What is ʀ and how do you say it?

Notes on symbols

If you’re familiar with IPA, you can skip this part, but I use a lot of funky brackets and things, and you’ll want to know what they are going in. This isn’t necessary, but it’ll make it easier to figure out what I’m trying to say later.

  • [skwɛəɹ bɹækɪts]  contain phonetic transcription – that is, the actual sounds of a word or letter. I recommend opening up this page in another tab or window and keeping it handy while reading this page, so that you can hear audio for the symbols I use as they come up: http://en.wikipedia.org/wiki/IPA_pulmonic_consonant_chart_with_audio. I will also make each symbol a link to its own Wikipedia page as long as I remember to.
  • /ˈbækˌslæʃəz/ indicate phonemic transcription. This is not exactly the same as the sounds themselves. It’s kind of complicated – phonemes are the smallest contrastive units that your brain stores for piecing words together. For example in English when a vowel comes before a nasal consonant like /n/ or /m/ the vowel itself comes out nasal: [ã]. You probably don’t even hear it because it doesn’t make a difference to English – both [a] and [ã] are different realizations of one phoneme /a/. In French, nasality does make a difference, so there must be two separate phonemes /a/ and /ã/. As you will see, a phoneme described as /z/ does not necessarily have to sound like [z].

1. What is /ʀ/?

If you read enough about Old Norse or the languages it came from, especially if you’re interested in the runes, you are bound to encounter /ʀ/. You probably see it mostly at the ends of words, though not exclusively. In the runic alphabet it was represented by *alǥiʀ which tended to flip over to z2 until that became permanent (in the younger fuþark it is known as ýʀ).

Some ways to write the name of the /n/-rune ᚾ in Proto-Norse:
– *nauđiʀ
– *nauđiR
– *nauđiz

All of these are correct, and the only difference between the first and second is really just what kinds of special characters are available when you’re typing or encoding something. Even if you’re writing sentences in viking-age Norse there’s no danger of confusing ⟨R⟩ for a capital version of /r/, since /ʀ/ can never be the first letter in a word.

It persisted through early Old West and East Norse, and you may recognize it from runic inscriptions from the viking age.

Rök runestone, Ög 136, Östergötland, Sweden (the little half-line on the left is the “short twig” variant of z2 ýʀ)

(Theodoric the Great, king of the Ostrogoths, 5th-6th century)

As far as most linguists can tell, at one point a long time ago, the Germanic languages had a /z/ sound (details on how it arose from /s/ later). Other than Gothic which was attested very early and still had a /z/, in all of the other Germanic languages this /z/ somehow merged with /r/. This is something called rhotacism, and believe it or not, it’s not uncommon – the same thing happened in Latin.

Since most linguists figure it didn’t just flip from /z/ to /r/ overnight, many have assumed there was an intermediary period where it got closer to /r/, and the letter used to show that that intermediary sound, whatever it was, is ʀ (or R).

Usually, though not always, people use /z/ for Proto-Germanic, and /ʀ/ for Proto-Norse. That’s not a bad rule of thumb (though plenty of people still use /z/ for Proto-Norse.

Let’s go back to our word *naudiz, the name of the n-rune. In the younger fuþark it’s called nauðr. It’s development looked something like this:

  1. Proto-Germanic: nauđiz
  2. Proto-Norse: nauđiʀ
  3. Viking-age Norse: nauðʀ
  4. Old Icelandic: nauðr

2. How do you say it?

The answer is uncertain, but the most widely-accepted belief is that it moved from [z], like a normal English z, to something like [ʒ] (the s-sound in vision) or the voiced retroflex sibilant, marked in the IPA with [ʐ].

Listen to a recording of what the following verse might have sounded like when it was written in runes on the Rök stone, mentioned above: http://en.wikipedia.org/wiki/File:Theoderic_noise_removed.ogg

Reð ÞioðrikR
hinn þurmoði,
stilliR flutna,
strandu HraiðmaraR.
SitiR nu garuR
a guta sinum,
skialdi umb fatlaðR,
skati Mæringa.
Theodric the bold,
king of sea-warriors,
ruled over
Reid-sea shores.
Now he sits armed
on his Gothic horse,
shield strapped,
prince of Mærings.

Normalized text and translation by Samnordisk runtextdatabas.

