🇷🇺Nivkh for English speakers

1. Nivkh: the language of the lower Amur and northern Sakhalin

Nivkh is the Indigenous language of the lower Amur River, the Amgun basin and the northern half of Sakhalin, in the Russian Far East. Linguists file it as a small language family that is effectively an isolate: nothing outside it has ever been shown to be related to it, and Glottolog lists it as a one-member family, glottocode nivk1234, ISO 639-3 niv.¹¹⁷ Everything about it is unfamiliar if you come from English — the word order is subject–object–verb, there is no class of adjectives, there are two tenses rather than three, counting requires you to know what kind of thing you are counting, and the first consonant of a word changes depending on what stands in front of it.¹³¹⁷

The name of the language in the language is built on the word for a person, нивх (nivx). The dossier this guide is built on records the general form and two variety-specific ones:

• Нивхгу диф — nivxgu dif
• 'the Nivkh language' — the general citation form.²¹⁷
• мер нивх диф — mer nivx dif
• 'our Nivkh language' — the Amur form.¹⁷
• нивғӈ дуф — niɣvŋ duf
• 'the Nivkh language' — the south-east Sakhalin form. Even the name of the language changes shape across the varieties, which is the first honest thing to know about Nivkh.¹⁷

The independently harvested Wiktionary lexicon confirms the same spread of the word for 'person' — нивх (nivx), нивӈ (nivŋ), нивғӈ (nivɣŋ) and ниғвӈ (niɣvŋ) are all recorded — which is a useful cross-check that the dossier's dialect note and the lexical record describe the same reality.¹⁵¹⁷

Why an English speaker might learn Nivkh

• It is one of the very few languages on earth with Celtic-style initial consonant mutation outside the Celtic family — and in Nivkh the mutation is triggered by syntax, so you can literally hear phrase boundaries.⁹¹⁷
• Its numeral classifiers are the standard textbook example of language obsolescence: the documented system counts people, animals, objects, boats, sleds, fishing seines and rods strung with smelt with different words for 'one'.¹⁶¹⁸
• It has an unusually strong scholarly apparatus for a language this small — a sketch grammar, a syntax, a typological monograph, a comparative dictionary and a multi-volume audio archive — and a native-speaker scholar, Galina Otaina, among its principal describers.⁵⁶⁷⁸¹¹
• There is no course, no app, no textbook in English and no CEFR-aligned material anywhere. That is a statement of fact, not modesty: this guide exists because the gap is real.¹⁷

A warning about this guide, before you read any further

Nivkh documentation is thin at exactly the point where a language course needs it to be thick. There are excellent descriptions of the grammar and almost no glossed connected text in the public domain. The in-repo research dossier this guide draws on says so explicitly: no verified core wordlist was available when it was compiled, and it forbids presenting AI-generated Nivkh.¹⁷ This guide honours that. Every Nivkh form below is either quoted from the dossier's attested material or taken from a harvested Wiktionary lexicon whose entries derive chiefly from Savel'eva and Taksami's 1970 Nivkh–Russian dictionary and Fortescue's 2016 comparative dictionary — a community-edited reference tier, not a primary source.¹⁵¹⁹⁶ Where a reader text was assembled for teaching rather than recorded from a speaker, it says so at the top of the reader.

2. How many people speak Nivkh? Three numbers that disagree

You will see three very different figures quoted, sometimes in the same paragraph. All three are real; none of them can be used alone.

The 2010→2020 jump is not recovery. Nobody learned Nivkh natively in that decade. The 2020 census asked a different question in a different way, and more people answered 'yes, I know Nivkh' — which can mean anything from full fluency to a handful of words learned in a grade-3 classroom or claimed as an act of identity.⁴¹⁷ Specialists describe the fluent population as elderly, Russian-bilingual, mostly born in the 1920s to 1940s, and dispersed.⁵¹⁴ The honest formulation is: fewer than 200 fluent speakers, around 1,300 people who claim some knowledge, out of roughly 4,650 Nivkh. Present the series; do not pick a number.

