Tuesday, February 26, 2013

Grammar Rule: Beginning-Sound Rule [두음법칙]

Dear Korean,

Why is the Korean family name 노, as in 노태우 and 노무현, anglicized as "Roh"? It's both spelled and pronounced as "Noh" in Korean, and there's no reason it can't be anglicized as such in English (it's not like "Noh" is not a sound that's foreign to English).

Anonymous Coward

Basically, this happens as certain words go through two levels of transliteration--first from Chinese to Korean, then from Korean to English. Let's take a look at each step in turn.

First, the Chinese to Korean part. Korean language uses a great deal of Chinese-derived words, much like English uses a great deal of Latin-derived words. This is to be expected, given that Korea spent its entire history right next to the extremely influential Chinese civilization. But by accident of history, Korean language and Chinese language belong to two different "families"--Chinese language is Sino-Tibetan, while Korean language is Altaic. This means that Korean language actually has a vastly different grammatical style from the Chinese language. 

Because of the grammatical differences between Korean and Chinese, Chinese words go through certain modifications as they are incorporated into Korean. One of the modifications is called the Beginning-Sound Rule [두음법칙]. (Please note that this is the Korean's own translation and not the official one.) Altaic grammar tends to avoid beginning a word with "n" and "r/l" sounds in certain situations. But Chinese language has tons of words that begin with "n" and "r/l" sound. When those words are imported into Korean, they are modified according to the BSR.

If you can read Korean, you can read the official explanation of the BSR at the website of the National Institute of Korean Language, the ultimate authority on Korean grammar. Here is a quick summary of the rules:
(1) The "n" sound rule:  If a Sino-Korean word begins with 녀 [nyeo], 뇨 [nyo], 느 [neu], 니 [ni], those sounds are converted to 여 [yeo], 요 [yo], 으 [eu], 이 [yi]. 


- Korean word "woman" is a Sino-Korean word, spelled 女子 in Chinese. Read as it stands, 女子 should be written and pronounced as 녀자 [nyeoja]. But because the word begins with 녀, the beginning sound is converted to 여. Therefore, Korean word for "woman" is 여자 [yeoja].

-Similarly, Korean word for "pseudonym" is a Sino-Korean word spelled 匿名. This should be written and pronounced as 닉명 [nikmyeong], if the word is to be read as it stands. But because the word begins with 닉, the beginning sound is converted to 익. Therefore, Korean word for "pseudonym" is 익명 [ikmyeong].

(2) The "r/l" sound rule 1:  If a Sino-Korean word begins with 랴 [lya/rya], 려 [lyeo/ryeo], 례 [lye/rye], 료 [lyo/ryo], 류 [lyu/ryu], 리 [li/ri], those sounds are converted to 야 [ya], 여 [yeo], 예 [ye], 요 [yo], 유 [yu], 이 [yi].


- Korean word "manners" is a Sino-Korean word, spelled 禮儀. This should be written and pronounced as 례절 [lyejeol], but 례 is converted to 예 under this rule, making the correct word 예절 [yejeol].

- A very common Korean last name is 李, which should be written and pronounced as 리 [li]. But because of the rule, 리 is converted to 이. Therefore, although outgoing president's name should be strictly read as Lee Myeong-bak [리명박], Koreans pronounce his name as Yi Myeong-bak [이명박].

(3) The "r/l" sound rule 2:  If a Sino-Korean word begins with 라 [la/ra], 래 [lae/rae], 로 [lo/ro], 뢰 [loe/roe], 루 [lu/ru], 르 [leu/reu], those sounds are converted to 나 [na], 내 [nae], 노 [no], 뇌 [noe], 누 [nu], 느 [neu].


- Korean word for "paradise" is spelled in Chinese as 樂園, which should be read and written as 락원 [lakwon]. But the beginning 락 sound is converted into 낙, making the correct Korean word 낙원 [nakwon].

- And now, the mysterious last name of Roh Tae-woo and Roh Moo-hyun. In both cases, the last name is spelled in Chinese as 盧, pronounced 로 [lo]. But because 로 cannot start a word, the word is converted to 노. Therefore, the name is 노무현 [No Mu-hyeon], although the Chinese spelling reads as 로무현 [Lo Mu-hyeon].
Think these rules are arbitrary and without logic? You are not alone. Because these rules are completely based on language experience, there is little logic to be found in the BSR. (But then again, the same is the case for a lot of grammar rules in any language.) Because this rule is so arbitrary, there actually was a significant debate within Korean language scholars as to whether BSR should be continued in modern Korean language. 

When Korea split into North and South Korea, the linguists of North Korea and South Korea came to opposite conclusions: North Korea scrapped the BSR, while South Korea left it alone. (This is partly a function of regional dialects, as the BSR tendencies were stronger in southern Korean dialects.) Thus, Sino-Korean words that begin with "r/l", for example, are written as they sound in North Korea. Thus, North Korea's state newspaper, 勞動新聞 ["Worker's Daily"], is written in North Korea as 로동신문 [rodong shinmun], rather than 노동신문 [nodong shinmun].

