Surface processes, word minimality and stress assignment in Blanga

Considering word minimality and assuming prosodic structure maximalisation, Hayes’ (1995) metrical stress theory predicts that if a syllabic trochee language that does not necessarily distinguish segmental length phonologically shows some evidence of heavy syllables, then a heavy syllable at one edge of the parse constitutes a proper syllabic trochee. At the same time, the author points out that the prediction is difficult to check due to lack of available data. This paper provides supporting evidence from Blanga, a recently documented Austronesian language of the Solomon Islands with no contrastive segmental length and no underlying heavy syllables, in which underlying stress is always penultimate and thus uses the syllabic trochee, a non-quantity-sensitive type of foot, as the minimal metric constituent. However, surface processes generate heavy syllables in all positions and stress on the last syllable if heavy, suggesting that the language counts moras, rather than syllables, for purposes of stress assignment. Applying the principles and constraints postulated by Hayes, and thus confirming their validity, I will reject the moraic trochee hypothesis and show that the underlying stress pattern is preserved at the surface by allowing a heavy syllable to form a trochee by its own as long as it occurs either in parse-final or parse-initial position and it is the only heavy syllable available at the word level