Why the decline happened

The documented causes are policies with dates, not anything about the community or the language. Sakhalin was made a penal colony from 1873, bringing settlers and epidemics including smallpox.²¹²⁴ Stalin-era collectivization in the 1930s dismantled the subsistence economy the language was embedded in; families were resettled into centralized towns such as Nogliki; children were schooled in Russian, often away from home.²¹²⁴¹⁷ The 1920s–1940s birth cohorts became the last native generation.³¹⁷ Nivkh did not fail. It was displaced, on a schedule, by policy.

The site's own research dossier puts it plainly: 'Any guide should attribute the decline to these policies, not to any deficiency of the community.'¹⁷

3. Sounds and the Cyrillic alphabet

Nivkh has never had a script of its own. A Latin alphabet was introduced in the 1930s and replaced by a Cyrillic one in the 1950s; the modern orthography is Russian Cyrillic plus a set of special letters and apostrophe-marked aspirates.¹⁰¹⁷ Which shapes of those special letters count as correct is decided in practice not by a state body but by Нивх диф, the monthly Nivkh newspaper — a community publication is the de facto orthographic authority, and this guide and its vocabulary deck follow its allographs.¹⁰¹⁷

Vowels

Six vowels. Length is not phonemic; long vowels in the recordings come from prosody or compensatory lengthening.¹⁷

The special consonant letters

This is the table to keep open. The third column is the romanisation used throughout this guide and in the Nivkh vocabulary deck — it is a LinguaCommons letter-by-letter scheme mapping each Cyrillic grapheme to the IPA symbol the standard alphabet table assigns it, because no standardised Nivkh romanisation exists.¹⁷ Do not mistake it for a pronunciation guide to any individual speaker.

Two contrasts do most of the damage to English speakers. The first is velar versus uvular: к/х are made where English makes them, қ/ҳ are made a good centimetre further back, and the difference is lexical. The second is the three-way laryngeal contrast in the stops — aspirated (кʼ), plain (к), and voiced (г) — where English only reliably distinguishes two.¹⁷ Then there is the voiceless trill Р̌ /r̥/, which is genuinely rare in the world's languages, and Ы.¹⁷

Letters that write two sounds

Д, Н and Т are polyvalent. Each writes an alveolar consonant normally — /d n t/ — but a palatal or affricate one before an iotated vowel (я ё ю) or before a syllable-final ь.¹⁷ That is why иньдь (ińǯ) 'food; to eat' romanises with two palatals and why тяӄо (čaqo) 'knife' begins with an affricate. Е, Ё, Ю and Я are /je jo ju ja/ at the start of a syllable. Ё, Щ and Ь outside that role appear only in Russian loanwords such as остол (ostol) 'table', мыло (məlo) 'soap' and сета (seta) 'sugar'.¹⁵¹⁷

The font problem is real

Ӷ, Ӻ, Ӽ and especially Р̌ — which is an ordinary р plus a combining caron — are poorly supported. Janhunen and Gruzdeva devoted a 2016 paper to exactly this: bringing the orthography of an Indigenous language into the digital age, where the special letters either fail to render, render in the wrong shape, or get normalised away by software.¹⁰ Expect to see Р̌, Р̆ and РШ all standing for the same sound, and expect copy-paste to lose the caron. The Amur variety, separately, does not use Ӷ at all and writes the voiceless trill as РШ.¹⁷

Reader 1 (A1.2) — «Урла муғв кʼинӈула ӈафӄғу!»

This is an attested text, reproduced from the LinguaCommons research dossier, which reproduces it from a referenced overview of the language; it is associated with the Nivkh-authored text by A. A. Yushkina published by Gruzdeva and Bugaeva in 2021.¹²¹⁷ It is short — two sentences — because that is the length of the verified opening; the dossier flags the rest of the passage as truncated in the rendering it was taken from and instructs that only this much be quoted.¹⁷ Two sentences of genuine Nivkh greeting register is a better beginning than twenty sentences nobody has said.

