Nasal Harmony and Phonotactic Well-Formedness in Tupari

This paper describes, analyzes, and typologically contextualizes the nasal harmony system of Tupari, an understudied Tupian language of Brazil. In this language, voiceless stops obligatorily nasalize when in coda position following a nasal vowel. This process superficially defies typological claims made by Walker (2000; 2003), Piggott (2003), and others about the behavior of voiceless stops in nasal harmony systems. However, I argue that coda nasalization in Tupari should be analyzed as separate from nasal spreading proper: it is driven by an independent phonotactic principle requiring that all syllabic rhymes be uniformly nasal or oral. This principle has an articulatory basis in the fact that all coda stops are unreleased in Tupari, thereby lacking a release phase in terms of Steriade’s (1993) Aperture Theory. Separating nasal harmony and coda nasalization makes testable predictions about these processes’ mutual independence—predictions confirmed by two other Brazilian languages, Awetí (Mawetí-Guaraní) and Kaingang Paulista (Jê).