PHILLIP BACKLEY & TOYOMI TAKAHASHI
In most privative approaches to melodic structure, the only phonological operations deemed possible are those involving the linking and delinking of primes. However, within an Element Theory model where no independent ATR prime is recognized, this restriction must be relaxed in order to accommodate cases of tongue root harmony. In response, we propose a view of melodic structure in which every position contains a full set of elements; phonological contrasts are then encoded by means of a lexical instruction to 'activate' individual primes. In this paper we demonstrate how these assumptions can provide a satisfactory treatment not only of ATR harmony, but of vowel harmony in general.