The root CV-template as a property of the affix: Evidence from Yawelmani

In this article, I have provided support for a skeletal core independent of any phonemic material. This is not a new theoretical claim, but rather adds to a small but growing literature (McCarthy 1979, 1981, Halle and Vergnaud 1980, Harris 1980, Marantz 1982, Yip 1982). However, the analysis here is an important addition because the skeleta are added to the grammer in an unfamiliar manner: affixes may determine the skeletal template of a root; if not, a default template is supplied, determined by a lexical diacritic on each verb root. Interestingly, recent work on Norwegian tone by Withgott and Halvorsen (in prep) suggests that when a suffix bears tone in Norwegian, the suffixal tone pattern surfaces on the word. With no affixes or with a toneless suffix, the underlying (or default) tone of the word surfaces. This parallels in tone the example in templates that Yokuts provides. In section 3, a CV-template pool consisting of the three default templates of verbs in Yawelmani was established. Certain affixes supply templates from this pool, and the phonemic melody of the root associates with the selected template according to universal conventions and the rule of V Spread (43). The assumption of a pool containing only three templates accounts for the pairing of bi- and triconsonantal forms when a template is selected by an affix, that is the CVC-CVCC, CVVC-CVVCC, and CVCVV-CVCVVC pairings. The triconsonantal template is selected in all cases. With biconsonantal roots, the third C-slot of the template has no segment associated with it, and so cannot surface. This explanation is elegant and concise, but is not available without the existence of an independent skeletal tier.