Íslenskt mál og almenn málfræði - 01.01.1979, Qupperneq 247

Pronunciation of modern Icelandic rövl(a) and slafneskur 227
‘nonsense’ and the verb rövla [rœvla] ‘talk nonsense’. As Gunnlaugur
Ingólfsson of the OHÍ (== the University of Iceland Dictionary Project)
has informed me per litteras, the oldest example of rövla in the OHÍ is
in a private letter dated in 1887 (spelled röfla\ see Ólafur Davíðsson,
Ég lœt allt fjúka, Reykjavík, 1955:144), and of rövl in books published
in the mid-twentieth century, but written by persons bom in 1885 and
1887. (See Þórir Bergsson, Á veraldar vegum, Reykjavík, 1953:19;
Þórbergur Þórðarson, Viðfjarðarundrin, Reykjavík, 1943:140.) It can
be assumed that rövl(a) has been in the language for at least a century.
Rövl has its origin in the synonymous Danish noun vr0vl [vBœu?l],
and rövla in the synonymous Danish verb vr0vle ['vBœuIa]. The follow-
ing adaptations took place in the naturalisation process: (1) The Danish
initial v was dropped in Icelandic, because no Icelandic word begins
with vr. (2) The Danish uvular fricative [k] was replaced by the corre-
sponding Icelandic r-sound, the apical trill [r]. (3) The Danish diph-
thong [oeu], non-existent in Icelandic, was replaced by [œv]. (4) The
glottal stop, unknown to Icelandic, was dropped. (5) The Danish / has
‘remained’, but undergoes the rules of Icelandic external sandhi; speci-
fically, it is only half voiced before a pause. (6) Both lexemes have
assumed Icelandic inflection.
As a result, rövl and rövla now look native, except in one respect:
they contain [vl] instead of [1?1]/[|?1]. No tendency to replace the [vl]
with [þl]/[^l] has been observed so far, although such a tendency is
expected, seeing that there is a v-to-ó rule in the language. The only
way to describe this state of affairs in a generative phonology of Ice-
landic is to say that rövl and rövla are exceptions to rule (3). However,
this is not satisfying: the pronunciation of rövl(a) and rule (3) have now
been stable for a long time; if rövl and rövla were exceptions, we would
expect either rövl(a) to (begin to) assume [l?l]/[bl] or further exceptions
to accrue, or both.
Kiparsky’s universal makes it possible to treat rövl(a) as normal
lexical items. The universal says that rule (3), being a neutralisation
rule, applies only to derived forms. Rövl- is not a derived form accor-
ding to Kiparsky’s definition of the term (its vl is morpheme-internal
and present in the underlying representation of the two lexemes), ergo
the rule does not apply to it. By contrast, the /vl/ of /(Ijœvlar/, a stage
in the derivation (4) of djöflar, is derived in the sense that the /vl/ is