SIL in Eurasia

Kabardian – A Language Without Vowels?

Approximate distribution of the branches of the Northwest Caucasian languages:

Red: Circassian, Orange: Abazgi, Yellow: Ubykh (extinct)

By Marius Baak*

Picture this: you are studying linguistics at university and need to collect language data to write your thesis. Thus, you pack your things and travel to a remote place to study the language of a people nobody has ever contacted – nothing about their language is known. Well, that’s exciting! So many intriguing things are waiting to be discovered in this language. However, your mind is not a blank page when it comes to languages. You certainly have an idea of what is possible in a language, and what is not. For example, when you point at your nose, then at your arm, then at a stick, you will expect to hear different sound waves from the people you visit, i.e. there will be distinct words. And when you manipulate an item, e.g. when you throw the stick, you expect the description of this to be a longer sound wave than when you just point at the stick. But aren’t there many more such “universals” in spoken languages? Wouldn’t you also, for example, expect that such a word as wknrklsp cannot exist, since not only consonants are needed but also vowels?

For a few years, a language at the foot of the mighty Elbrus mountain in the Caucasus seemed to disprove the assumption that a language needs vowels. This language is Kabardian or East-Circassian. It is spoken by more than 500,000 people in the North Caucasus in Russia and by many more in other countries. Like the other Northwest Caucasian languages, Kabardian is famous for having a large number of consonants – it has twice as many as English – and for verbs that need to be represented by whole sentences in English. Consider this Kabardian verb: Уакъыхуезгъэшыжыфатэкъыми, which, in a desperate attempt to represent these difficult sounds, could be transliterated as: waqɨxwezʁeshɨzhɨfateqɨmi. In English, you need a full thirteen words to describe what Kabardian communicates with one single verb: Since I could not make him bring you back here then for them.

But aside from this unusual grammatical structure (for speakers of Indo-European languages, that is), you see plenty of vowels in this example, don’t you? So why did the Dutch linguist A. H. Kuipers describe Kabardian as a vowelless language in his fascinating 1960 analysis “Phoneme and Morpheme in Kabardian”? The answer is complicated, but will be presented in a simplified way below.

 

Not everything we utter carries meaning - The basic idea-r-is this…

First of all, we have to be precise: Kuipers did not argue that you cannot hear vowels in Kabardian. What he did say was that vowels play no role in the interpretation of Kabardian, i.e. they do not affect the meaning of words. It is true for most languages that we utter things that are not important for what we try to convey. For instance, some people, when they say ‘the idea is’ actually pronounce this phrase as ‘the idea-r-is’. They slip an r-sound between the two words to connect a word that ends with a vowel with a word that begins with a vowel. However, this r-sound carries no meaning at all – and no English speaker would expect it to carry meaning. 

Thus, Kuipers was well aware that you do hear vowels in Kabardian, even very familiar vowels such as i, e, a, o, u… However, according to him these vowels are not important in distinguishing different meanings of a word. How can that be? After all, you have Kabardian words where the only difference seems to be a different vowel, i.e. shɨ ‘horse’ and shə ‘milk’!

Types of vowels - Tennis vowels and other extremes

First, let us look at how you can distinguish different vowels and then how vowels and consonants are less different than you might think. So, how is an [i] (as in teeth) different from an [a] (as in after)? With [a] your mouth is wide open while with [i] it is more closed. But when you look at [i] and [u] (as in food) they do not differ in terms of their openness, both are so-called close vowels, you don’t open your mouth much. But for [u] you round your lips and your tongue is much further back than for [i] which is more front and unrounded. Between those ‘extreme’ vowels (meaning front or back here) are the central vowels for which you do not move your tongue forward or backward, you simply open your mouth and lift your tongue to varying degrees. You hear those a lot while watching a tennis match: we do not bother moving our tongue too much when grunting or moaning. Thus openness, backness and rounding are the three features which help to distinguish vowels from each other. We can classify the main Kabardian vowels according to these features (the two back vowels o and u are rounded which is not marked here):

Kabardian vowels and their position in the mouth

Thus, we see seven vowels and for all of them we have examples like those above where only the vowel distinguishes the meaning. But things are not as simple as they seem – vowels and at least some consonants are not as distinct from each other as we may think. 

