Introduction

In Indonesian, the voicing of the root-initial obstruent determines the outcome of /məŋ-/ prefixation. When that consonant is voiceless, it coalesces with the prefix-final nasal to produce a nasal with the same place of articulation as the obstruent, in a process referred to as nasal substitution (e.g., /məŋ + paksa/ → [məmaksa] ‘to force’). When the root-initial obstruent is voiced, though, simple place assimilation results (/məŋ + buat/ → [məmbuat] ‘to make/do’).

In traditional analyses of nasal substitution, the limitation to voiceless consonants is expressed as a featural restriction on the scope of the relevant rule. No attempt is made to derive this restriction from principles active elsewhere in Indonesian, or in other languages. Pater (1999) points out that Indonesian is far from alone in its avoidance of nasal-voiceless obstruent clusters and invokes a substantive output constraint against these clusters, *NC, as the formal driving force behind processes like nasal substitution, postnasal voicing, nasal deletion, and denasalization. The Optimality Theoretic ranking between *NC and faithfulness constraints determines which of these routes a language chooses to take to eliminate nasal-voiceless obstruent clusters.

Despite the relative success of the *NC account in formally connecting nasal substitution to a crosslinguistic range of phonological processes, in this chapter I will argue for a reanalysis, based on evidence from other Austronesian languages as well as Indonesian itself.