Gripla - 01.01.1977, Qupperneq 170

166
GRIPLA
ignored by the u-umlaut. Moreover, these stems are the only ones in
which a u that has not resulted from the u-umlaut of a intervenes
between the umlauted vowel and the end of the stem, cf. par. 3 below.
How these stems are to be treated in the descriptive grammar of Ice-
landic as far as u-umlaut is concerned, depends for one thing on the
way that vowel syncope is treated. To mention one possibility, suppose
the native learners of Icelandic analyze the stems not as losing a vowel
through a rule, but as containing a special set of endings, say, /faq +
Yr/, /faq + ran/, etc. In that case there would be no problem as far as
u-umlaut is concemed, for the u of the non-contracted forms would be
moved into the desinence, and thus any trace of u-umlaut operating
upon vowels separated from the end of the stem by the string /C0yC0/,
where the /y/ does not itself result from u-umlauting, would be gone.
(That a part of a stem becomes a part of an ending, is not unusual. Cf.
Middle English cheris(e analyzed as cherí + s, whence a new singular
cherry. The ultimate reason for the retraction of the morphological cut
in the fagur type would be the learner’s inability to determine where
the border is between SAME and DIFFERENT in the paradigm of
jagur.) However, the correct treatment of vowel syncope is still un-
known.
2.5. A number of polysyllabic stems display sometimes Final, some-
times Initial u-umlaut, more rarely also umlaut of the höföld type. All
such exceptional cases are listed in (19), q.v.
(19) List of stems displaying several kinds of u-umlaut
hérað ‘district’: nom./acc. pl. héruð (Initial umlaut) and héröð
(Final umlaut). My impression is that héruð is more common
than héröð.5
meðal ‘medicine, drug’: nom./acc. pl. meðul (Initial umlaut) and
meðöl (Final umlaut). My impression is that both are equally
used.
líkan ‘picture, model’: nom./acc. pl. líkön (Final umlaut) and
líkun (Initial umlaut). My impression is that líkön is the normal
form, whereas líkun is rare and found in the written language
only.
5 My impressions here and below are based on the data in the OHÍ and on my
work with informants.