Урла муғв кʼинӈула ӈафӄғу!

Чʼыӈ ордь ни маӈгут эзмудьра.

English: 'Hello, dear friends! I am very pleased to meet you.'¹⁷

The dossier gives its own transcription of the passage, which differs in small ways from this guide's letter-by-letter romanisation because it transcribes speech rather than spelling — note муғв romanised there as muɣf, with the final consonant devoiced:

Urla muɣf kʰinŋula ŋafq-ɣu! · Cʰəŋ or-ɟ ɲi maŋgut e-zmu-ɟ=ra.

Compare our spelling-based romanisation of the same words: урла (urla), муғв (muɣv), ни (ni). Both are legitimate; they answer different questions.¹⁷

Character count of the attested Nivkh in this reader: 53 characters of Cyrillic plus the 61-character transcription. The house style for these guides asks for a 1,200-character reader. For Nivkh that is not available, and padding it with invented sentences would be worse than useless — see section 11.

4. Initial consonant alternation — grammar you can hear

In Nivkh the first consonant of a morpheme is not fixed. If a morpheme has a phrase-mate to its left — a possessor, an object, a modifier, a prefix — then an initial stop turns into the matching fricative or trill. The system is explicitly compared to Celtic initial mutation in the literature; Shiraishi's phonological work is the current specialist analysis.⁹¹⁷ Because it is triggered by what is syntactically next to the word, it is, in a real sense, audible syntax.⁷⁹

The mutation is blocked when the preceding morpheme itself ends in a fricative, a trill, a nasal, or /l/.⁹¹⁷ So the trigger is not 'is there a word in front of me' alone — it is 'is there a word in front of me, and does it end in a sound that lets the mutation through'. Learn the four attested phrases below by heart before you try to apply the rule; that is the sequence the dossier recommends, and it is the right one.¹⁷

Reader 2 (A2.1) — the soup set

This is attested material, reproduced from the LinguaCommons research dossier, which takes it from a referenced overview citing Shiraishi's analysis.⁹¹⁷ It is a phrase set, not connected prose, and it is presented that way honestly: four noun phrases plus two control words. It is the single most-quoted piece of Nivkh data in the typological literature, and it is worth more to you than a page of anything invented.

пыңх · пынрай выңх · амсп выңх · тьхыф пыңх

/pəŋx/ · /pənraj‿vəŋx/ · /amsp‿vəŋx/ · /cxəf‿pəŋx/

English: 'soup · duck soup · seal soup · bear soup'.¹⁷

Two things follow immediately. First, dictionary lookup is hard in Nivkh, because the word you hear may not begin with the letter the word is filed under — выңх is пыңх. Second, a mutation you can hear tells you that two words belong to the same phrase, which in a pro-drop SOV language is genuinely useful parsing information.⁹¹⁷

A dispute worth knowing about

Some transitive verbs look as though the alternation runs backwards — a fricative where you expect a stop. Older descriptions treated this as a separate hardening process. Shiraishi argues it is the same lenition all along, with an underlying stop that was lenited by a historical *i- prefix, in a paper whose title is the argument: 'Nivkh Consonant Alternation Does Not Involve Hardening'.⁹ Present-day specialist opinion follows Shiraishi; the older description is not thereby erased.¹⁷ Noun-initial fricatives, for what it is worth, never alternate at all.⁹

5. Building sentences: SOV, pro-drop, no adjectives, two tenses

Nivkh is agglutinative and synthetic, overwhelmingly suffixing, with a developed case system and a high degree of incorporation — spatial relators, the equivalent of English prepositions, incorporate into the noun they govern rather than standing separately.⁷¹³ Mattissen's typological monograph coins the label 'dependent-head synthesis' for this profile, and it is her book that made Nivkh a standing reference point in the polysynthesis literature.⁷

