Western Caucasus tribal composition, 1774–1780
One of the World’s Most Phonologically Complex Language Families
Five languages. Eighty-four consonants. Two vowels. And a civilisation’s worth of human expression.
The Northwest Caucasian languages — known equally as Abkhaz–Adyghe, West Caucasian, or, in older literature, Pontic — form one of the primary language families of the world. They are indigenous to the north-western flank of the Caucasus range and the north-eastern shore of the Black Sea, a territory their speakers held compactly until 1864. The family comprises five languages in three branches, and despite more than a century and a half of sustained comparative work, no genetic relationship to any other family on Earth has been demonstrated to general satisfaction. Proposals linking Northwest Caucasian to Northeast Caucasian, to the ancient Anatolian language Hattic, or to a wider Nostratic grouping have been advanced repeatedly and remain, in each case, contested. For practical purposes the family stands alone.
The Northwest Caucasian languages stand as one of the great enigmas of comparative linguistics — an isolate family of extraordinary phonological elaboration with no known relatives.
What sets the family apart is not merely its isolation but its shape. Northwest Caucasian languages carry some of the largest consonant inventories ever recorded — from roughly forty-eight in standard Kabardian to more than eighty in Ubykh — set against vowel systems of two or three phonemes. They are ergative, verb-final, and radically polysynthetic: a single verb may host a dozen prefixal slots encoding subject, object, indirect object, location, direction, negation, modality and evidentiality, so that one word does the work of an English sentence. These are not exotic curiosities but the ordinary daily speech of communities from Maykop to Amman.
The Abkhaz–Abaza branch comprises Abkhaz (аԥсуа бызшәа), the state language of the de facto independent Republic of Abkhazia, with roughly 120,000 speakers, and Abaza (абаза бызшва), co-official in the Russian republic of Karachay-Cherkessia, with some 51,000 speakers and a UNESCO classification of Definitely Endangered. The two are closely related enough that the boundary between them is partly a political and literary one: Abaza's Ashkharywa dialect is linguistically closer to Abkhaz than it is to Tapanta, the other Abaza dialect and the basis of the Abaza literary standard.
The Circassian branch divides into Adyghe or West Circassian (адыгабзэ), official in the Republic of Adygea, with roughly 590,000 speakers worldwide, and Kabardian or East Circassian (къэбэрдеибзэ), official in Kabardino-Balkaria and Karachay-Cherkessia, with roughly 1,685,000 — the largest Northwest Caucasian language by a wide margin. Their speakers overwhelmingly regard them as two forms of a single Circassian language, notwithstanding the absence of mutual intelligibility between the standard varieties.
Ubykh (twaxəbza) constitutes a branch of its own, and it is extinct. Tevfik Esenç, its last fully competent speaker, died in Turkey in October 1992. The language he carried had the largest consonant inventory ever documented for a language without click phonemes.
Hover or tap any node for speaker numbers, official status, script and ISO code. Ubykh is placed centrally rather than at the margin: in a number of phonological and morphological features it occupies an intermediate position between the Abkhaz–Abaza and Circassian branches, and the question of which it stands nearer to has occupied comparativists since Dumézil.
Dialect divisions in this family are conventional simplifications. Sadz and Shapsugh are treated by some scholars as languages in their own right; Guma is strictly a sub-dialect rather than a dialect; Besleney is classified here with Kabardian following recent literature, though it has long been described as transitional.
Adyghe is the western of the two Circassian standards, and the one whose consonant system has stayed closest to the Circassian starting point. Where Kabardian has simplified, Adyghe has held: it keeps a full series of retroflex sibilants and their labialized counterparts, and its southern dialects — Bzhedugh and Shapsugh — preserve a four-way stop contrast of plain, aspirated, voiced and ejective reconstructed for Proto-Circassian and lost almost everywhere else in the family. A Bzhedugh speaker distinguishes tʰəʁe ‘given’ from təʁe ‘sun’ by aspiration alone — and makes the same contrast in fricatives, where šʼʰe ‘milk’ opposes šʼe ‘sell’.
