Íslenskt mál og almenn málfræði - 01.01.1984, Blaðsíða 139
Model ofModern Icelandic Syllable Types 137
short, can be accounted for in a simple way with the aid of the syllabi-
cation described above (cf. Árnason 1980).
Although the division hest-ur etc. looks very much like a morpho-
logical analysis in many cases, there are other cases that show that
there is not a complete correspondence between this syllabication and
morphological analysis. There are disyllabic morphemes like biskup
‘bishop’, ketill (acc. ketil) ‘kettle’, kastali ‘castle’, akur ‘field’, and
even trisyllabic ones like almanak ‘almanac’. Also, portmanteu rep-
resentations in forms like börn ‘children’, where both ‘plural’ and
‘child’ are represented by one syllable show that the mapping is not
direct. Still, it must be admitted that the stressed syllable comes close
to being a canonical form for units of lexical meaning and the un-
stressed syllable a canonical form for units of grammatical meaning.
It seems wise, then, not to exclude the possibility that morphology
is partly responsible for the organization that will be investigated in
the following. But the organization is basically phonological although
it has morphological correlations.
On the grounds of the above considerations, then, the syllabication
adopted for the purposes of this paper will be the maximal one sug-
gested in Árnason (1977b), although this is not crucial, and the basic
approach examined here could be used within a different approach
to the syllabication problem.1
1 One fact needs special mention. There are bisyllabic forms with complex intervocal-
ic consonantism that are reluctant of formation of monosyllables with the whole conson-
antism retained. These are forms like flysja [flnsja] ‘to peel’, belja [belja] ‘to bellow’,
heiðra [heiðra] ‘to honour’, stöðva [sdœðva] ‘to stop’. All of these have in common
that they end in /v/, /j/, or /r/. Forms like flysj, belj, heiðr, stöðv are either marginal
or not permitted at all (cf., e.g., Bernódusson, 1978). There is a split in the phonological
patterns of these forms, though (cf. Árnason 1981), in that those with /p/, /tj, /k/, or
/s/ in front of the /j/, /v/, /r/, have a long vocalism, and the other ones a short one.
The former forms could easily be syllabified flys.ja, nep.ja etc., and then the first syllable
could have the same phonological status as the potential (semantically empty) monosyl-
lables flys [fli:s] and nep [ne:b]. But this sort of analysis runs into difficulties with forms
with a short vocalism. There is no monosyllabic belj to motivate the syllabication belj.a
and there is no monosyllabic bel (with a short vowel and a single consonant) to motivate
the syllabication bel.ja.