From there it may have become more r-like but of course the next step would depend on what the “regular” /r/, written with the r1 *raiđō/reið rune, sounded like. Most people have believed that it sounded like [r], an alveolar trill, known better as a “rolled r.”

Based on that, some theories regarding /ʀ/ have included that, shortly before the merger, it was a uvular [ʁ] like in many dialects of French or German, or (surprise!) [ɹ], an alveolar approximant – a normal American English /r/.

However, a recent dissertation by Dr. Robert Kenneth Painter has demonstrated that the slow progression toward rhotacism is in no way necessary – it easily could have happened without much of an intermediary. He points out that we don’t actually know for sure the quality of the original /z/, which has been assumed to be [z] because it was originally derived by voicing an [s]. However, pre-Proto-Germanic – before /z/ arose – had only one /s/-sound even related to [s], which gives it a lot of freedom to wobble between sounds without being confused for something else (see: Finnish, where speakers sometimes pronounce their /s/ in a way an English-speaker would identify it as sh /ʃ/, because they have no /z/ or /ʃ/ or /ʒ/ for it to be mistaken for… Icelanders do this sometimes too).

So the original /z/ may already have sounded something like [ʐ]. It’s also possible it did come into being as a [z], but made a small change, staying consistent for most of its existence.

Using a fascinating experiment, Dr. Painter went on to prove that some [ʒ]-, [ʐ]-, and even [z]-like sounds can easily be confused for and switched with r-sounds in normal everyday speech, especially when following certain other sounds in the same word. I won’t go into all of the details here, but the link above will bring you to a page where you can download the thesis (warning: steep learning curve).

Update 11 November 2013: I’ve just read, in Frå urnordisk til norrønt språk by Odd Einar Haugen, that Harald Bjorvand, a Norwegian historical linguist, pointed out that the Proto-Norse word *diuʀą (“animal”; cognate to deer; ON dýr) was adopted into the Saami languages as divra. I’m afraid I don’t know much about Saami languages. Looking through an article, On Germanic-Saami contacts and Saami prehistory by Ante Aikio, which notes a variety of North Germanic loanwords into older Saami languages at various points in time, it seems most instances of Germanic /ʀ/ became Saami /s/ (Proto-Norse *haitaʀ ‘hot’> Northwest Saami *hājttēs > Northern Saami háittis ‘very hot (of a stove)’).

However I did find two other interesting examples mentioned…

  • Proto-Norse(?) *hrauʀōʀ ‘groin’ (plural) > Pre-Saami *rawša ‘udder’
  • Proto-Norse(?) *nāƀiʀōʀ ‘birch bark’ > Pre-Saami *napra ‘birch bark’

The š, I would have to believe, represents /ʃ/, which is the English ‘sh’ sound. That’s what it usually represents anyway.

The reason I question-marked the ‘Proto-Norse’ is that the dating of the first seems uncertain, and because Aikio protests that the second was actually *ƀirōʀ, with a “regular” /r/ instead of a /ʀ/, on the grounds that Saami wouldn’t have taken it as an /r/. But if Harald is right about divra, I wonder if that may be evidence that /ʀ/was analyzed by Saami speakers in different ways depending on whether or not it came at the end of a word.

Harald Bjorvand considers the sound of /ʀ/ to be the raised alveolar non-sonorant trill [], which is the ř-sound in the Czech name Dvořák, like the composer.

3. Verner’s Law and evidence for /ʀ/

The complete story of this letter takes us all the way back to Proto-Indo-European. Like Norse, PIE words all had endings that indicated their grammatical position in a sentence. Grammatical endings are NOT the only place that /ʀ/ appeared, but it makes for an easy demonstration (many strong verbs – those that show past tense by changing the vowel, like run/ran, often show Verner’s Law affects as well, for example). Take for example the word *deywós, the word for a god.

*Deywós is the nominative form of the word, but if you were praying to a god and wanted to address him or her directly, you would use the vocative: *Deywé. Something which belonged to the god was described with the genitive *deywósyo (“*deywós‘s” or “of *deywós“).