The literary standard rests on Temirgoy, spoken in Adygea, but the demographic centre of gravity lies elsewhere. Most Adyghe speakers alive today live outside the Caucasus, and among them the dominant variety is Abzakh — a minority voice in the homeland and the majority voice in Turkey.
| Dialect | Region | Notes |
|---|---|---|
| Temirgoy | Adygea | Basis of the literary standard; also called Chemgwi |
| Abzakh | Turkey, Jordan, Syria | The largest diaspora dialect; also written Abadzekh |
| Shapsugh | Black Sea coast; Turkey | Four-way stop contrast; occasionally classed as a separate language |
| Bzhedugh | Adygea | Retains the plain/aspirated contrast in fricatives as well as stops |
| Hakuchi | Turkey | A Shapsugh variety spoken in villages of Ubykh descent |
| Natukhay | Turkey; a few villages in Russia | Formerly of the north-western coast; now severely reduced |
Kabardian is the demographic giant of the family and, simultaneously, its most streamlined member. Its inventory of roughly forty-eight consonants is the smallest in Northwest Caucasian — which is to say it is still more than twice the size of Spanish. The simplification is systematic rather than random: where Adyghe has a labialized alveolar sibilant, Kabardian regularly shows a labiodental, so that a speaker of one can often reconstruct the other on the fly.
‘Fire’ and ‘wife’. The Adyghe labialized sibilant becomes a Kabardian labiodental — the clearest single diagnostic separating the two standards.
| Dialect | Region | Notes |
|---|---|---|
| Baksan | Kabardino-Balkaria | Basis of the literary standard; often called Greater Kabardian |
| Terek | Eastern Kabarda | Closely allied to the standard |
| Mozdok | North Ossetia | Spoken by a historically Christian Kabardian community |
| Kuban | Karachay-Cherkessia | Western varieties; known locally as Cherkess |
| Besleney | Karachay-Cherkessia; Turkey | Transitional between Kabardian and Adyghe |
| Malka | Northern Kabardino-Balkaria | Distinguished chiefly by vowel realisation |
Abkhaz holds the family’s starkest ratio. Against fifty-eight to sixty-seven consonants — the conservative Bzyp dialect sits at the upper end — it fields exactly two vowel phonemes: the low /a/ and the mid-high /ə/. That is not a shortage of vowel sounds. Abkhaz speech is full of them, but they are predictable colourings produced by neighbouring consonants rather than contrasts a learner must memorise. A labialized consonant rounds the vowel beside it; a palatalized one fronts it. The vowel does what it is told.
Abkhaz also has the family’s longest scholarly history. Baron Pyotr Uslar produced the first grammar in 1862 — lithographed then, printed properly only in 1887 — working, like other early commentators, on Bzyp. Ivan Bartolomei directed the first Abkhaz primer on Uslar’s alphabet a few years later. Every Abkhaz Cyrillic system since has been, in some measure, a revision of Uslar’s.
Until the mid-nineteenth century Abkhaz was spoken in a chain of varieties running south-east to north-west along the coast: Abzhywa in the east, Tsabal in the upper Kodor, Guma between the Kodor and present-day New Athos, Bzyp from New Athos to the Bzyp river, Sadz from Gagra to the Matsesta, and Ahchypsy in the upper Mdzymta. The mukhajirstvo — the forced migration to the Ottoman Empire — removed most of the speakers of most of them. Two survive in Abkhazia today: Abzhywa in the east and Bzyp in the west.
The exception is worth dwelling on. Guma, a sub-dialect rather than a dialect in Chirikba’s classification, occupied the middle of the chain and behaved accordingly, retaining features of both Abzhywa and Bzyp — a genuinely central idiom. It also happens to be the first form of Abkhaz anyone wrote down. The French traveller Jacques-François Gamba compiled a small Abkhaz–French dictionary from Guma speech at Sukhum in 1822; five years later Lieutenant Vladimir Romanov of the Russian fleet recorded some two thousand Guma words and phrases in the same city. Guma underlies Bartolomei’s Abkhaz ABC book and an early partial translation of the Bible — the first printed Abkhaz there was.
