Alternative to #544

Made on behalf of @Annyonghaseyo a Korean native that lives in Japan and regularly attends the Tokyo Bitcoin Meetup. She is also the translator of a lot of Korean bitcoin apps.

I don't really have much to say in the matter except the limited knowledge I have of Korean based on similarities to Japanese. She echoed the concerns of using -hada verbs in the list which was also echoed by others and myself.

This list is NFKD normalized and sorted in order of UTF-8 binary so as to allow for binary search.

nice work @junderw @Annyonghaseyo

All words on the list are not difficult, easy to remember and well-following the rule mentioned by @Annyonghaseyo, and bip39

this is much better than #544

This list looks good. Non-Korean speaking ACK.

@luke-jr Looks like we have an ACK from BIP author @voisine and Korean natives @Annyonghaseyo and @gnujoow both see this wordlist as sufficient. Please merge if everything seems in order.

Thanks everyone.

Just generated 10 phrases using bitcore-mnemonic.

Can't understand it, but looks good.

역시 삼십 하드웨어 보통 세월 고궁 시골 제한 한마디 장미 청춘 관념
암컷 영국 같이 토마토 법률 탤런트 고전 마음 갈색 공업 점점 공동
예선 고구려 목숨 흥분 모퉁이 단골 시아버지 바탕 울음 사슴 의학 원인
신비 명예 내일 모양 교직 연출 전문 팝송 부문 부탁 사나이 창고
정비 왕자 태권도 만화 번개 전체 중학교 막걸리 입사 서명 승용차 전시
탤런트 책가방 형부 주택 퇴근 강아지 인공 사냥 어려움 극히 엔진 기원
무릎 테니스 전혀 알루미늄 간섭 즉석 칠십 밥솥 예전 엔진 줄거리 임무
행복 도저히 주말 벌레 공부 백화점 혹시 추천 교과서 자연 별도 법률
촛불 의논 단추 지진 고양이 습기 가뭄 시아버지 전통 정성 매장 가족
진짜 홍수 상자 강도 유적 지리산 막걸리 닭고기 햄버거 사슴 클래식 진료

@junderw it looks good for korean as well :)

ㅎㅏㄴㄱㅡㄹ, I mean, 한글.

How should one treat un-normalized Hangeul inputs^{(e.g. 가격 (U+AC00 U+ACA9))}, which are more commonly used than normalized ones?

Should one normalize them into NFKD_{(Normalization Form: Compatibility Decomposition.)}^{(e.g. 가격 (U+1100 U+1161 U+1100 U+1167 U+11A8))} before processing, or refuse to accept them?

@wlzla000

1. User enters string.
2. Your app NFKD normalizes the string.
3. Your app splits the string on ' ' space (0x20) into an array.
4. Your app checks if each word is contained within a single wordlist you support.
5. If so, your app converts the indices of each word into binary to reassemble the data and verifies the checksum. If the checksum fails, throw a failure. If checksum passes, run the NKFD normalized phrase from step 2 through the pbkdf2 rounds.
6. If not, Ask your user "We could not verify the validity of this phrase. Would you like to use it anyways?" and if they say no. Throw a failure. If they say yes, run the NFKD normalized string from step 2 through the pbkdf2 rounds.

Since you will NFKD normalize the user's input, 한글 will become ㅎㅏㄴㄱㅡㄹ automatically.

That's the whole point of NFKD normalization. You do it on the backend. The user only interacts with the frontend.

Should Korean mnemonics be displayed with ideographic spaces (as per Japanese) or ASCII spaces (as per Chinese)?

There is no mention of this in the bip39 wordlists document.

Any other Korean language considerations that should be added to that document?

Should Korean mnemonics be displayed with ideographic spaces (as per Japanese) or ASCII spaces (as per Chinese)?

If nothing is specified, ASCII.

If you google any Korean website at all, they use ASCII spaces in their writing.

Japanese and Chinese don't use any spaces at all normally.