Inkuntri
Chinese CJK crossover

Building a Cross-CJK Cognate Deck Without Teaching Yourself Errors

The reader can build a Chinese-Japanese-Korean vocabulary comparison system that captures useful cognates without reinforcing false friends or fake equivalences.

Published March 31, 2026 Chinese

Slug: cross-cjk-cognate-deck-without-errors

Why this deck is worth building

Shared character vocabulary can accelerate learning. A Mandarin learner who also studies Japanese or Korean can recognize broad families of meaning: society, economy, law, research, education, nation, culture, science. But the same advantage can create durable errors because familiar-looking terms feel more trustworthy than they are.

A good cross-CJK deck must separate form, reading, meaning, register, grammar, and domain. A bad deck just puts three “equivalents” on one card and teaches the learner to overtrust them.

Six categories

Use these labels.

LabelMeaningExample type
Safe cognateSimilar meaning and register in common use社会/社会/사회 in many formal contexts
Partial cognateRelated but not identical经济/経済/경제 may vary by phrase and domain
False friendFamiliar form, different meaning手紙/手纸, 勉強/勉强
Register mismatchSimilar meaning, different formality先生 across languages
Script-form differenceSame root, different standard forms学/學/学, 国/國
Pronunciation trapCharacter looks familiar but reading differs sharply学 xué / gaku / hak

Card fields

Every card should include:

  • Chinese simplified form
  • Chinese traditional form if useful
  • Mandarin Pinyin and audio
  • Japanese form, reading, and example sentence
  • Korean Hangul, Hanja if relevant, and example sentence
  • English gloss
  • domain
  • register
  • word class
  • collocations
  • danger label
  • “do not confuse with” field
  • one real sentence for each language, not just isolated words

Example card: 勉強 / 勉强

FieldChineseJapanese
Form勉强勉強
Basic meaningreluctant, forced, barely adequatestudy; also effort in some contexts
DangerMajor false friendMajor false friend
Chinese example这个解释很勉强。means “forced/unconvincing”
Japanese warningDo not map Japanese “study” to Mandarin 勉强Use 学习 for Mandarin “study”

This is a high-priority false-friend card because the forms are so visually convincing.

Example card: 手纸 / 手紙

In modern Mandarin 手纸 can mean toilet paper in some contexts and is not the ordinary word for a letter. In Japanese 手紙 means letter. The card should have a red false-friend warning. Do not bury this in a neutral cognate list.

Example card: 法律 / 법률

This is useful but should still be domain-labeled. Mandarin 法律 and Korean 법률/法律 both relate to law, but legal systems, collocations, and document genres differ. The card label should be “useful formal cognate, verify in legal context.”

Maintenance workflow

A cross-CJK deck is not a one-time import. Use this workflow.

  1. Add the word only after seeing it in a sentence.
  2. Add at least one native example per language.
  3. Mark whether the word is common, formal, technical, literary, old-fashioned, or internet-specific.
  4. Add a danger label if meaning or register diverges.
  5. Review false friends separately.
  6. Recheck cards after real reading exposes new collocations.

Corpus check

Before calling something a safe cognate, search for it in real context. Is it used in news, academic writing, conversation, law, product copy, or only dictionaries? Does it take the same objects or modifiers? Does it appear in the same word class? Does it sound natural to a native speaker?

Deck design rule

Do not put more than one danger on a beginner card. If a word has form difference, pronunciation trap, and false-friend meaning, split it into separate notes. The deck should train caution, not overwhelm.

Build a Cross-CJK Cognate Deck Builder. The tool should require a danger label before export. It should prevent users from saving a card with only “Chinese = Japanese = Korean = English” fields. Add spaced-review tags: safe, partial, false friend, register mismatch, script difference, pronunciation trap.

Remediation and upgrade layer

This capstone article should be more operational than explanatory. The upgraded version needs to give readers a deck architecture that prevents false friends from hardening into memory.

Cognate-deck failure diagnostic

Bad card designWhy it failsRepair
Front: 経済 / 经济 / 경제. Back: economy.It hides pronunciation, grammar, and register.Add local readings, sentence frames, and collocations.
“Same characters = same word.”It turns partial cognates into false certainty.Label match type: safe, partial, false friend, register mismatch.
No example sentences.The learner memorizes dictionary glosses, not usage.Require one natural sentence per language.
No review of old cards.False assumptions persist.Schedule a “false-friend audit.”
Mixing domains randomly.Legal/medical/academic words need special caution.Tag domain and risk level.

Upgraded card template

Each cross-CJK card should include:

  • Concept label, not just English gloss.
  • Chinese form, simplified/traditional if relevant.
  • Mandarin pronunciation with tone.
  • Japanese form and reading, including kana when needed.
  • Korean Hangul and Hanja, when Hanja is relevant.
  • Word class in each language.
  • Example sentence in each language.
  • Register/domain tag: everyday, academic, legal, business, Buddhist, slang.
  • Risk label: safe clue, partial overlap, false friend, rare, archaic, jurisdiction-specific.
  • Do-not-infer note: what the learner must not transfer.

Article-level repairs

Weak version: “Cognate decks can speed up CJK learning.”

Upgraded version: “Cognate decks speed up recognition only when they encode danger. A deck that stores resemblance without risk labels trains overconfidence.”

Weak workflow: “Add a word when you see the same characters.”

Repaired workflow: “Add a word only after checking word status, example sentence, register, and at least one negative-transfer risk.”

Audit protocol

Every month, review cards under these prompts:

  1. Can I produce a sentence in Mandarin without looking at Japanese/Korean?
  2. Does the term have the same word class in every language?
  3. Is the term equally common, or just recognizable?
  4. Is there a false friend or register mismatch?
  5. Would this word be risky in law, medicine, finance, names, or official documents?

Use Unicode Unihan for character readings, national dictionaries for modern word use, and corpus examples for register. For Korean, reveal Hanja only where reliable; for Japanese, distinguish kanji form from actual word usage.

Related reading