Looking at Korean transcription rules for Japanese, I noticed that for some consonants, the hangul transcription would change if it was in the initial position:


カ /ka/ is transcribed 가 in the inital position, but 카 in medial of final positions. So the word カタカナ is transcribed 가타카나. ガ /ga/ is always transcribed as 가.

The same thing happens with /t/ and /tɕ/, so: タ /ta/ is 다 initally, but 타 medially and finally. So たれ (tare, "sauce") and だれ (dare, "who") are both transcribed 다레. The same happens with チ, 지 initially and 치 in other positions. The Korean Wikipedia article for ちょんまげ "chonmage", features both 촌마게 and 존마게 as a transcription.

Curiously, it does not happen with /p/, that is always transcribed as ㅍ.

This does not happen in the transcription rules for other languages, where seens to be /k/,/t/, tɕ/tʃ are consistently written as ㅋ,ㅌ,ㅊ even in the inital position:

English: cat 캣, time 타임, chart 차트.

Spanish: colcren 콜크렌, chicharra 치차라, tetraetro 테트라에트로.

Italian: Como 코모, Torino 토리노.

Portuguese: Cabral 카브랄, Coelho 코엘류, Tavira 타비라

So why Korean transcriptions of Japanese words uses the letters ㄱ,ㄷ,ㅈ for initial /k/, /t/, /tɕ/ while using ㅋ,ㅌ,ㅊ for other languages? What are the phonological or historical reasons for this different treatment when transcribing Japanese to Hangul specifically?

Other links:




Edit: I think @jogloran explained well why ㄱ,ㄷ,ㅈ are used for the Japanese initial unvoiced stops and the article explains these and other transcription conventions in depth. The use of ㅋ,ㅌ,ㅊ for English initial unvoiced stops, that are more aspirated than the Japanese ones, is clear to me now too. But the question remains as why other languages that have initial stops with even less aspiration than their Japanese counterparts, such as Portuguese, use ㅋ,ㅌ,ㅊ in their official hangulizations.

  • English initial stops are more aspirated than the Japanese ones, so ㅋ,ㅌ for initial stops does make sense. AFAIK, Italian and Spanish initial stops are not more strongly aspirated than Japanese, and in Portuguese I am sure they aren't (native speaker). So why not ㄱ,ㄷ? I do not think it is just an influence of English transcription rules, as in the Portuguese words hangulization, the jamo chosen to represent r, t, x, ch, lh, nh, rr make sense, including when the sounds change because of position inside the word. So it's clear to me that the transcription rules are phonetically informed.
    – kanazoshi
    Apr 24, 2021 at 5:03
  • Honestly, being korean myself, although I've already known this japanese translation rule for first letter, I really think this rule makes no sense. Furthermore, this rule is abided by only official or professionals. Just normal korean people usually even don't know and don't follow this rule when referring to japanese words. Of course you can ask why professionals and officials made the rule. There must be some reasoning in terms of academics, but I mean, the reasoning seems trivial or not legit to many average korean. Or is this forum for asking for that pedantic reasoning? Then I'm sorry
    – Arbitrary
    Apr 24, 2021 at 10:54

3 Answers 3


As you may know, "single" stops in Korean are weakly aspirated in initial position only (audio example), so Japanese stops in the unvoiced series (such as た) correspond to Korean "single" stops (e.g. ㄷ) in initial position. They are not perceived as ㅌ because the aspirated series in initial position in Korean is much more strongly aspirated than in Japanese.

Section 5 of Ito et al. was an attempt to explain the second part of your question -- why loanwords from other languages do utilise the aspirated series to correspond to aspirated/non-aspirated distinctions in those languages. The paper mentions French and English as languages where the mapping of aspirated stops seems to differ to that of Japanese. The key observation on p. 93 was:

Japanese voiceless stops have a VOT that is too brief to be identified with a Korean aspirated stop

They argued that even though this left Korean unable to distinguish the aspirated from the unaspirated series in initial position in Japanese loanwords, the perceptual match was large enough for Koreans that they would both map onto the Korean "single" stop series.


Ito et al. (2006): The adaptation of Japanese loanwords into Korean. Studies in Loanword Phonology (MITWPL 52).

  • 3
    But in Italian and Spanish for example, which according to @kanazoshi are transcribed with "double" stops in all positions, unvoiced plosives are basically not aspirated at all... or at least, they're more weakly aspirated than in Japanese (especially thinking about か).
    – LjL
    Apr 24, 2021 at 0:12
  • 2
    I added a reference that offers an explanation for why Japanese might be distinguished from loanwords from other languages.
    – jogloran
    Apr 24, 2021 at 0:13
  • 1
    I guess aside from English (with its definitely voiced stops), this hypothesis on French is relevant: "Japanese is a pitch accent system where lexical contrasts are signaled by F0 contour while F0 plays no such role in French. Given the difference in the status of F0 in the two languages, there may be a difference in the extent to which the consonant voicing-induced F0 difference persists into the syllable—namely, the consonantal effect on F0 may be more extensive in French than Japanese, making the F0 raising following French voiceless stops more extensive than that in Japanese."
    – LjL
    Apr 24, 2021 at 0:31
  • 1
    There is still a question (not directly addressed by the paper) as to why the labial series might constitute an exception. The data in the paper suggests that [p] in Japanese has an even shorter VOT than the other two places of articulation, so it’s not clear to me either why it might consistently map to an aspirated stop in Korean. Presumably some other factor is needed to explain this…
    – jogloran
    Apr 24, 2021 at 0:38
  • 1
    [p] is relatively uncommon in Japanese, ocurring mostly medially and with sokuon (e.g. にっぽん Nippon). Initial [p] is rare, found mostly in onomatopoeia and in western loanwords. This is pointed out in the article at page 24. Perhaps it's consistent mapping to ㅍ might have some relation with that.
    – kanazoshi
    Apr 24, 2021 at 3:02

