Íslenskt mál og almenn málfræði - 01.01.1984, Page 147

Model ofModern Icelandic Syllable Types
145
C LATERAL
tal [tha:l]/[tha:l]
d NASAL
von [vo:n]/[v3:n]
auðn [œyðn]/[œyðn]
e TRILL
tár [thau:r]/[thau:r]
tafl [thaþl]/[thaþl]
safn [saþn]/[saþn]
Köln [khœln]/[khœln]
sífr [sivr]/[sivr]
bruðl [þrYðl]/[þrvðl]
hjálm [?aulm]/[gaulm]
The types of complex coda that occur are: [s$, sg, s$l, s$n, $n,
gn, þn, þl, gl, $1]. Forms exemplifying this are: kemst [chems$], heimsk
[heimsg], ærsl [aijs^l], ræksn [raixs^n], botn [þohdn], hölkn [hœjgn],
vakna [vahgna], vopn [vohþn], epli [ehþll], jökli [jœhgll], skutla
[sgYhgla].
There are several things that need further investigation here. For
example, we have seen that constraints prevail with respect to the
coocurrence of codas and nuclei, creating some gaps that are not im-
mediately explained within the framework examined here. Most im-
portant is the relation between what Haugen (1958) calls consonantal
vs. vocalic nuclei and the coda, in other terms between vowel length
and the following consonantism. Within the present framework, this
could be stated in terms of a constraint to the effect that nuclei that
are not ‘open’ and ‘held’ (=vocalic), demand a coda following them,
in other words consonantal codas cannot form full syllables. If the
vowel is short and the rest of the nucleus filled by some non-vocalic
material, either a consonant (voiced or voiceless) as in varð [varþ],
and verk [vejg] or by preaspiration (checked voicing and open articula-
tion) as in hattur [hah^Yr], then the nucleus must be ‘supported by’
a coda; [var], [vaj] or [hah] cannot occur alone. This relates back to
what is discussed at the end of Section 2, that is, the question of
how to deal with forms like sess and lögg. Two possibilities were ment-
ioned there. One was to assume that in the underlying structure the
codas are empty and the above mentioned constraint is satisfied by
a phonological rule (or transformation) that ‘copies’ the consonant-
ism. (‘Underlying’ ses becomes sess.) The other possibility considered
was to assume that the coda is filled by a consonant identical to the
one in the nucleus, that is, that sess and lögg have the same underlying
structure as hest and sögn, except that in the former the consonant
in the coda is identical to the one in the nucleus. In either case, the
íslenskt mál VI 10