An honesty note about paradigms

This guide contains no case-suffix table and no verb-agreement paradigm. That is not an oversight. The research dossier states that no paradigm table was available from a source consulted directly, and that inventing cells is prohibited; the paradigms exist in Nedjalkov and Otaina's syntax of Amur Nivkh and in Gruzdeva's sketch grammar, and the correct move is to send you there rather than to guess.⁵⁸¹⁷ Verb forms below are cited exactly as the lexicon gives them, in the -д / -дь finite form, and no form has been inflected, derived or altered.

Core verbs

Pronouns

Nivkh distinguishes inclusive from exclusive 'we' and has a dual — a distinction English lost long ago and Russian never had.¹³¹⁵

Reader 3 (A2.2) — people and actions

This is a constructed teaching text — every word in it is attested in the sources cited below, but the passage itself was assembled for this guide. Specifically: every lexeme comes from the harvested Wiktionary lexicon in its citation form, the order is the documented subject–object–verb, and nothing has been inflected, derived or invented.¹³¹⁵ It has been built under one extra self-imposed constraint that you should know about: every noun and verb in it begins with a vowel, a fricative or a sonorant, never with a stop. That is deliberate — a stop-initial word in object position might or might not mutate, and this guide is not going to guess (see section 4). No speaker has vetted these sentences. Treat them as decodable, not as models of idiomatic Nivkh.

Ни видь. Чи видь. Иф видь. Ньыӈ видь. Имӈ видь. Ни иньдь. Ни леп иньдь. Умгу леп иньдь. Оӻла леп иньдь. Умгу леп ед. Ни лу лудь. Ньыӈ лу лудь. Ни му ивдь. Умгу му ивдь. Ни мрыдь. Ни итть.

English: I go. You go. He goes. We go. They go. I eat. I eat bread. The woman eats bread. The child eats bread. The woman cooks bread. I sing a song. We sing a song. I have a boat. The woman has a boat. I swim. I speak.

6. Counting: the numeral classifier system

WALS classifies Nivkh as having obligatory numeral classifiers.¹³ What that means here is stronger than in, say, Mandarin or Japanese, where a classifier word sits between the numeral and the noun. In Nivkh the numeral itself is different. 'One' counting people is нин; 'one' counting animals is ньын; 'one' counting a flat object is няӄр̌; 'one' counting boats is ним; 'one' counting sleds is нир̌; 'one' counting fishing seines is нео; and 'one' counting rods strung with smelt is ньӈаӄ.¹⁶ The kaikki.org mirror of English Wiktionary records 58 numeral senses for Nivkh, and the categories they encode are a portrait of a subsistence economy: boats, sleds, seines, fishing nets, holes in a seine, strands of rope, bunches of yukola (dried fish), bunches of dog food, and tackle for seal and kaluga sturgeon.¹⁶

Read the dashes correctly. A dash means the harvest did not record a form for that cell — not that the form did not exist. The sources record no classifier-specific variants at all for six, seven and eight, which may be a real property of the system or may be a gap in the record; this guide is not going to fill it in.¹⁶ Fortescue's comparative dictionary is where a serious answer lives.⁶

Why this table is also an obituary

Nivkh classifiers are the standard example of obsolescence in the language-death literature. David Crystal's Language Death uses precisely the fine distinctions in this table — separate counting words for sledges and for fishnets — as the illustration of what erodes first when a language stops being the working language of a community.¹⁸ Modern speakers have collapsed the fine categories into a small number of general ones; the sledge-and-fishnet distinctions have effectively disappeared in practice.¹⁷ You are therefore learning two things at once: the documented system, which is what the written record and the dictionaries contain, and the fact that a living speaker will very likely not use it that way. Both are true. Neither replaces the other.