Then its speakers were expelled, and the gap they left in the written language was filled by Abzhywa, which became the norm for everything published afterwards. Guma survives in three villages in Abkhazia — Bagmaran, Merkheul, and partly Eshera — where the oldest generation still uses it, and in about ten villages in Turkey. Chirikba, who has recorded the sub-dialect in the diaspora, plans to publish the texts.
| Dialect | Region | Notes |
|---|---|---|
| Abzhywa | Eastern Abkhazia | Basis of the literary standard |
| Guma | Kodor to New Athos | A sub-dialect, central between Abzhywa and Bzyp; the first Abkhaz speech ever recorded |
| Bzyp | New Athos to the Bzyp river | The most phonologically conservative variety; largest consonant inventory |
| Sadz | Gagra to the Matsesta; now Turkey | Nearly extinct; sometimes treated as a separate language |
| Ahchypsy | Upper Mdzymta; now Turkey | Highland variety closely allied to Sadz |
| Tsabal | Upper Kodor; now Turkey | Displaced from the Tsabal highlands in the 1860s |
| Samurzaqan | Gal district | Transitional zone with long-standing Mingrelian contact |
Abaza is the northern face of the Abkhaz–Abaza branch, and it arrived there by migration. In the fourteenth century a movement across the Klukhor Pass carried Abaza speakers onto the northern slope of the range, settling along the Laba, Urup and Zelenchuk rivers; those communities became today’s Tapanta speakers. A second wave in the seventeenth and eighteenth centuries produced the Ashkharywa. The two dialects have been separated by centuries of divergent contact, and the result is a classification puzzle: Ashkharywa is measurably closer to Abkhaz than it is to Tapanta, the dialect on which the Abaza literary standard is built.
Whether Abaza is a language or a divergent form of Abkhaz is therefore a question with no purely linguistic answer. The two are partially mutually intelligible; they have separate literary standards, separate republics and separate orthographic traditions. Scholars generally treat them as two languages while acknowledging the continuum underneath.
Abaza occupies the family’s most exposed position. Roughly fifty-one thousand speakers, no second homeland, and a Turkish diaspora that has largely shifted to Circassian or Turkish. Out-migration from Abaza villages to Cherkessk and beyond breaks the village-level transmission the language has always depended on, while Russian occupies every domain above the household.
In October 1992, Tevfik Esenç died in Hacı Osman, a village near Istanbul. He was 88. With him died the Ubykh language — and the largest consonant inventory ever documented in a human language without clicks.
Counts range from about 80 to 84 depending on the analyst; John Colarusso devoted an entire paper to the question of how many consonants Ubykh actually had, and the answer turns on how one treats the pharyngealized and labialized series. Against them stood three vowels — /a/, /e/, /ə/. No language on record has shown a wider gap between the two inventories.
The Ubykhs held the coast and hinterland around Sochi, between the Abkhazians to the south and the Circassian tribes to the north. When the Caucasian War ended in 1864 with the surrender at Krasnaya Polyana, the Ubykhs — all of them, without exception — chose exile in Ottoman lands over resettlement away from their mountains. Not one Ubykh-speaking community remained in the Caucasus.
In Anatolia the community fragmented. Ubykh villages found themselves scattered among Circassians, Abkhazians, Turks, Kurds, Laz, Greeks and Armenians, and small groups lost contact with one another. The language was vulnerable in a way that is easy to miss: most Ubykhs had been bilingual in Circassian even before the exile, so a ready alternative was already in every mouth. Within roughly a century and a half of the exodus, transmission had ended.
What survives is documentation, and it is unusually good. Georges Dumézil worked on Ubykh from the 1920s into the 1980s, much of it in collaboration with Esenç, who is credited as co-author on a series of publications — an extraordinary arrangement for a field consultant of that era, and a measure of how completely he had become a scholar of his own language. Hans Vogt published the Ubykh–French dictionary in 1963; Julius von Mészáros produced a German dictionary in 1934; Adolf Dirr described the language in 1928; Rohan Fenwick’s 2011 grammar consolidated the tradition. Georges Charachidzé, Rieks Smeets and George Hewitt extended the comparative work.
