Unmasking the Bantu Orthographic Vowels

Bantu vowel phonemes are reflexes of the Proto-Bantu seven-vowel system /*i *ɪ * ε‎ *a *ɔ *ʊ *u/. While lax high vowels were supplanted in some systems because of vowel mergers in the first two degrees /*i *ɪ/ and /*u *ʊ/, lax mid vowels / ε‎ ɔ/ are attested across most Bantu languages either underlyingly or at surface. Widespread use of roman orthographic vowels has left the phonemic status of mid vowels fuzzy. Here the orthography is treated as a “mask” disguising the phonetic quality of vowels, to be “unmasked” with the help of proper documentation and description. With examples from endangered Bantu languages of Tanzania and from Swahili current vowel documentation methodologies and theoretical approaches for unmasking are discussed. The distribution of mid vowels is characterized with a theory of markedness which contributes to understanding why lax mid vowels may be either triggers or targets of harmony and why a low vowel may be opaque or transparent to harmony.