Mary Paster

It is widely accepted that phonological rules/constraints do not ‘count’ past two. This is said to follow from locality restrictions, which do not allow anything to intervene between the target and trigger of a rule. In the domain of stress, it also follows from foot binarity. However, as I will discuss, there exist some phonological phenomena that are problematic for the standard notion of locality, especially in Bantu tone systems. In this talk I will give an overview of some phenomena that seem to require counting past two, focusing on particularly challenging cases from Kuria and Manyika. I will discuss ways of analyzing those two cases, present results of a pilot study designed to investigate the learnability of a counting rule, and conclude with arguments for allowing counting within our theory of phonology despite the apparent rarity of the phenomenon.