My language is like a bird that has flown away. You can describe it, but you cannot bring it back. Attributed to Tevfik Esenç — a paraphrase in wide circulation rather than a verified quotation
Ubykh’s position within the family remains genuinely open. Some features align it with Abkhaz–Abaza — the palatalized uvular stops it shares with that branch, for instance — while its lexicon shows deep Circassian layering, part inherited and part borrowed across centuries of bilingualism. Disentangling the two is the central problem of Ubykh comparative work, and it is why the language sits in the middle of the family tree on this page rather than at its edge.
Ubykh was never written by its speakers. Provisional orthographies exist: phonetic transcriptions in the Caucasological tradition, IPA-based systems, and modified Latin and Cyrillic alphabets devised by Adam Dean for community use. Interest in relearning from the recordings has grown among Ubykh descendants in Turkey, and the archival material is substantial enough to support it. What cannot be recovered is a speaker to correct you.
Set the numbers side by side and the scale of the thing becomes plain. English works with roughly twenty-four consonants and twenty vowels; Spanish with nineteen and five. Ubykh had somewhere near eighty-four consonants and three vowels. That is not a difference of degree but of architecture: in a Northwest Caucasian language, almost the entire burden of lexical distinction is carried by the consonants, and the vowels are along for the ride. The richness comes from two sources — an unusually large set of sibilant fricatives and affricates, and the systematic use of secondary articulations, which multiply every basic place of articulation into two, three or four contrasting phonemes.
| Language | Consonants | Count | Vowels |
|---|---|---|---|
| Ubykh † | ~84 | 3 | |
| Abkhaz | 58–67 | 2 | |
| Abaza | ~65 | 2 | |
| Adyghe | 50–60 | 3 | |
| Kabardian | ~48 | 3 | |
| English | ~24 | ~20 | |
| Spanish | ~19 | 5 |
Counts vary between analysts, sometimes considerably. The Ubykh figure has been argued over at length; Abkhaz varies by dialect, with conservative Bzyp at the upper end. Treat every number here as a range with a scholar’s name attached to each endpoint.
Every Northwest Caucasian language has ejectives, and they are the first thing a newcomer hears. An ejective is a voiceless consonant produced not with air from the lungs but with air compressed by a closed glottis rising in the throat — the vocal folds shut, the larynx lifts, and the pressurised pocket of air behind the closure is released in a sharp pop. Think of the difference between an ordinary p and a tight, pressurised p made as though you were spitting out a seed. In IPA the apostrophe marks it.
What is unusual about this family is not the ejective stops — those are common enough worldwide — but the ejective fricatives. Both West Circassian and Kabardian have a glottalized “hissing-hushing” sibilant and a glottalized lateral, sounds rare enough that most phoneticians will go a career without recording one. Ubykh and Abkhaz–Abaza add palatalized uvular stops and fricatives; the Abzakh dialect of West Circassian even contrasts a palatalized glottal stop.
Here is the engine that built these inventories. A consonant can be produced with the lips simultaneously rounded — labialization, written with a raised ʷ — or with the tongue body simultaneously raised toward the hard palate — palatalisation, written with a raised ʲ. Northwest Caucasian uses both, systematically, across almost the whole consonant system. From a single velar k a language derives kʲ and kʷ; Ubykh went further and added pharyngealization, giving a third dimension.
The historical process is the crucial part, because it explains the vowels as well. Rounding and fronting were originally features of vowels. Over time those features were absorbed into the neighbouring consonant, which took on the secondary articulation while the vowel collapsed toward a neutral schwa. Proto-Northwest Caucasian *ki became kʲə; *ku became kʷə. The information did not disappear — it migrated. Every labialized consonant in a modern Northwest Caucasian word is a fossil of a rounded vowel that used to stand beside it.
