Íslenskt mál og almenn málfræði - 01.01.1984, Side 141

Model ofModern Icelandic Syllable Types 139
ends in a consonantal element and the vocalism is short, the latter
type is exemplified by forms like vor and te, where there is no conson-
antism in the nucleus and the vowel is long. There is interdependence
between the nucleus and the coda in that if material is added at the
end of a form like vor, with a ‘vocalic’ nucleus and a simple coda,
the [r] has to appear in the nucleus, and thus the nucleus becomes
consonantal. That is, the distribution of vocalic and consonantal nu-
clei is predictable on the basis of the following consonantism accor-
ding to the Icelandic ‘length-rule’ (cf. Árnason 1977b; 1980, Ch. 2).
Consonantal nuclei need a consonant in the coda, and we can say
that vocalic nuclei do not allow more than one consonant in the coda.
(Twill return shortly to the interrelation between nucleus and coda
in ‘geminate’ forms like mann, sess, ball and lögg.) In the forms men-
tioned in footnote 1 (like flys.ja), the assumed syllabication accounts
for the distribution of length, i.e. the same holds for these forms as
for vor, and vor.ið. This can be seen as motivation for the intermedi-
ate unit rhyme in the hierarchy. There is a closer unity between the
nucleus and the coda than between onset and nucleus or between
onset and coda. (This is of course nothing new, it is also reflected
in poetic rhyme; the things that ‘rhyme’ are the ‘rhymes’.)
As for motivation for the division of the rhyme into two units, it
may be mentioned that there are strict limits as to what sort of relation
the vocalism may enter into with a following consonantism. For exam-
ple the addition of more complexity into the consonantism does not
seem to affect the nucleus in forms like lagst [laxs^] ‘lie down’ (past
participle) or ræksn [raixs^n], ‘mesh, rag’, the phonological ‘length’
of the vocalism is unchanged (although one might find phonetic short-
ening of duration conditioned by the added material), and one might
say that the nucleus was ‘saturated’ by the one consonantal unit [x].
Thus the parsing assumed here is: s[o[l]o r[n[3X]n c[s(?]c ]r ]s and
s[ o[r]o r[ N[aix]N cNn]c] r ]s- It is assumed that the nucleus is
independent and has limited room, and any consonantal material that
is added in excess of one segment will have to appear in the coda,
and if there is only one segment, the nucleus will, as we saw, be
vocalic, or we might want to call it ‘free’.
We have seen forms with an empty coda and a vocalic nucleus,
but we might wonder whether forms with an empty coda and a con-
sonantal nucleus exist. It might be suggested that forms like mann