Why underlying representations?

Abstract

Phonology is a rapidly changing and increasingly varied field, having traveled quite some distance from its original structuralist and generative underpinnings. In this overview I address the status of underlying representations (URs) in phonology, which have been rejected by a number of researchers working in different frameworks. After briefly discussing the current state of phonology, I survey the arguments in favor of vs. against URs, considering recent surface-oriented critiques and alternatives. I contrast three straightforward abstract tonal analyses against the potential arguments which accuse URs of being (i) wrong, (ii) redundant, (iii) indeterminate, (iv) insufficient, or (v) uninteresting. Identifying two distinct goals in linguistics which I refer to as determining ‘what’s in the head?’ vs. ‘what’s in the language?’, I suggest, responding to some rather strong opinions to the contrary, that URs are an indispensable and welcome tool offering important insights into the typology of phonological systems, if not beyond.