The consequence is a “vertical” vowel system: the surviving vowels contrast only in height, not in frontness or rounding. Abkhaz and Abaza have two, /a/ and /ə/. Circassian and Ubykh add a mid /e/, giving three. Nikolai Yakovlev first proposed this analysis for Kabardian in 1923, and it has been argued over ever since — Kuipers and Allen went as far as a one-vowel analysis, while Kumakhov, Choi and Fenwick defend three. The three-vowel position rests on minimal triplets that are hard to explain away, such as Besleney Kabardian šxə ‘eat it!’, šxe ‘eat!’ and šxa ‘s/he ate’.
Phonetically, Northwest Caucasian speech contains nearly every vowel quality there is. The point is that those qualities are predictable from the surrounding consonants rather than contrastive — the difference between having few vowels and making few vowel sounds.
All five languages are ergative–absolutive. In English, the subject of “she sleeps” and the subject of “she sees him” are marked identically, while the object stands apart. Northwest Caucasian cuts the other way: the single argument of an intransitive verb is grouped with the object of a transitive one, both taking absolutive marking, while the transitive agent takes a distinct ergative or oblique form. It is the same set of relations described by a different geometry.
And all five are radically polysynthetic. The verb is the sentence. Kabardian has thirteen prefix positions and five suffix positions, each in fixed order relative to the root, encoding subject, object, indirect object, direction, location, negation, causation, modality and more. Roots themselves are short — one or two syllables — so a long word is a long word because of everything piled in front of it.
‘You (pl.) ask us!’
Temirgoy West Circassian · one word, five morphemes, four grammatical relationsAbkhaz builds the same way, and adds gender to the mix. The single Abkhaz word də-z-s-mə́-sə-jṭ — absolutive third person human, potential, first person singular indirect object, negation, the root ‘hit’, declarative — means ‘s/he couldn’t hit me’. Six English words; one Abkhaz one.
Until 1864 the speakers of all five languages lived compactly, in a single continuous territory running from the Kuban plain to the Ingur. Everything after that date is a map of dispersal.
This is the world the rest of this page describes. Read the coast from north-west to south-east and the family tree lays itself out as geography: the Circassian tribes — Natukhay, Shapsug, Abadzekh, Bzhedug, Temirgoy, Besleney — occupy the Black Sea littoral and the Kuban basin; Ubykhs hold a narrow wedge of coast between them and the Abkhazians; Abaza districts sit inland in the central highlands, where the fourteenth-century migration over the Klukhor Pass left them; and Great and Lesser Kabarda stretch east toward the Terek.
The Abkhazian coast repays close reading. Every dialect named in the table above appears here as a territory: Sadzes and Ahibga and Akhchipsou in the north-west, then Bzyb, then Guma and Tsebelda, then Abzhua in the east. Guma sits precisely where Chirikba places it, between the Kodor and the Bzyp — the middle link in the chain, and the first form of Abkhaz that anyone wrote down. Within a century of this map, its speakers were in Anatolia.
Modern linguistic mapping of Western Caucasus tribal composition, 1774–1780. Reproduced with permission — replace this line with the cartographer’s name and licence before publication.
In Turkey the descendants of all three branches are known collectively as Çerkes — Circassians — regardless of whether their ancestors spoke Adyghe, Abkhaz or Ubykh. The label has outlived the distinctions it once tracked.
No language in this family has had a stable writing system for as long as a century. The Soviet alphabet policy of the 1920s and 30s put every Northwest Caucasian language through at least two changes of script within a single generation — Arabic to Latin under the Latinisation campaign, then Latin to Cyrillic when Moscow reversed course in the mid-1930s. Abkhaz went through more turns than any of them, and each change severed a cohort of readers from what had been printed before. The consequence is a literature that has to be re-transcribed to be read.
First written in Arabic script in 1918, Latinised in 1927, and moved to Cyrillic in 1938. The modern orthography leans heavily on digraphs and trigraphs — гъу, шӏу, кӏу — because there is no other way to fit sixty-odd consonants into a Russian typecase. The palochka Ӏ marks glottalization.