Now, the second step--going from Korean to English. If (South) Koreans spell 盧 as 노 [no] rather than 로 [ro], why do the former presidents transliterate their names as Roh Moo-hyeon, rather than Noh Moo-hyeon? Here, we are dealing with an exception in the Romanization rules. The Revised Romanization rules require that Korean words are to be Romanized as they are pronounced in Korean language. Therefore, the BSR-ed words are Romanized with their changes intact. (That is, the word 낙원 would be transliterated as "nakwon", not "lakwon".) 

However, the Revised Romanization rules provided an exception for people's names. The exception is simple--people may transliterate their names however they want. For historical figures who never had a reason to write their names in English, the Revised Romanization rules stand. (Thus, the famous admiral 이순신 is Yi Sun-shin, not Lee Sunshin.) But Koreans who had the time to consider how to transliterate their names into English do not really have to follow any rule. Thus, Korea's first president 이승만 (who studied and lived in the United States for a significant amount of time) chose a rather peculiar Romanization of "Syngman Rhee," although his name would be transliterated as "Yi Seung-man" under the Revised Rominization rule.

So, to sum up, why is it "Roh Moo-hyun" instead of "Noh Moo-hyun"? Because president Roh, when he decided to Romanize his name, decided to ditch Korean grammar rule that is the Beginning-Sound Rule. This is commonly done for Koreans whose last names fall under the BSR, i.e. Lee/Yee, Roh/Noh, Ra/Na, etc.

Got a question or a comment for the Korean? Email away at askakorean@gmail.com.


  1. great explanation! makes so much more sense now!

  2. Very minor nitpick - whether or not the Altaic language family exists is controversial, and (for those who believe that it exists) whether or not to include Korean in it is also disputed. Korean may be a language isolate.

    1. In my linguistic training, Korean was never definitively categorized as Altaic and in my preparation to move to Korea, I read an academic book, written by a Korean, on the history of Korean which also made it clear that we don't know if Korean is Altaic.

      In Korea, Korean is overwhelmingly categorized as Altaic. I live near the 전곡 prehistoric museum and they explicitly state (and have a map of the languages geographical history) that Korean is Altaic. I read a news article where they stated it plainly as well (of course, they also categorized Finnish as Altaic.. so.. bad journalism) I've tried to ask a few people about this, but no one I have met is anywhere near qualified to talk about it. Or even knows what I am talking about. It is a nitpick, but I have been wondering why this is the case.

  3. It should be noted that historically the /r/ and /n/ initials were actually pronounced that way and changed later on (I think during early 20th century). This pronunciation still remains at least to some extent in some dialects even in the South. For example, at least in my experience, those who speak the Gyeongsangdo dialect pronounce 六 as 륙 (ryuk), not 육 (yuk).

  4. Thanks for this post!
    --from a "Lee"

  5. Thanks for the explanation. I always thought it was weird that romanization favored North Korean pronunciation, but it makes sense to me now. Personally, I think the beginning sound rule makes Korean easier to pronounce, but not using it for romanized names makes them seem more familiar as names in English. The name "Yi" just doesn't sound like a family name in English (at least to me), while "Lee" does. And perhaps Mr. Roh saw the movie Dr. No and wanted to avoid the bad jokes :)

  6. Although it doesn't undermine the Korean's explanation to the original question, FarFromKorea makes an important point: despite decades of trying, the Altaic language family hypothesis has failed to be proven and few comparative linguists accept it anymore. That is to say, there is no established genetic connection between the 'Altaic' languages groups (Turkic, Mongolic, Tungusic and possibly Koreanic and Japonic), meaning they were not descended from a common proto-Altaic ancestor language. However, likely due to areal contact (centered around southern Manchuria-northern Korea), they do share similar morphological (structural) features and so it is possible to talk of an 'Altaic typology', but this is not equivalent to being an established genetic language family such as Sino-Tibetan.

    Having been introduced by Sin Chaeho (1880-1936) and Choe Namseon 1890-1957), the Altaic hypothesis remains popularly accepted in South Korea largely because it was included in school textbooks from the 1950s onwards. It has also been adopted by popular historians (재야사학팍), none of whom are comparative linguists. To the Korean and any other Korean language reader, I would strongly recommend Song Ki-Joong's (송기중) 역사비교언어학과국어계통론 (Historical comparative linguistics and theories on the genealogy of the Korean language, 2004).