From 7 vowels to 3 – The vowel countdown starts

Have you ever noticed that the consonant [w] (as in wow or flow) is almost the same sound as the vowel [u] (as in food)? When you pronounce the word woo you will feel that your tongue barely moves. In Kabardian, every instance of an [u]-vowel occurs after – or before – a [w]-sound or a similar sound. The same is true for the vowel [o] (as in board). The following examples always show the actual pronunciation rather than the spelling:

ku:wu   ‘deep‘

pso:      ‘alive‘   (written: psəw; from psə ‘soul‘ + adverbial ending -w)

For now, just note how the vowels [u] and [o] not only sound similar to the [w]-consonant but are restricted to its immediate context in Kabardian. In a very similar way, the consonant [j] (as in yes) is almost the same sound as the vowel [i] (as in feet). When you say yeast you barely move your tongue from [j] to [i]. And guess what? Every instance of a Kabardian [i]- and the similar [e]-vowel (as in Australian English bed) comes after – or before – such a consonant:

jin        ‘big‘

dej       ‘nut tree’

So, what is happening here? At first, it sounds contradictory: The vowels [i] and [u] come from the same central vowel [ɨ], just as [e] and [o] are essentially variants of the central vowel [ə]. Let this sink in a bit: In Kabardian, [i] and [u] are the same vowel! How can this be? Think back to the beginning: Just because you hear a difference that is not to say that the difference is meaningful – the difference might have other causes. And indeed, in Kabardian, those causes are physiological: remember that for the central vowels [ɨ] or [ə] you do not move your tongue forward or backward; you just lift it and open your mouth. When such a vowel meets the consonant [j] for which the tongue moves quite a long way forward, this pulls the vowel along: you get [i] or [e]. Likewise, when such a central vowel occurs beside [w] which is quite far back, this pulls the vowel backward, resulting in [u] and [o]: 

How [j] and [w] change vowels

The close vowel [ɨ] becomes [i] around the consonant [j], and [u] around the consonant [w]. The same happens with the mid vowel [ə] that changes to [e] and [o] according to its environment. We can thus rewrite the above examples to show the pattern here (written as underlying form --> what you hear):

 

                  around [j]                                      around [w]

close vowel [ɨ]:              jɨn --> jin           ‘big‘                  kɨwwɨ -->  ku:wu           ‘deep‘

mid vowel [ə]:               dəj --> dej         ‘nut tree’          psəw --> pso:                ‘alive‘

In other words: We hear the four vowels i, e, o, and u, but they are so-called ‘contextual variants’ of the two vowels ɨ and ə. The best example is the word psəw that you hear as pso: (with a long o): as soon as the suffix -w is added to the word psə, the ə-vowel changes to o.

From initially seven vowels we now have arrived at just three (ɨ, ə, a) that are only distinguished by height (and maybe length, see below). This analysis is widely accepted, with some linguists proposing just two vowels for Kabardian. But that still leaves us with two or three vowels! Let us continue from here.

From 3 vowels to 2 – And there goes number five

Until today linguists argue about the vowel [a]: Some say it is not different in height from the ə-vowel, but is just its long variant. Kuipers established his own theory: Since the four vowels i, e, o, and u are actually central vowels that are altered by consonants, then the same might be true for [a]. Again, make the test: say the word hard and notice how your tongue does not move at all from h to a: for h you only narrow your vocal cords a bit, that is the only difference. And again he would say that a h before or after ɨ or ə will change the vowel, basically pulling it down to become an open vowel a:

 underlying form --> what you hear

 həzhə --> azhə2   ‘male goat’

The problem here is while [j] and [w] actually occur in Kabardian, [h] does not. Kuipers needs to argue that what was once a consonant [h] is gone in modern Kabardian, and what is left is only the effect that consonant had on the neighbouring vowel, in pulling a high vowel down to a. In other words, to maintain his theory, he needs to claim that [h] totally assimilated (i.e. disappeared) in every single context without a trace, or at least only a slight trace (sometimes, there is a slight glottal friction on the a-vowel). However, he can use this theory to show why a is the only vowel that stands at the beginning of a word: there used to be a consonant that assimilated to the vowel over time.