There is a second, quieter contact effect. Under Russian influence Nivkh has innovated plural marking on counted nouns and on pairs, which the older system did not have.⁵¹⁷ Gruzdeva's work on Russian–Nivkh grammatical interference documents parallel restructuring in the imperative.⁵

Reader 4 (B1.1) — the day, the weather, the year

Another constructed teaching text — every word in it is attested in the sources cited below, but the passage itself was assembled for this guide. It exists to make one grammatical point that a word list cannot: Nivkh has two tenses, future and non-future, so the same verb form does present and past, and it is the adverb that tells you which.⁸¹³ Наўх 'today' and намр̌ 'yesterday' are doing the work that English does with -ed. The same constraints as Reader 3 apply: citation forms only, SOV order, no stop-initial words in a mutation-ambiguous position, no speaker verification. One line deliberately breaks the stop rule — лых кыд, 'rain falls' — because лых ends in the fricative -х, which blocks mutation, exactly as тьхыф does in Reader 2.⁹

Наўх лых кыд. Намр̌ лых кыд. Ырты-ырты лых кыд. Наўх ни видь. Намр̌ ни мрыдь. Ығрӿагин ни лудь. Наўх ньыӈ лудь. Оӻла лудь. Ниғвӈ иньдь. Умгу леп ед. Ныкы ни видь. Ни ивдь.

English: Today it rains. Yesterday it rained. It rains constantly. Today I go. Yesterday I swam. I always sing. Today we sing. The child sings. The people eat. The woman cooks bread. I go a little way. I have (it).

Reader 5 (B1.2) — counting to ten, three times over

This is a constructed teaching text in a weak sense: every numeral in it is individually attested in the sources cited below, but the sequences were assembled for this guide. Listing numerals in order makes no grammatical claim — there is no syntax here to get wrong — which is why it is the one place this guide is comfortable producing several hundred characters of Nivkh. Read each row aloud as a unit; the point is that the three rows are the same numbers.¹⁶

Ниғвӈ ырӈ: Нин, менӈ, тяӄр̌, нырӈ, тʼорӈ, ӈах, ӈамк, минр̌, няндорӈ, мӽоӈ.

Counting people: 'one, two, three, four, five, six, seven, eight, nine, ten.' (The label Ниғвӈ ырӈ is this guide's descriptive tag, built from ниғвӈ 'person' and ырӈ, the element attested in ань ырӈ 'season'; treat it as a heading, not as an attested phrase.¹⁵)

Ӈа ырӈ: Ньын, мар̌, тёр̌, нур̌, тʼор̌, ӈах, ӈамк, минр̌, няндор̌, мӽос.

Counting animals: the same ten numbers, different words. Note that six, seven and eight do not change — the record has no classifier-specific forms for them.¹⁶

Ырӈ: Няӄр̌, меӄр̌, тяӄр̌, ныкр̌, тʼоӄр̌, ӈах, ӈамк, минр̌, няндор̌, мӽоӄр̌.

Counting objects: again the same ten numbers.

Ним … няндорӈ. Нир̌ … няндорӈ. Нео, мевор … мӽоу.

And the specialist series, partially recorded: ним 'one boat' … няндорӈ 'nine boats'; нир̌ 'one sled' … няндорӈ 'nine sleds'; нео 'one seine', мевор 'two seines' … мӽоу 'ten seines'. The ellipses are honest — the intervening cells are not in the record.¹⁶

7. Everyday words: people, kin, time, land, food

All of the following come from the harvested lexicon, which is the kaikki.org mirror of English Wiktionary — reference tier, deriving chiefly from Savel'eva and Taksami's 1970 Nivkh–Russian dictionary and Fortescue's 2016 comparative dictionary, but community-edited.¹⁵¹⁹⁶ Where two forms are given for one meaning they are attested doublets; the source does not label which variety each belongs to, so neither does this guide. The full 407-term deck is linked at the end.

