Glide formation is not motivated by onset requirement in Malawian Tonga, glide epenthesis is

Abstract

This article discusses the distinct motivations for glide epenthesis and glide formation in hiatus contexts in Malawian Tonga. It argues that glide epenthesis is triggered by the need to create a canonical syllable structure, one that has a proper onset, while glide formation (and deletion where glide formation is disallowed) derives from an avoidance of vowel sequences. Furthermore, glide epenthesis seems to apply where the prefixes are affixed to the prosodified constituents, while glide formation is used when the prefix has been affixed to a morphological constituent. Empirical evidence has been derived from the realisation that the constituent whose onset is the epenthetic glide can also stand on its own as a phonological word/stem such as in citation or imperative form. In other words, glide insertion may occur where there is hiatus involving a prefix and a stem, but it also occurs before vowel-initial words/stems in situations where there is no hiatus. The second piece of evidence is based on optional insertion of featurally unspecified syllables between prefixes and monosyllabic verbs, which is an indication that certain prefixes optionally require that they be affixed to a proper prosodic constituent, one that has a proper onset or one that is minimally bisyllabic.