Íslenskt mál og almenn málfræði - 01.01.1979, Page 248

228
Janez Oresnik
not present in the underlying representation of the form, but has come
into existence through the application of another rule.
If this treatment is correct, all the existent cases of non-altemating
morpheme-intemal [þl]/þl] from /vl/ must have arisen when rale
(3) was not yet a neutralisation rale, or in the special ways discussed
below in section 4. The universal predicts that all new words (neolo-
gisms, loanwords) containing non-alternating morpheme-intemal [vl]
will keep the cluster unchanged unless something artificial intervenes
(spelling pronunciation, fiat by those who think that [vl] is not genu-
inely Icelandic, etc.).
3. Modern Icelandic phonology also contains the rale
(5) v -» þ / V—n
I.e. any /v/ is replaced by /þ/ if immediately preceded by a vowel and
immediately followed by /n/ + vowel or by /n/ + word boundary.
— The rules (3) and (5) can be collapsed, but this is irrelevant in the
present context.
Examples: nom. pl. m. gefnir [-Ijn-] of gefinn [-v-] ‘given’, and gen.
pl. hújna [-þn-] of húfa [-v-] ‘cap’. Their derivations are stated sub (6):
(6) gjev + in + ir huv + na
vowel syncope g-jcvmr -
v -> þ é'i2!?1111 huþna
gefnir
húfna
These derivations do not contradict Kiparsky’s universal: in the deriva-
tion of gefnir, /vn/ is not underlying, but has come into being through
the operation of the vowel syncope rule; in the derivation of húfna the
cluster /vn/ is underlying, but its segments are crucially separated by
a morpheme boundary. Thus both instances of /vn/ satisfy Kiparsky’s
definition (2) of a derived form/input, and consequently the neutral-
ising rale (5) can apply in the derivations (6). (Rule (5) is neutralising
because of the many examples, such as höfn [hœþn] ‘harbour’, in which
[|?n]/[ljn] is realised in all the forms of their respective paradigms, i.e.
the result of the historical process has been lexicalised.)
However, rule (5) does not apply in the adjective slafneskur [-vn-]