The distribution of u-umlaut in North Germanic
Short note
North Germanic famously has many types of umlaut, or occurrences of vowel harmony where a preceding vowel is affected by a following vowel. In this post I’ll look at u-umlaut, whereby an original a (or its long form á) is rounded before an old u. The resulting vowel is typically written as ǫ (long form ǫ́), and survives in modern Icelandic as ö (but in modern Icelandic long ǫ́ merges with á and reflexes of both vowels are written á).
Mediæval North Germanic is usually split into three branches:
Western Scandinavian, spoken in Norway and her Atlantic colonies, most importantly Iceland
Eastern Scandinavian, spoken in Sweden and Denmark
Old Gutnish, spoken on Gotland
U-umlaut has a fairly similar distribution in the first two branches in words like hǫfuð ‘head’ (Icelandic höfuð, Swedish huvud; from Proto-North Germanic *haβuðą) and jǫrð ‘earth’ (Icelandic jörð, Swedish jord; from Proto-North Germanic *erþu).
The word jǫrð is also an example of u-breaking, where a Proto-North Germanic e has been “broken” into eǫ before the u. The exact development here is disputed. Some argue that the e was first broken into ea (as in words like hjalpa < *helpan) and then underwent u-umlaut to eǫ, while others think it broke directly to eǫ.
The latter view might find some support in Old Gutnish, which has hafuþ ‘head’ (< *haβuðą), but iorþ ‘earth’ (< *eǫrþu < *erþu).
In words like these Eastern and Western Scandinavian mostly have the same distribution of u-umlaut, but things get more interesting when we look at other words. As an example, let us take the Proto-North Germanic adjectival declension paradigm for the adjective *spakaʀ 'wise' in the singular:
masc. fem. neutr. nom. *spakaʀ *spaku *spakat acc. *spakanó *spakó *spakat dat. *spakumʀ *spakiʀé *spakú gen. *spakas *spakiʀóʀ *spakas
This develops into the following Old Norse (i.e. Western Scandinavian) paradigm:
spakr spǫk spakt spakan spaka spakt spǫkum spakri spǫku spaks spakrar spaks
Here the vowel ǫ occurs in the places we’d expect, namely the feminine nominative singular and neuter/masculine dative. It’s interesting that the old -u in the feminine nominative singular is still visible through the umlaut vowel—this vowel was lost before the Wiking Age, and so the umlaut must be fairly old.
Things are different with Old East Scandinavian, here the archaic Old Swedish paradigm (which is also representative of Old Gutnish):
spakr spak spakt spakan spaka spakt spakom spakre spako spaks spakrar spaks
One might assume that East Scandinavian never had the specifically Norse phonemic type of u-umlaut, where a and ǫ vary phonemically within the same word depending on the Proto-North Germanic distribution of *u.
Luckily we can see that this was indeed once the case also in Sweden and Denmark from the pronoun hann/hón ‘he/she’. This pronoun is fairly young and was originally declined just like an adjective, albeit with a defective paradigm, lacking neuter and plural forms. The Proto-North Germanic paradigm (all forms are reconstructed):
*hánaʀ *hánu *háninó *hánó *hánumʀ *hániʀé *hánas *hániʀóʀ
Due to the low stress of the pronoun the long vowel á was generally shortened to a. The cluster nʀ undergoes regular assimilation to nn (compare Proto-North Germanic stainaʀ > Old Norse stęinn), and the feminine dative and genitive forms have irregular i-umlaut. Finally there is some syncope, where unstressed vowels are dropped. These changes give the archaic Old Norse paradigm:
hann hǫ́n hann hana hǫ́num hęnni hans hęnnaʀ
Here the distribution of ǫ́ is the same as that of ǫ in the declension of spakr above; not very surprising.
The nasal consonant n has an curious property in that it sometimes causes fronting of a preceding long ǫ́ > ó. An example of this is the Old Norse word vǫ́n 'hope' (< Proto-North Germanic *wánu), which sometimes takes the form ón. Here the fronting would give the form **vón, but this is not phonologically permissible in North Germanic as v (pronounced /w/) is always lost before o and u (cf. ON orð ‘word’, ulfr ‘wolf’ < Proto-North Germanic *wordą, *wulfaʀ).
This change has an interesting effect on the Old Swedish paradigm of the same word, which means that the u-umlaut is uniquely preserved in a way that looks almost identical to the Old Norse:
hann hón hann hana hónom hęnne hans hęnnar
But Old Gutnish still has no traces of the u-umlaut!
hann hán hann hana hánum hęnni hans hęnnar