Based upon the fact that the 'mora nasal' in Japanese behaves as a vowel in terms of metre, that nasal is often assumed to occupy a 'weight-bearing' position (i. e. nucleus or mora) in a syllable. The purpose of this paper is to put forward an alternative view of the representation of the 'mora nasal'.
The distribution of the nasal is restricted to the position preceding a homorganic consonant, prevocalic position and word-final position.
(1) (a) ondo 'temperature' (b) hoNi 'one's real intention' (b) hoN 'book'
tô (1986) claims that all nasals in (1) should be syllabified in a moraic coda position. Yoshida (1991), on the other hand, suggests that the nasal should occupy a nucleus position in the case of (1b) and (1c).
The problem shared by both of these analyses is that morphological concatenation must trigger resyllabification. ln Itô's case, for example, a verb such as sin 'to die' is inflected as follows:
(2) sin 'to die' (coda)
si.nu '(somebody) dies' (onset)
sin.da '(somebody)died' (coda)
In Yoshida's case, as shown below, the 'mora nasal' resyllabifies from a nucleus into a coda position in cases of compounding.
(3) si.N'new'+ka 'turning'→sin.ka 'evolution'
However, this kind of strudure-changing operation entails a degree of arbitrariness and weakens the retrictiveness of a theory. It would be desirable, therefore, to eliminate this altogether.
I will purpose, with recourse to the notion of Phonological Licensing developed in Harris (1994), that the 'mora nasal' should always appear in an onset position. My analysis posits no resyllabification procedure, and therefore, gains an advantage over the above analyses in terms of restrictiveness.