    1. Thumbs up for answering a question about Korean culture that I have had!

    2. Largely correct, excepting the statement "...Meaning they were not descended from a common proto-Altaic ancestory language." This negative claim, of non-relationship, is almost impossible to prove between any languages at all, except within a certain time-frame, let alone one like Altaic Korean, with a small body of hypothetical, debatable protos forms identified. While non-genetic areal influence is quite likely, not just for typology but a set of apparent grammatical borrowings (or, possibly, inheritances), this does not preclude an earlier genetic relationship. What we must say instead is something like, "Despite suggestions of a proto-Altaic ancestor of Korean from archaeology, geography, and linguistics, it may or may not have existed, and there are other possibilities; we cannot currently show, at least to the usual standards of better attested language groups, that there was a clear genetic relationship, and if there was, we can only speculate on how long ago it would have been, to have become so eroded". The linguists who reject Altaicism are really only rejecting that it has been indubitably demonstrated from the provided evidence, not that it is possible or even likely as a stand-alone hypothesis.

  7. Thanks for answering this question. I really wanted to ask it, but was too cowardly since it didn't have anything to do with the topic of presidential power.

  8. Simplified and very informative.
    Thank you for this explanation about transcribtions.

  9. TK, for someone who gets infuriated at the ignorant legal views of people with no legal training, I'd say you sure have a lot to learn about linguistics. This post has quite a few striking errors, some perhaps a matter of technical terminology but others that suggest serious confusion. I'll try to take it point-by-point.

    (1) this happens as certain words go through two levels of transliteration--first from Chinese to Korean, then from Korean to English

    The first change you refer to is not mere "transliteration" (representing a word in one language with a script not primarily used for that language); rather, it is lexical borrowing from one language into another. The difference between 습근평 and 시진핑 is that the former uses Chinese > Korean borrowed morphemes (which have subsequently undergone divergent sound changes in both languages), while the latter is merely transliterated into Hangeul. This distinction is perhaps less obvious because of the unusual 1-to-1 relation of script to language in modern Korea; since most scripts are used for many languages and many languages regularly use multiple scripts, transliteration can happen within a language and cross-linguistic borrowing often requires no transliteration. (Of course, even today many languages are mostly not written, but borrowing and phonetic change take place just the same – language does not equal literacy.)

  10. (2) Think these rules are arbitrary and without logic? You are not alone. Because these rules are completely based on language experience, there is little logic to be found in the BSR. 

    It's neither arbitrary nor surprising to anyone trained in phonology. It may be a bit hard for an introductory linguistics course, but students in an upper division phonology course would surely be expected to handle it, with no prior knowledge of Korean required.

    What's going on in n-deletion is that alveolar consonants in (Seoul) Korean are heavily palatalized before high front vowels (so 시 sounds like she rather than see to English speakers, 같이 sounds like 가치, etc.). Palatal nasals that are not between vowels are hard to articulate (keeping the middle of the tongue high while the velum is open). Vietnamese is quite unusual in its frequency of word-initial palatal nasals – a more common pattern is seen in Spanish, with a tiny handful of mostly rare words with the initial tilde-n. A student should be able to recognize the phonetic environment right away (before a high front vowel), and then given the information about Korean palatalization, the explanation of easing articulation ought to at least occur to them.

    Next, we notice that the deletion doesn't occur in Pyeongan Korean. Given the hypothesis, what should we look for?

    Bingo, Pyeongan Korean doesn't palatalize to nearly the same extent. 시 sounds a lot more like “see” than in the South, 덩거당 instead of 정거장, etc. This same phenomenon explains why the basis for n-deletion didn't exist there.

    ㄹ- to ㄴ- seems to be about the lack of initial ㄹ- at the time of borrowing, and ㄴ- being the closest sound, although I'm not sure to what extent the North Korean standard ㄹ- is natural or an artificial contrivance since 1960. In any case, though,

    (3) Altaic grammar tends to avoid beginning a word with "n" and "r/l" sounds in certain situations.

    Phonology is not normally referred to as “grammar”, but in any case, Turkish, Mongolian and Japanese all have lots of native words that start with /r/, so I'm not sure where this claim comes from. “Certain situations” is way too vague to be useful – English also avoids beginning a word with these sounds in certain situations, such as before a [k]. ;-) With the speed of sound change and the minimal possible age for a common ancestral tongue, it would be extraordinary to find any good generalizations about word-initial consonants across such a broad grouping as Greater Altaic.

    Beginning-Sound Rule [두음법칙]

    It's usually called the “Initial Law” in official government translations, although linguists are more likely to talk about “n-deletion” or “initial sonorant variation” or some such.

    1. Thanks! Yes, I do have much to learn about linguistics.

  11. Long-winded bullshit. The reason that people with the last name "Noh" spell it as "Roh" or "Rho" is very simple. It is awkward to be called President "No" or Mr. "No." Hence, most people with that name spell it with an "R" instead to avoid such awkwardness. That's the real explanation, in 2 sentences.

    1. That's not the real explanation, but go ahead and believe that if you want.


Comments are not available on posts older than 60 days.

Related Posts Plugin for WordPress, Blogger...