On the Structure of /iu/ and /ui/ in Sixian Hakka

Chung (1989:87) claimed that /iu/ and /ui/ are both falling diphthongs in Liudui Hakka, which is similar to Sixian Hakka, the dialect to be discussed here, in most aspects. This paper takes a partially different position that /iu/ is a falling diphthong, while /ui/ is a rising diphthong from the angles of co-occurrence restriction, rhyming and acoustic measurement. Chung (1989:121) argued that every Hakka syllable has an XXX template, with the nucleus in the middle. The ability to precede a consonant distinguishes a rising diphthong from a falling diphthong. On the premise that the former occupies one skeletal slot, while the latter occupies two, no extra slot is available to accommodate a consonant after a falling diphthong. Sixian Hakka makes no exception, as shown below (Yu, 1984; Luo, 1985). (1) rising diphthongs falling diphthongs ia – iam, iaN, iap, iak (N = velar nasal) ai -*aiC ie – ien, iet au -*auC io – ion, ioN, iok eu -*euC ua – uan, uat oi -*oiC ue – uen , uet One thing that merits attention in (1) concerns ien and iet. We argue that they are just phonetic realizations of /ian/ and /iat/ due to some sort of co-articulation effect. In Sixian Hakka, ien rhymes with /am/ and /aN/. If ien occurs phonemically, the consequence is: members of the rhyming set can be different at any level of phonological representation. An immediate question lies in what motivates rhyming then. Suppose the line of thinking is correct, the rising diphthong /ie/ happens to be not followed by any consonant. Accidental gaps occur. To take a step further, the presence of a following consonant is an indicator of a rising diphthong, and yet the absence of a following consonant does not necessarily identify a falling diphthong. By analogy, /iu/ is a rising diphthong, while the compositional structure of /ui/ is obscure. (2) iu – iun, iuN, iut, iuk ui -*uiC In fact, the phonemicization of /iuC/ is somewhat misleading. Following Chung (1989:81-84), we contend that /iu/ is characterized by a falling diphthong instead. More precisely, /iu/ has two phonetic variants. [y] (the front counterpart of [u]) occurs before a consonant, and [iu] never does. [y] results from fusion of the two vocoids in /iu/ just as Particle Phonology (Schane, 1984) claims. In Sixian Hakka, the vowel coalescence is probably triggered by the demand that a falling diphthong should not precede a consonant. Rhyming furnishes additional argument for /iu/ to be a falling diphthong. An observation by the present author