This ending survived in many of the descendant languages:
Proto-Celtic: *dēwos
Latin: deus (Old Latin deiuos)
Balto-Slavic: *deiwas
Proto-Germanic *tīwaz (both a generic word for “god” and the god Týr/Tīu)

Something happened during the transition from PIE to Proto-Germanic which is called Verner’s Law. This changed the sounds in a word based on the syllable which was stressed. In Proto-Germanic, every word had its stress on the first syllable (unless it had a prefix) but the stress could move around in PIE like in Greek and Latin (or like modern English — say “I’m working on a project,” and then “to project and image”). Since Verner’s Law didn’t hit every word in the same position, that means that it must have happened before Proto-Germanic developed its fixed stress pattern – the stress still had to be moving around like in Greek and Latin for Verner’s Law to have the results it did.

Verner’s Law states that after an unstressed syllable, voiceless fricatives become voiced fricatives. Voiceless fricatives are consonant sounds that partially, but not fully, obstruct airflow, and do not make use of the vocal cords – the important ones for Proto-Germanic being f, þ, s, and h [x]. After Verner’s law these all became, respectively, ƀ, đ, z, and ǥ.

“What’s with those lines,” you ask? Normally when transcribing Proto-Germanic linguists just use regular b, d, and g, but some mark them like I did to remind the reader that they act a little funny. The /b/ usually sounded sort of like an English /v/, but with both lips instead of the lips and teeth [β]. The /d/ sounded like the /th/ in then, and is the ancestor of Icelandic /ð/, which is its sound [ð]. The /g/ was kind of like a Scottish or German /ch/ but with the vocal chords vibrating [ɣ].

The /b/ and /d/ could sound like our normal equivalents in some positions, like the beginnings of words, and that might be true of the /g/ as well but some linguists think it was always a fricative [ɣ].

Before Verner’s Law, *tīwaz was probably *tīwas. It had a stress on the first syllable (say “TEE-wass”). That means the *-was part, being unstressed, triggered Verner’s law, and the unvoiced s became a z. The genitive form however, had a stress on the *-was part (say “tee-WASS”). Because Verner’s Law didn’t have anything to operate on in that form, it stayed the same (since it only effects syllables coming after the unstressed one… if the last syllable in a word was stressed, there was nothing for it to do). So by the time of Proto-Germanic, when *Tīwaz owned something, then it was *Tīwas (“Týr’s,” like in *Tīwas dagaz, “Tuesday”).

Throughout the period during which runic writing was used, the   *algiʀ rune (then z2 ýʀ) consistently marks this phoneme up until a few hundred years before Norse languages started being written using the Latin alphabet. Around the year 900 AD, rune-writers start to look a little confused; they start using z2 where r1 is expected, and r1 where z2 is expected.

Tryggevælde stone, D 230, Sjælland, Denmark (about 900 AD)

Tryggevelde_raknhiltr sustiʀ

Ragnhildr, systiʀ
“Ragnhildr, sister.”

Etymologically, Ragnhildr’s name should end with a z2/ʀ/ (hildr from PG *hilđiz), and systir should end with r1 /r/ (from PG *swestēr). By this point we can basically say that the sounds have merged, at least in the right environments.

4. Bonus Round

By the way, have you ever wondered why in English we say “I was” but “we were?” Why the alternation between /s/ and……… OH!

Verner’s Law affected all of the Germanic languages, including English. English lost all of it’s grammatical endings, but like I said, Verner’s Law affects ALL voiceless fricatives following unstressed syllables.

In Proto-Germanic to say “I was” you would say *ek was, and to say “we were” you would say *wīz wezum (which could also be written *wīʀ weʀum). The r/s alternation between are and is comes from the same source.

In Old Icelandic they leveled that alternation out (which means we English speakers have something more archaic in our language than the Icelanders!). “To be” is að vera, and it’s forms are:

 að vera Present Past
Singular ek er
þú ert
hann er
ek var
þú vart (later varst)
hann var
Plural vit erum
þit eruð
þeir eru
vit várum/vórum
þit váruð/vóruð
þeir váru/vóru

But in early texts they still used að vesa with a slightly more archaic conjugation:

 að vesa Present Past
Singular ek em
þú ert
hann es
ek vas
þú vast
hann vas
Plural vit erum
þit eruð
þeir eru
vit vǫ́rum
þit vǫ́ruð
þeir vǫ́ru

5. Sources


Leave a Reply

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

WordPress.com Logo

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

Google+ photo

You are commenting using your Google+ 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