Korean Translator: Why One Text Is Really Two Languages
October 27, 2026 · KoryoTalk · 7 min read
Short answer: “Korean” in practice means two state norms: southern 표준어 and northern 문화어. A generic Korean translator is trained mostly on southern text: it reads a North Korean text through southern meanings and guesses or drops unfamiliar northern words. A northern text translates correctly only with a tool whose base includes the DPRK norm — the KoryoTalk bot, with a dictionary base of 2.2 million Korean–Russian entries plus a separately collected layer of northern correspondences.
Why “Korean” is two languages of norms
After 1945 the written norm developed separately. South Korea codified 표준어, the DPRK codified 문화어. The differences go well beyond accents: different spellings of the same word, different words for the same thing, different politeness forms. A text natural for a Seoul reader is spotted as “foreign” by a Pyongyang reader from the first line — and the other way around. See “one language, two norms”.
For a translator this means one thing: identify the norm before translating, because the norm changes what words mean. The same syllable can be two different words in the two norms.
Trap table: what a translator reads wrong
| Word | DPRK norm meaning | What a southern-trained translator does |
|---|---|---|
| 로동자 | worker (spelled with “로”) | finds nothing or outputs “?” — the southern spelling 노동자 is all it knows |
| 력사 | history (spelled with “레”) | finds nothing — the southern spelling is 역사 |
| 동무 | comrade — a live, neutral address | flags it as archaic or translates it literally |
| 택시 | taxi (northern spelling of the loanword) | finds nothing — absent from a southern base |
| 감사합니다 / 옳소 | “thank you” / “that’s right” — bookish politeness forms | translates the words but loses the politeness level |
* The spellings 로동자, 력사, 동무, 택시 are dictionary lemmas of the 문화어 norm; southern spellings (노동자, 역사) are not listed as separate words. The forms 감사합니다 and 옳소 are politeness forms of the lemmas 감사 and 옳다.
The first two rows are the most common case: because of the initial-sound rule (두음법칙) adopted in South Korea and not applied in the DPRK, half of the service vocabulary is spelled differently. See the article on 두음법칙 and the North.
Three error types on northern text
Type one — “word not found”. A northern spelling is absent from a southern-trained base: 로동자, 력사, 택시 do not translate at all. Part of the text simply disappears from the output.
Type two — “southern meaning”. The word exists in both norms but means different things. The classic is 동무: in the North it is an ordinary address “comrade”, while a southern-trained model renders it with an archaic shade. The output looks coherent but the tone is wrong.
Type three — “politeness level”. The northern norm leans on the bookish endings -습니까 / -습니다; conversational constructions are rare in its written language. A model trained on southern text smooths these into a conversational register — the word meanings survive, the distance between the speakers does not.
The order for translating a northern text
The proven order is four steps, as in “how to translate Korean into Russian”:
- identify the norm by marker words (로동자 vs 노동자, 력사 vs 역사, 동무 as an address);
- if the norm is northern, translate with a northern-base tool: KoryoTalk, a bot translator of the North Korean language (조선말);
- check the trap-table words above in the output — they cause most errors;
- to reply to a DPRK addressee, build the phrase in the DPRK norm — see “translating Russian into Korean”.
For a quick norm check and translation: open the KoryoTalk bot. More reading aids: the basic Korean word list and “how to say it in Korean”.
Conclusion
A Korean translator answers correctly only when the norm of the text matches the norm of the base. A northern text (문화어) in a southern-trained translator loses words, meanings and politeness levels. The fix: identify the norm with markers, translate with a northern-base tool, and check the trap-table words.