People and kinship

Time, seasons and weather

Land, water and living things

The lexicon is dense in exactly the places a river-and-coast subsistence economy would make it dense: fish, seals, boats, nets, birds. That density is data about how people lived, and it is why the classifier system looks the way it does.¹⁵¹⁶

House, food and things

The body

Noises people make

Interjections are the closest thing the harvested lexicon has to recorded speech, and several of them are commands to a dog team — which tells you something about what the language was used for.¹⁵

• ӿы · ӿыи · ыныя · ӿана · ӿай
• 'uh-huh (agreement)' · 'whoa! (surprise)' · 'yay! hurray!' · 'come on, well (encouragement)' · '(fear or fright)'¹⁵
• хей хей · фай фай · ыпып · ӿу ӿу · пы пы
• '(regret, pity)' · 'ew (disgust)' · 'brr (cold)' · 'phew (weariness)' · 'hushaby (to a baby)'¹⁵
• порр̌ · тюй · ухта
• 'stop / slow down (to a dog team)' · 'left! (to a dog team)' · '(delight after catching a bear)'¹⁵

A note on that last one. Nivkh bear ceremonialism is culturally significant and sensitive, and the research dossier deliberately keeps it out of scope pending community-appropriate sources.¹⁷ This guide does the same: the word is listed because it is in the lexicon, and nothing further is said about the practice.

8. Common learner difficulties

There is no study of Nivkh learners anywhere. The research dossier says so and labels its own list of difficulties as pedagogical inference from the documented structure, not measured error data.¹⁷ The same caveat applies to everything in this section. These are the places the structure predicts trouble.

• Initial consonant alternation wrecks word recognition and dictionary lookup at the same time. The word you hear is выңх; the word in the dictionary is пыңх. Until the system is internalised you will fail to find words that are right in front of you.⁹¹⁷
• The laryngeal and place inventory is larger than English or Russian. Aspirated versus plain versus voiced stops, velar versus uvular, and the voiceless trill /r̥/ are four contrasts an English ear does not have wired.¹⁷
• Ы. English has no high central vowel and English speakers reliably substitute /ɪ/ or /ə/ in the wrong places.¹⁷
• The polyvalent letters Д, Н and Т mean you cannot read Nivkh Cyrillic phonetically until you know the ь and iotated-vowel rules cold.¹⁷
• No adjectives means rethinking, not translating. 'The long river' is not adjective + noun; the property word кылдь is a verb.¹³¹⁷
• Two tenses means you must stop looking for a past marker. It is not there; the adverb is carrying the load.⁸¹³
• Classifiers require you to hold two systems at once: the documented one, which is what you will read, and the collapsed modern one, which is what you would hear.¹⁷¹⁸
• Variety mixing is the silent killer. Amur and Sakhalin forms differ enough that assembling a sentence from two sources can produce a variety nobody speaks. Label every form you learn.⁵⁶¹⁷
• Font and encoding problems will corrupt your notes. Р̌ is a combining sequence; some software normalises it away, some renders it as a box.¹⁰

If you already speak Russian

Russian is the contact language and the language of most of the scholarship, which is an enormous practical advantage. It is also the source of the documented interference: Russian-model plural marking on counted nouns and restructuring in the imperative are the specific changes attrition produced.⁵¹⁷ Material should flag those as contact forms rather than presenting them as the older system. The shared Cyrillic script helps less than it looks like it will, because the special letters and the polyvalent Д/Н/Т rules are exactly where Russian habits mislead.

