Features, segments, and the sources of phonological primitives

I review the role of distinctive features in early generative theory, focusing on their multifaceted role as defined by Chomsky Halle (1968) for capturing contrast, phonological patterns, and the phonetic realization of these patterns. Based on evidence from these multiple aspects of phonological representation, I conclude that the characterization of segments as combinations of universally-defined distinctive features is approximately, but not literally, correct. This leads to the question of how young children learn the elements of their phonology to achieve the knowledge of an adult phonological system? Crucially the evidence suggests that how we learn is not the same as what we know. Rather, an approximately categorical and compositional system is learned out of a more continuous, gradient one.