" is poorly supported by comparative data. Regarding speech production, human peripheral vocal anatomy includes several derived characteristics (permanently descended larynx, loss of air sacs), but their importance has been overestimated. In contrast, the central neural mechanisms underlying speech production involve crucial derived characteristics (direct monosynaptic connections from motor cortex to laryngeal motor neurons, derived intracortical dorsal circuitry between auditory and motor regions). Paleo-DNA from fossil hominins provides an exciting new opportunity to determine when these derived speech production mechanisms arose during evolution.