Kabardian was the first Northwest Caucasian language to be Latinised — Bekmurza Khuranov’s alphabet was introduced in 1923 — and the first to be returned to Cyrillic, in 1936, two years before its neighbours. Note фӏ, the ejective labiodental that answers to the Adyghe labialized sibilant.
The most turbulent history in the family. Uslar’s Cyrillic alphabet of 1862 ran to 55 characters and underpinned every later Cyrillic system. Mustafa Butba published a Latin alphabet for the diaspora in Istanbul in 1919 — nine years before Turkey itself Latinised.
Then came the churn: Nikolai Marr’s “analytical alphabet” of 76 Latin-based signs, adopted in 1926 and abandoned as unusable; a simpler unified Latin alphabet of 63 letters from 1929; and in 1938, uniquely, a shift not to Cyrillic but to a Georgian-based script — a preliminary to the abolition of Abkhaz-language schooling in 1945. Cyrillic returned only in 1954, after Stalin’s death.
Abaza followed the standard Soviet trajectory — Arabic, then Latin from 1932, then Cyrillic from 1938 — and its orthography is built on the Tapanta dialect. Because Abaza and Abkhaz use different Cyrillic conventions for closely related sounds, the two written traditions look far less alike than the spoken languages do.
Ubykh was never written by its speakers. Everything recorded is in phonetic transcription: the Caucasological roman system used by Vogt and Dumézil, IPA in later work, and Cyrillic renderings in Russian scholarship. Adam Dean produced modified Latin and Cyrillic alphabets intended for community use, the specimen above being a simplified Latin base to which diacritics for glottalization, labialization, palatalisation and pharyngealization are added.
The practical difficulty is arithmetic. Eighty-odd consonants cannot be written with a bare Latin alphabet, so any usable Ubykh orthography ends up either heavily diacritic or heavily digraphic — and there is no longer a speech community to settle which.
Reading about eighty consonants is not the same as hearing four of them in a row. One sample for each of the five languages; four of them are living speech, and the fifth is not.
Connected speech in the western Circassian standard — retroflex sibilants and their labialized partners throughout.
A poem read aloud. Recitation slows the consonant clusters down, which makes it a good register for hearing the ejectives cleanly.
Two vowel phonemes against sixty-odd consonants. Listen for how much of the colour sits in the consonants rather than the vowels.
Abkhaz’s northern relative, spoken in Karachay-Cherkessia — close enough to be partly intelligible to an Abkhaz ear.
Tevfik Esenç, recorded by George Hewitt in the field. There is no other kind of Ubykh recording, and there will not be.
One sample per language rather than per dialect: recordings of individual dialects are not yet available in usable form. Archival Ubykh material derives from the George Hewitt field documentation collection and will be presented with appropriate attribution and permissions.
Every question about the future of these languages runs back through a single year. The end of the Caucasian War in 1864 produced the mass expulsion and flight of the Circassian, Abkhaz–Abaza and Ubykh populations to the Ottoman Empire — a displacement so complete that in the Ubykh case not one village remained behind. What followed in the diaspora was not deliberate suppression so much as structural impossibility. Communities that had transmitted their languages through village life for centuries were resettled among strangers, in a state whose administrative and religious life ran in Turkish, with no schools, no press and no standard orthography. Two or three generations was all it took in most places. Ubykh took slightly longer and ended completely.
UNESCO Atlas classifications. “Vulnerable” means most children still speak the language, but its use is restricted to certain domains; “Definitely Endangered” means children are no longer learning it at home as a mother tongue.
Every language is an irreplaceable way of seeing the world. The Ubykh tragedy was not inevitable — it was the final echo of a nineteenth-century catastrophe whose consequences are still being counted.
The documentary record of this family is the work of a small number of people, several of whom gave most of a working life to it. What is known about Ubykh in particular exists because two men — one a French comparativist, the other a farmer from a village near Istanbul — sat down together, repeatedly, for forty years.
Commissioned to produce an ethnographic description of the Caucasus, Uslar instead produced its first scientific grammars. His Abkhaz grammar of 1862 — lithographed then, properly printed in 1887 — was the first of any Northwest Caucasian language, and he left the first grammatical sketch of Ubykh. Every Abkhaz Cyrillic alphabet since descends from his.
