Studia Islandica - 01.06.1949, Blaðsíða 31

29
is it? Since Primitive Old Norse times Od has not occurred in
uncompounded words, as the verbs of the type breiOa, breiddi
show. This fact leads me to think that this change too is ana-
logical.
It is instructive to look at the changes taking place in com-
pound words of the type haf-gola, vit-laus from this point of view.
In most words of this type the (long) vowel of the first element
is shortened and the consonant lengthened, if the second element
begins with a consonant. If it begins with a vowel or an h +
vowel no change takes place in the first element. Our examples
thus become havvgola, vittlaus, while hafátt remains unchanged.
Obviously this is in analogy to conditions in uncompounded words,
where a vowel is mostly short before clusters of consonants and
where the first consonant of two is always longer comparatively
speaking than the following one. E. g. höfgi (hövvgi), gutla
(guttla).
But there are exceptions. As a rule p, t, k, and s do not shorten
the preceding vowel and they remain short themselves The rea-
son must be that in mono- and disyllabic words p, t, k, s are kept
strictly distinct from pp, tt, kk, ss; e. g. hop : hopp, hrat : hratt,
tak : takk, haus : hauss.
This is a clear case of preservative analogy. But, why did this
analogy not work in other cases? Because for a number of them
there was no such pattern to oppose the assimilatory forces of
the uncompounded words. The voiced spirants occur only short
after a vowel or intervocally: haf (hav), baO, sag; hafa (hava),
baOa, saga. They have no contrasting long sounds to maintain the
pattem. Of the liquids and nasals l is treated like the voiced
spirants, probably because original long ll has always developed
into dl and, as such, is too different to act as a contrasting ele-
ment and keep its opposition in check.
But in case of the other liquids and nasals our theory breaks
down, and even to some extent for l too. For we have, to be sure,
such contrasts as ala : Alla, dyra : kyrra, lama : skamma, mana :
rnanna and likewise in monosyllabic words, exactly as was the
case with the p:pp, t:tt, k:kk, s:ss above. In spite of that we
find compounds like al-vís, ber-bakt, fim-leikar, man-söngur as
a rule with shortened vowel and lengthened consonant. Why,
we do not know. Is it because the liquids and nasals, like the
spirants above, are voiced?
In the preceding lines we have dealt with systems or patterns
made up of two contrasting elements. We have seen that such
a pattern may have preservative influence on its members. Con-
versely, if the preservative influence of oppositions is lacking,
changes are more liable to happen.
French spellers must keep distinct their accent grave and ac-
cent aigu, but Icelandic writers, having only an accent aigu, can
indulge in any style until caught by a foreign printer.
One of the chief changes in Southern Icelandic is the weaken-
ing of intervocalic (or postvocalic) p, t, k to b, d, g. Initially we
have the contrasting elements p:b,t:d, k:g, but such is not the
case in inter- or postvocalic position. Here only one element oc-