How Was the Old Norse ʀ Pronounced?

Written by Dyami Millarson

A syllable is divided into three parts: an onset, nucleus, and coda. Simply put, the onset and coda are consonantal, the nucleus is vocalic. There was an r-sound in Old Norse which only occurred in the coda of word stems and otherwise could only occur in the onset in inflected forms of the definite type (Old Norse had an indefinite and definite distinction in inflected forms). This Old Norse r-sound, which could mostly be expected to occur in the syllable coda unless part of a definite declension, was the ʀ.

Let me share a few facts about the ʀ:

  1. Runic Norse distinguished between ʀ and r as the yr rune (ᛦ) was used in Runic Norse to represent the ʀ and the reið rune (ᚱ) was used to represent the normal r.
  2. The ʀ occurred in both Proto-Norse and Early Old Norse.
  3. The ʀ was mostly found in the suffixes of word inflections. These environments often necessitate a syllabic pronunciation.
  4. The ʀ developed into r in a later stage of the Old Norse language, and as word-final -r, it later developed into -ur in Icelandic and Faroese.
  5. The ʀ came from Proto-Germanic /z/, which came from Indogermanic /s/, which is a historical sound change known as Verner’s law.
  6. The ʀ corresponds to /s/ and /z/ in the Gothic language. The Proto-Germanic /z/ had been devoiced to /s/ in most environments in Gothic yet it remained /z/ in a few environments in Gothic.
  7. The ʀ produced a vowel change in some words known as ʀ-umlaut or ʀ-mutation. This vowel change made a vowel sound more ʀ-like by moving it closer to the place of articulation of the ʀ. As a result, the ʀ-mutation should make words with ʀ noticeably easier to pronounce with ʀ-mutated vowels than without. Examples of the ʀ-umlaut include kýʀ and and kęʀ, which stand for ʀ and kaʀ.
  8. I have seen the ʀ being variously interpreted as alveolar trill [r], retroflex approximant [ɻ], retroflex fricative [ʐ], alveolar approximant [ɹ], post-alveolar non-sibilant fricative [ɹ̠˔], alveolar retracted sibilant [z̠]. Personally, I have considered the alveolar fricative trill [r̝] as an option as well.
  9. Vesa and es, where the s-sound was likely pronounced /z/, changed to vera and er in Late Old Norse, which is reminiscent of the correspondence between Proto-Germanic /z/, Proto-Norse and Early Old Norse ʀ, and Late Old Norse r.

Being aware of fact 4, I am absolutely certain that the ʀ was pronounced at a late stage of Old Norse as a syllabic alveolar trill [r̩] in word-final positions which warranted a syllabic pronunciation and otherwise it would be pronounced as a non-syllabic alveolar trill [r]. The former corresponds to Sanskrit ऋ  [r̩]. Indogermanic is reconstructed to have had syllabic liquids and nasals as well. Just like Indogermanic, Old Norse had syllabic liquids and nasals. The word-final -r, which I believe to have been pronounced as a syllabic alveolar trill, later developed into -ur in Icelandic and Faroese as the final -r became non-syllabic. Fact 9 also demonstrates to me that /z/ could change to /r/ direcly without an intermediate stage. However, the /z/ > /r/ change as represented by vesa/es and vera/er does not involve a non-syllabic r-sound and so the environment is not entirely comparable to that of ʀ, which often occurs in places which necessitate a syllabic pronunciation (see fact 3).

At first sight, whilst observing that ʀ occurs in word-final positions necessitating a syllabic pronunciation, my phonetic intuition would be tempted into trying to make sense of ʀ as a syllabic alveolar trill first. However, this interpretation does not work since it cannot account for nominative plural -aʀ and other instances where a syllabic interpretation is clearly not appropriate. What such instances show is that ʀ contrasted with r in terms of sound value and that the ʀ/r was not a simple distinction of non-syllabic and syllabic r. Namely, when only taking the instances into account of where ʀ has to be interpreted as a syllabic consonant, it makes perfect sense to try interpreting ʀ as just the syllabic cointerpart of r. It would have been interesting if Old Norse rune masters made a distinction between a syllabic and non-syllabic r. In any case, such an interpretation of the facts makes only sense at first glance, because upon deeper inspection, it would become clear that ʀ is not just the syllabic counterpart of r, which is a vital point to prove in our journey in search of the phonetic value of ʀ.