Better known for his work on Indo-European mythology, Dumézil was also the principal documenter of Ubykh. From the 1920s to the 1980s he published grammars, texts, tales, verb studies and etymological notes, a great many co-authored with Tevfik Esenç. Without him the language would survive as a wordlist.
Author of the Dictionnaire de la langue oubykh (Oslo, 1963), still the standard lexicon, complete with a phonological introduction and Ubykh texts. Dumézil later published a full revision of it — a measure of how central the work remains.
His The Northwest Caucasian Languages: A Phonological Survey (1988) remains the reference work on the family’s sound systems, alongside A Grammar of the Kabardian Language (1992) and extensive work on Proto-Northwest Caucasian reconstruction and Nart mythology.
Author of the standard English-language Abkhaz grammar (1979) and editor of the two-volume Indigenous Languages of the Caucasus. He has also written extensively, and controversially, on the politics of Abkhazia.
Common West Caucasian (1996) is the most thorough attempt to reconstruct the family’s phonological system and parts of its lexicon and morphology. His work on Abkhaz dialects, on Ubykh onomastics, and on the history of the Abkhaz alphabets is foundational.
A consonant made with air compressed by the larynx rather than the lungs. The glottis closes, the larynx rises, and the trapped air is released in a sharp pop. Written in IPA with an apostrophe: pʼ, tʼ, kʼ. All Northwest Caucasian languages have them, including — unusually — ejective fricatives.
A secondary articulation in which the lips round at the same time as the main consonant is produced. Written with a raised ʷ: kʷ, gʷ, tʷ. In Northwest Caucasian these are separate phonemes, not variants, and they arose historically from the absorption of rounded vowels into neighbouring consonants.
A secondary articulation in which the body of the tongue raises toward the hard palate during the main consonant. Written with a raised ʲ: kʲ, gʲ. Like labialization, it multiplies the consonant inventory and reflects the earlier presence of a front vowel.
An alignment system in which the single argument of an intransitive verb is marked like the object of a transitive verb (both absolutive), while the transitive agent is marked differently (ergative). English does the opposite. All five Northwest Caucasian languages are ergative–absolutive.
A morphological type in which single words carry as much grammatical information as a whole clause in a language like English. Kabardian verbs have thirteen prefix positions and five suffix positions, encoding every argument of the verb along with tense, negation, causation, direction and more.
A sound unit that distinguishes meaning in a given language. Abkhaz has many vowel sounds but only two vowel phonemes, because the others are predictable from context and never distinguish one word from another.
The International Phonetic Alphabet, a standard notation in which each symbol represents one sound. Indispensable for Northwest Caucasian, where the Cyrillic orthographies rely on digraphs and trigraphs that obscure the underlying phonemes.
A language with no demonstrated genetic relationship to any other. Northwest Caucasian is not an isolate language but an isolate family: its five members are clearly related to one another, and to nothing else that has been shown to general satisfaction.
A chain of varieties in which neighbours understand each other but the two ends do not. The Circassian and Abkhaz–Abaza situations both approximate this, which is why the boundary between “language” and “dialect” here is partly a matter of standards, literatures and borders rather than of linguistics.
A language with no remaining speakers. Ubykh became extinct in October 1992. Documentation may survive in quantity, as it does for Ubykh, but no one remains who can produce a new sentence and judge whether it is right.
A language at risk of ceasing to be spoken. UNESCO grades the risk: vulnerable (children speak it, but only in limited settings), definitely endangered (children no longer learn it at home), severely and critically endangered, then extinct.
The variety selected for writing, publishing and schooling. Adyghe uses Temirgoy, Kabardian uses Baksan, Abkhaz uses Abzhywa, Abaza uses Tapanta. The choice confers prestige on one dialect and, over time, quietly erodes the others.
A vowel system whose members contrast only in height, with frontness and rounding supplied by neighbouring consonants. Yakovlev proposed the analysis for Kabardian in 1923; it remains the standard description of Northwest Caucasian vocalism and is typologically very rare.