From 2 vowels to 1 – Last vowel standing

That leaves us with only two vowels, ɨ and ə. And again, things are not as simple as they may seem. Make a test: If you were to read the following made-up sequence of English consonants, where would you insert vowels? Let us take only the English ə-vowel (as in assume or petition) because it is similar to the Kabardian one:

stzk 

We can have st in English, but no tz, certainly not stz. Also, there is no zk-sequence in English. So we can say either sətəzək or stəzək. It is not possible in English to pronounce stzək or stəzk. Languages will differ in where exactly such vowels can be inserted, but human physiology makes it necessary to insert at least some. Kuipers argued that you can predict every occurrence of the ɨ-vowel in Kabardian simply by taking the underlying consonant-sequence of a word and putting ɨ where it would otherwise result in impossible consonant combinations (let us leave the ə-vowel in the example, since we haven’t talked about it yet):

sʁrtəqm --> sɨʁɨrtəqɨm   ‘I was not weeping’

In this example, and in many other words, there is only one way to insert ɨ-vowels. The details of this argument are very complex and highly debated – but generally, it is a valid observation that holds also for the other Northwest Caucasian languages. 

From 1 vowel to 0 – At the bottom of the barrel

This leaves us with just one vowel, ə. But in the light of pairs like shɨ ‘horse’ and shə ‘milk’ it should be impossible to dismiss this vowel, shouldn’t it? Sure, the argument goes that shɨ has the ɨ-vowel just because you cannot pronounce a word that consists of only a consonant in Kabardian: practically, the root is just sh according to Kuipers. But how do you explain the ə in shə?

Well, ə is different from ɨ only in terms of openness: for ə you open your mouth more. In order to arrive at a zero-vowel analysis, Kuipers takes the openness feature of the vowel and makes it a feature of a consonant (marked with a). Thus, there is a normal sh-consonant, and there is sha, a consonant after which you open your mouth a bit more than usual. Both consonants cannot stand for themselves, so a ɨ-vowel is added (this would also work in longer words where certain consonant combinations aren’t allowed). The vowel then either changes or not depending on the preceding consonant:

sh + ɨ --> shɨ ‘horse’

sha + ɨ --> shə ‘milk’

Thus, Kuipers would argue that from a linguistic point of view you only have chains of consonants in Kabardian. To be able to pronounce those consonant chains, you need to insert ɨ-vowels every now and then. These are extremely sensitive to their surroundings and can change to all kinds of other vowels in actual speech.

Wrapping up: There and back again

Congratulations: you have followed Kuipers on his journey from seven, to three, to two, to one, to zero vowels. I trust it was an interesting journey! Later detailed critiques by linguists like Halle (1970) clearly showed how the last two steps, the reduction of ɨ and ə, as well as the h-hypothesis, are taken too far. If Kuiper’s theory had been true, Kabardian would have been the only language where the openness feature is consonantal rather than part of the vowel system. Also, it is much simpler to assume that there are occurrences where ɨ is part of the root and cannot be predicted from context.

Yes, the extreme theory of Kuipers needs to be rejected. However, the theory does show beautifully how Kabardian is a consonant-based language where the few vowels readily adapt to their environment and take a whole different flavour. Thus we see not only how the boundaries between vowels and consonants are blurred in this example, but also how our preconceived ideas about language can be challenged once we open ourselves to the uniqueness of the many languages we still have on our planet. While we still have them.

*pseudonym

1Note: It is well known that there can be words without vowels, but not entire languages; the Salishan languages for example, which are spoken in North America, have many words without vowels.

2The zh is pronounced like the si in vision.

Resources

Halle, M. (1970) 'Is Kabardian a Vowel-Less Language?', Foundations of Language, 6(1), pp. 95–103.

Kuipers, A.H. (1960) Phoneme and Morpheme in Kabardian: (Eastern Adyghe). Mouton.

Photos

Map of Northwest Caucasian Languages 

Photo by Gaga.vaa, CC BY-SA 3.0 

Kabardians in Traditional Costume
Photo by Itamar Grinberg for the Israeli Ministry of Tourism, CC BY-SA 2.0

Elbrus Mountain
Photo by IlyaKuban from Pixabay