We know now that the alveolar trill interpretation of ʀ does not work. So we are still left with interpretations of ʀ as some kind approximant, sibilant, or non-sibilant fricative. The r, which is ultimately descended from ʀ, is pronounced as an approximant in Faroese. However, both the Old Norse r and ʀ have merged in Faroese, and so it cannot be used as definite evidence that the Old Norse ʀ was pronounced as an approximant. Furthermore, given facts 4 and 5 which may be summarised as that ʀ merged with /r/ in Late Old Norse and ʀ originally comes from Proto-Germanic /z/, I am inclined to believe that the approximant interpretations are too much unlike the value of the sound it later merged with and too much unlike the sound it came from originally. One would expect a sound that is both /z/-like and /r/-like. Furthermore, fact 7 is an important consideration too: could approximants really have produced the ʀ-mutations we know of? Do the approximants meet the basic definition of ʀ-mutation: are the mutated vowels indeed closer to approximants in terms of pronunciation? If we were to interpret ʀ as some kind of approximant, would the ʀ-mutation really make the word noticeably easier to pronounce? These questions are relevant to ask in relation to fact 7.

Personally, I have never been that drawn to interpretations of ʀ as approximant. I do not find approximants to be /r/ and /z/-like enough. I am also not that attracted to interpreting ʀ as a sibilant. It would make sense in light of fact 9. However, as I explained earlier, ʀ involves a sound which occurs in both syllabic and non-syllabic environments, while the example of vesa/es > vera/er only involves a non-syllabic /z/ and /r/. Furthermore, ʀ apparently contrasted with /r/ as well as /z/ as an allophone of /s/ (namely, I am not aware of vesa/es being spelled as veʀa/ in Runic Norse, which should have been tempting if ʀ were a sibilant).

I have always noticed that I may pronounce the final -r of Dutch sterker in various ways, namely one where the r is a syllabic alveolar trill and one where the r is a syllabic fricative of some kind. I always compared this allophonic fricative pronunciation of the alveolar trill /r/ to the sibilant /z/ in my mind since it sounded similar to the sibilant /z/ but it was not quite a /z/. It actually took me a while to realise that the syllabic fricative I was hearing was a post-alveolar non-sibilant fricative, and this allophone occurs in non-syllabic environments in my Dutch as well. This allophone specifically occurs in my Dutch in word-final r‘s. I have always intuitively felt that my pronunciation of the final -r in sterker might offer a clue for how the Old Norse stęrkr/stęrkʀ was pronounced.

Apparently, my pronunciation of the final -r of sterker as a syllabic alveolar trill might match that of Late Old Norse stęrkr and my pronunciation of the final -r of sterker as a syllabic post-alveolar non-sibilant fricative might match that of Early Old Norse stęrkʀ. What attracts me to the interpretation of ʀ as a post-alveolar non-sibilant fricative is not just that it seems familiar to me, but also that the sound seems sufficiently /r/-like as well as /z/-like to me. I had initially analysed the sound as a kind of /z/, since it sounded like that to me, but later came to identify it with the symbol [ɹ̠˔] in the International Phonetic Alphabet (IPA) upon closer inspection of the exact sound value.

In other words, I believe [ɹ̠˔] fits the phonetic picture the best, as we needed to find a sound value for ʀ that would be intermediate between /z/ and /r/; the requirements of being /z/-like and /r/-like were predicted by facts 4 and 5, which touched on sound history. Furthermore, the /z/-like quality of [ɹ̠˔] is definitely needed in order to yield the expected results of the ʀ-umlaut as explained in fact 7 (also see this study). I should mention that whilst preparing this article, I also briefly read an amazing presentation which delved into the Lautverhältnisse or sound relationships of the ʀ (or as the presentation author described it: “phonological evidence from participating in phonological activity”), Norse loanwords in Sámi and Finnic, and the analysis of (mis)spellings, and which mentioned the [ɹ̠˔] interpretation of ʀ. I genuinely felt amazed with that presentation because the phonological arguments the author made followed a logical structure that I particularly like and feel comfortable with and I know that I would employ a similar logical structure if I were to write a presentation like that on the issue of ʀ; so I really commend the author for his work. The presentation also made a good point that the period between Proto-Germanic /z/ and Late Old Norse /r/, which comprises the intermediate period of ʀ, lasted over a millennium. Having read said presentation, it further confirmed to me in any case that my attraction to the post-alveolar non-sibilant fricative [ɹ̠˔] as an interpretation of ʀ makes phonetic sense.


Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s