9. Dialects and variation

Nivkh varies by village, by clan, and by individual. The varieties are conventionally grouped into four geographic clusters, and how many languages those clusters add up to is genuinely disputed.⁵⁶¹⁷

Three framings, all in current use

The third row is not a mistake that the first two rows correct. It is a documented, reasoned choice by the people whose language it is, made for a stated reason, and it governs how the language is presented in revitalization contexts.¹⁷ LinguaCommons follows community practice by offering Nivkh as one language, while stating plainly that specialists count two or three mutually unintelligible ones and labelling forms by variety wherever a source labels them. The comparative-method literature, which works on regular sound correspondences across the varieties, generally follows the two-or-three analysis for technical reasons.²³

Concrete differences

Doublets in the deck, and what to do with them

The harvested lexicon does not tag its entries by variety, so the vocabulary deck cannot either. What it can do is show you where the doublets are, so you know which words to check before you use them. These are the most conspicuous pairs:

Notice the pattern in several of these: an -р̌ form beside an -с form, a -в beside a -ф, an а beside an ы. Those are the regular correspondences the comparative work on Nivkh is built on, and they are the reason Proto-Nivkh can be reconstructed at all.⁶²³

10. Media, culture and community

This is where the honest inventory matters most, because for Nivkh several of the usual categories are genuinely empty.

Print

Нивх диф (Nivkh Dif, 'Nivkh Language') is a monthly newspaper published on Sakhalin, and it is the only Nivkh-language newspaper. Its importance goes well beyond circulation: because it must choose between competing allograph shapes for the special Cyrillic letters every time it sets a page, it has become the de facto orthographic authority for the language, and the standard scholarly treatment of Nivkh orthography takes its choices as the reference point.¹⁰¹⁷ A community publication deciding the spelling standard is unusual and worth stating plainly. Nivkh-language book publication also continues in Russia.¹⁷

Audio — the standout resource

Sound Materials of the Nivkh Language, recorded and edited by Hidetoshi Shiraishi with Galina Lok, is the largest Nivkh sound archive on the web and describes itself that way. The research dossier records eight volumes (2002–2013) totalling around five and a half hours of transcribed conversation; the site itself, checked on 4 August 2026, advertises an eleventh volume (Natal'ia Dem'ianovna Vorbon) and was last modified in July 2015, so the series ran further than the dossier's count.¹¹¹⁷ The site also posts sample sound files specifically for palatalization and for consonant mutation — which means you can hear the process described in section 4 rather than only read about it.¹¹ If you do one thing from this guide, do that.

Published Nivkh-authored text

Gruzdeva and Bugaeva published a linguistic text in Nivkh by A. A. Yushkina in the Journal of Ainu and Indigenous Studies in 2021 — a Nivkh-authored text entering the scholarly record, which is where the greeting in Reader 1 comes from.¹²¹⁷ It is worth naming Galina Otaina here too: she co-authored the standard syntax of Amur Nivkh with Vladimir Nedjalkov, and she was herself Nivkh.⁸¹⁷ The language's description is not something that only happened to it from outside.

Schooling

Nivkh is taught in grades 1–3 in several schools in Sakhalin Oblast and Khabarovsk Krai, as of the sources' reporting in the 2010s.¹⁷ Three years of primary-school exposure is not language transmission, and describing it as such would be dishonest; it is, however, the institutional foothold that exists.

What does not exist

There is no Nivkh feature film, no commercial music catalogue, no podcast, no broadcast radio or television service in the language, no mainstream course, no app, no CEFR-aligned material, and no substantial English-medium textbook.¹⁷ Every one of those is a real absence, not a failure of this guide's research. The Endangered Languages Project entry, the census record and the scholarly literature all describe a language whose remaining public life is a monthly newspaper, a school programme in three grades, an audio archive maintained by a Japanese phonologist, and a shelf of reference grammars.¹¹¹⁴¹⁷

Reading at C1: what 'unedited text' means here

For most languages, C1 means reading a newspaper without help. For Nivkh you can attempt that literally — Нивх диф exists — but you will be doing it without a bilingual dictionary in English, without a spell-checker, without a corpus, and against initial consonant alternation that changes the letter a word is filed under. The realistic C1 skill for Nivkh is different: it is reading glossed text in the scholarly literature, following an interlinear analysis in Nedjalkov and Otaina or Mattissen, and listening to Shiraishi's recordings with the transcriptions open.⁷⁸¹¹ That is a real advanced skill and this guide can honestly point you at it.

