Íslenskt mál og almenn málfræði - 01.01.1999, Blaðsíða 177
175
Flugur, smágreinar og umræðuefni
sonority from the nucleus. This might be illustrated as shown in (3),
where the vowels are in the nucleus and the point of the < sign points
towards lesser sonority):
(3) The Sonority Sequencing Principle in syllables:
obstruents < nasals < liquids < glides < VOWELS > glides > liquids >
nasals > obstruents
The SSR predicts that the onset cluster /kl-/ in klænt [khlant] ‘thin’ is
a good cluster, since the liquid /1/ is more sonorous than the obstruent
^ and it is closer to the nucleus. The same holds for the sequence
'nt/ in the coda, where the nasal /n/ is closer to the nucleus than the
obstruent /tI. Conversely, the SSR rules out an onset sequence like /lk-
’ where the liquid is in front of the obstruent in the onset. Consequent-
y a word like *lkænt is impossible.4
Now if we compare the actual output form of metathesis in words
'he feskt [fekst] ‘fresh’ to the output that would be expected if no
^etathesis occurred, namely something like *[feskt], the SSH as
escribed in (3) does not rate one of the forms higher than the other
^nce no distinction is drawn between obstruents (fricatives and stops).
ne might think, however, that a fricative should if anything be more
s°norous than a stop and thus tend to occur closer to the nucleus of the
syllable. Then the non-metathesized version, with the [s] preceding
e [k] would correspond better to the SSP than the actual output does.
ut ibis is not entirely clear. Thus it has been known for a long time
j at /s/ has a special status among fricatives in certain respects. Thus
q ls Ihe only fricative that can precede a stop in the syllable onset in
rrnanic languages, such as Faroese, Icelandic and English for in-
p^nce’ i e. onsets like [sp-, st-, sk-] are fine in all these languages.
r°ese (and Icelandic!) examples include words like spinna ‘spin’,
Sjjp "ibere are examples in Faroese, as well as other languages that violate
Wo i actchhon, the language has various ways of “fixing” sequences that
sh ^ f Vlotate the SSP. Thus a sequence like the coda /-tn/ in vatn ‘water’
the a- not be possible. What happens is that the nasal becomes syllabic (cf.
ay . lscussion of syllabic sonorants in Sandpy 1994) and thus the violation is
0b 60' The same can happen in words with final sequences involving
ruents+liquids, such as /-tl/ in œtl ‘consideration’.