Rit (Vísindafélag Íslendinga) - 01.06.1970, Blaðsíða 103

IOI
two points to recommend it: (i) In contradistinction to the
umlaut proper the other two changes—a-umlaut and the
t-umlaut of c > i—afíected only the short vowels, not the cor-
responding long ones; and (2) the earliest Old High German
sources, in which the i-umlaut (of a>é) is not yet regular,
show both these changes.2 In addition, as regards the change
e >1 in particular, this alternative is supported by the following
three facts: (a) this would be the only case of i-umlaut af-
fecting a front vowel, and the result unique in not being the
creation of a new phoneme, but, instead, a merger with a
preexistent vowel; (b) this change occurs in forms which are
otherwise without z-umlaut, e.g. ON pret. skilþe (of skilja ‘to
divide’) vs. talþe (of telja ‘to count’), OHG pret. filta (offillen
to flay’) vs. stalta (of stellen ‘to place’), or pret. rihta (of rihten
to set right’) as compared with maht ‘power’, plur. mahti
(without i-umlaut before ht as opposed, for instance, to anst
grace’, plur. ensti); and (c) the same change, e > i, occurs also
under other conditions, viz., before a nasal + a consonant, cf.,
e-g., ON binda ‘to bind’, OHG bintan, OS OE Goth. bindan<
*bend-, cf. Lat. of-fend-ix ‘band’, etc.; as will be shown below,
there is a cogent reason for not separating this change from
the other cases of e > i.
For these reasons the two changes—a-umlaut and the change
e > i—cannot properly be regarded as an integral part of um-
laut. In order to place these changes in their proper per-
spective, we may take the following two Germanic vowel dia-
grams as a starting-point:
2The earliest North Germanic evidence is a little less unequivocal; cf., e.g.,
horna (Gallehus), but erilan (Lindholm, Kragehul, etc.). The latter form, however,
being trisyllabic, is a special case; in its development it may have followed a
different pattern, which also gave the by-form irilaR (Veblungsnes, By). For
irilan, if it is a genuine form, is not necessarily a later equivalent of erilart since
both occur in sixth-century inscriptions (on the problems involved, see, fur-
ther, H. Benediktsson 1967:188—190).