jogloran's answer is a good explanation on why it's possible to transcribe word-initial /k/, /t/ to Korean ㄱ and ㄷ. As for why it had to be that way, there's no logical answer - IMHO the other way (always transcribing /k/ to ㅋ, for example) makes just as much sense, but when the rule was decided, apparently the scholars liked the current scheme better.

If we had consulted a different group of scholars, the decision could have gone the other way, but once these decisions were made, millions of books would start to use the official scheme - just by being official, it becomes harder and harder to change.

That also partly explains your question about "Why not Spanish or Italian?" Obviously, the transcription rules for Spanish were made by experts on Spanish (or at least I hope so) - they were a totally different group of people, and there would be no reason to expect them to talk to Japanese experts to keep things consistent.

In fact, the official Japanese transcription rule is still controversial and not universally accepted. You can easily see it in the Korean fandom of Japanese popular culture (games or anime), where people frequently use ad-hoc schemes that are more faithful to the original Japanese sound system. For example, here's a Korean wiki site listing Japanese animation directors: you can see many names starting with ㅋ or ㅌ, which would be written with ㄱ/ㄷ under the official scheme.

  • 1
    BTW, if that looks chaotic, you should take a look at the Chinese transcription rules! People/Place names before the 1911 Xinhai Revolution are written with traditional reading of Chinese characters, while more modern names follow modern Mandarin sound. So Beijing becomes 북경 when talking about Qing dynasty, but 베이징 when talking about Mao.
    – jick
    Apr 24, 2021 at 4:46
  • Thanks for your answer. I don't think the perceptual mapping was done "by scholars" as you propose -- after all, nobody compiled any official way of mapping from Japanese loans to Korean phonotactics. Rather, this is mainly about the lines in native speakers' minds that map one phonology's sounds to another.
    – jogloran
    Apr 24, 2021 at 4:52
  • 2
    As another example, consider how the unvoiced interdental fricative [θ] from English is mapped onto different phonemes in different varieties of English -- from Indian English [tʰ] to HK English [f]. No authoritative body decided that it should be assigned one way or the other. Instead, the phonologies of the target languages, and native speakers' perceptual judgements, informed the mapping.
    – jogloran
    Apr 24, 2021 at 4:56
  • 3
    Well, these are two related but different things. Obviously the official rules must be grounded on native speakers' perception, but perceptions can vary while official rules are official rules - so, if you ask "Why is Tokyo written as 도쿄, not 도꾜, 토쿄 or 토오쿄오?" then the answer is "Official rules." You can get it straight out of the horse's mouth here in the web page of National Institute of Korean Language: korean.go.kr/front/page/… (Transcription Rules - Japanese).
    – jick
    Apr 24, 2021 at 5:04
  • Sure, perhaps the official Japanese hangulization scheme is older, and using initial ㅋ/ㅌ in Japanese has the advantage of not merging inital か and が, even if Koreans would have a hard time trying to distingish the two sounds in initial position. The same where there are different romanization schemes for Korean and Japanese, sure, why not more than one hangulization scheme? That said, do you know of transcriptions of Portuguese, Spanish, Italian words (or any other language with voiced/unvoiced contrast as Japanese) where initial stops are written as ㄱ/ㄷ instead of the "usual" ㅋ/ㅌ ?
    – kanazoshi
    Apr 24, 2021 at 5:33

For what it's worth, I think all comments made here to the effect that the Korean spelling of カタカナ as 가타카나 is due to VOT in either language are not correct or at least insufficient for modern speakers of Korean (it might have been different in the past). The indication for this is that not only does the Korean Wikipedia page for Tokyo say

도쿄도(일본어: 東京都 발음: '토쿄토')

(i.e. effectively "Tokyo is written in Korean as 도쿄도 and pronounced as '토쿄토'), but also the page about Katakana has consistently transcribed カ as 카 and ガ as 가. So writing 가 for カ is a historical artifact rather than what people would reach for if they had to devise a transcription today.

By the way the surprising Korean spelling may be compared to what is done in German; when you visit the German page about Tokyo you will find "Tokio (auch Tokyo, japanisch 東京 Tōkyō [to̞ːkjo̞ː])" next to spellings like "Tokugawa Ieyasu" and "Kyōto". The last one would be highly unusual for newspapers (Wikipedia opines "Kyōto, im Deutschen meist Kyoto, seltener auch Kioto geschrieben", but in my experience both Kioto and Kyoto can be frequently found). The fact is that a lot of times, German articles about Japan will use a different spelling for the new and the old capital but prefer the international / English spellings for all lesser place names (the same being true for Pyongyang, which is rendered as Pjöngjang, retaining an older spelling, which is add odds with the usage for all other Korean place names).

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.