11. What mastery looks like, and where this guide stops

At C2 in a well-resourced language you would be reading literature, handling register and idiom, and writing at length. This guide does not offer a C1 or C2 reader, and it will not, because constructing several hundred words of advanced Nivkh prose would require inventing morphology, and inventing Nivkh is exactly what the source material forbids.¹⁷ Five readers appear above. Two of them are attested and short. Three are constructed from citation forms in SOV order and are labelled as such at the top. Their combined length is a few hundred characters, not the 1,200-per-band the house style asks for. That gap is the honest state of the evidence, and closing it with fabricated sentences would make this guide worse, not better.

What C2 actually consists of for Nivkh is command of the literature and of the open questions in it.

Three disputes you should be able to state

And one about this guide

12. Resources

Everything below was checked on 4 August 2026 unless noted. The audio archive is the one you should open first.

• Nivkh vocabulary — the full 407-term deck, with SRS trainerA1–B2 — Cyrillic plus Latin romanisation, browsable as a table, with CSV / TSV / Anki / .apkg downloads.
• Sound Materials of the Nivkh Language (Shiraishi & Lok)All levels — The largest Nivkh sound archive on the web: transcribed conversation across multiple volumes, plus sample files for palatalization and for consonant mutation. Live 2026-08-04.
• Glottolog 5.3 — Nivkh (nivk1234)Reference — Classification, the 'insufficient' judgment on Chukotko-Kamchatkan–Amuric, and the reference list behind it. Live 2026-08-04.
• kaikki.org — Nivkh machine-readable dictionaryReference — 696 word forms extracted from English Wiktionary; the direct source of this guide's vocabulary deck. Reference tier: community-edited, ultimately from Savel'eva & Taksami (1970) and Fortescue (2016).
• kaikki.org — Nivkh numerals (58 senses)B1+ — The classifier series in full, including the ones no summary reproduces: strands of rope, bunches of yukola, seal and kaluga tackle.
• WALS Online — NivkhReference — Typological profile: SOV, eight or nine cases, obligatory numeral classifiers, two tenses, no gender, inclusive/exclusive pronouns.
• Endangered Languages Project — Sakhalin NivkhReference — Severely endangered; census figures and script history. The page is JavaScript-rendered and returned no text to an automated fetch on 2026-08-04 — open it in a browser.
• Gruzdeva & Bugaeva, 'A unique linguistic text in Nivkh by A. A. Yushkina' (2021)C1+ — Nivkh-authored text in the scholarly record. PDF; not opened for this guide.
• Shiraishi, 'Nivkh Consonant Alternation Does Not Involve Hardening' (2000)C1+ — The paper behind section 4. PDF; a fetch attempt during the dossier's compilation returned empty content, so it is cited via summary.
• Wikipedia — Nivkh languagesReference — Caution tier, but densely referenced to Gruzdeva, Fortescue, Mattissen, Shiraishi, Nedjalkov & Otaina and Arefiev; the conduit through which the research dossier obtained the attested greeting, the soup set and the alphabet table.

Where to go next

• Open the audio archive and listen to the consonant-mutation samples while re-reading section 4. Everything else in this guide is downstream of hearing that once.¹¹
• Work through the 407-term deck. It is not a frequency list — no frequency data exists for Nivkh — but it covers people, kin, land, water, food, time, the body, core verbs, the pronoun system and the whole recorded classifier apparatus.¹⁵¹⁶
• If you read Russian, Panfilov and Savel'eva & Taksami are the foundation, and the Russian-language scholarship is where most of the detail is.¹⁷¹⁹
• If you can get to a library, Gruzdeva (1998) is the single most efficient next step, and Fortescue (2016) is the one that will let you label a form by variety.⁵⁶
• Use the name Nivkh. Search under Gilyak.¹⁷