This article contributes to the case for reconstructing verb agreement for Proto-Tibeto-Burman. It shows first that, given the distribution of cognate agreement systems across the family, there is no alternative to reconstructing it for the proto-language. Secondly it describes the paths by which agreement has been lost in those languages where it is absent.
Evidence is presented to demonstrate the prevalence of evidence for the PTB paradigm in languages across the family. It is shown that, contrary to assertions which have been made in the literature, the agreement systems of Jinghpaw, Nocte, and Northern Chin are cognate to those of the socalled “Rung” branches (Kiranti, rGyalrongic-Qiangic, Nungish, and West Himalayan), and that even without, but especially with, this evidence the “Rung” hypothesis is inconsistent with other proposals for subgrouping Tibeto-Burman. Once the cognacy of the Jinghpaw and Nocte systems is recognized, there is no further reason to believe in a genetic “Rung” unit. Several case studies are presented which show that agreement systems can be quickly and easily lost in TB languages, as a result of intense language contact and/or through the replacement of older finite structures by innovative new constructions based on clausal nominalization.