Tonal (non-)transfer in Kunming Reduplication

In this paper, I discuss tonal reduplication in Kunming, which has received little attention in the formal linguistics literature. Based on first hand data, I show that tonal reduplication in Kunming is interesting in two respects. First, a Base tone is not always transferred to the RED (reduplicant); while a Hr (high register) Base tone (i.e., yinping 44, shangsheng 53) is transferred to the RED, the RED corresponding to a Lr (low register) Base tone (i.e., yangping 31, qusheng 11) surfaces with a non-correspondent high tone. I propose that the non-corresponding high tone comes from a floating tone that accompanies reduplication and is only realized on the RED corresponding to Lr Base tones, but not to Hr Base tones, because Hr Base tones are prominent and more easily preserved on the RED. Second, as far as yinping tone sandhi is concerned, though it applies normally on the Base and outside the BR complex, on the RED it unexpectedly underapplies when the RED carries a yinping tone copied from the Base. I propose that tone sandhi underapplies only on a copied yinping tone, and not elsewhere, because only underapplication of yinping tone sandhi in this case can help improve BR Identity.