Monovalency and the status of RTR

Cases of ATR harmony are numerous and widely documented in the literature, whereas languages in which RTR is harmonically active (e.g. Wolof) have received a good deal less attention. A description based on the bivalent feature [±ATR] is capable of representing both harmony types, but fails to encode naturally the clear difference in typological markedness separating the two. On the other hand, an Element Theory approach — employing melodic headship as the basis of tongue root distinctions — is unable to provide an adequate account of RTR harmony without compromising privativeness. In response, I propose a tier geometry analysis (Backley 1995) which succeeds in capturing the alternation facts of both ATR and RTR harmony systems, together with the markedness characteristics of each.