Hangul as alphasyllabary
« previous post | next post »
After visiting the massive National Museum of Korea in Seoul, I was eager to go to the National Hangeul Museum nearby. Alas, it is under renovation, so I was unable to enter it this time, but I will go back on some future occasion when I travel to Korea. I did, however, manage to buy two facsimile versions of the Hunminjeongeum 훈민정음 / 訓民正音 ("The Correct / Proper Sounds for the Instruction of the People), a 15th-century manuscript that introduced the Korean script Hangul, one for the populace and one for the literati.
Several of the comments to this post, "How to say 'Seoul'" (5/12/25), prompted me to think some more about a problem that had perplexed me from the time I did a review of The Korean Buddhist Canon: A Descriptive Catalogue, by Lewis R. Lancaster, in collaboration with Sung-bae Park (Berkeley and Los Angeles: University of California Press, 1979). That was nearly half a century ago, but I still remember keenly how difficult it was to romanize the titles and the proper nouns. The hardest part of that was dealing with what happened at syllable boundaries. It was obvious that different authorities romanized the sounds in discrepant ways.
As I wrestled with that large tome (724 pages) having more than a thousand detailed bibliographical entries, I grew increasingly frustrated and exasperated at not being able to get clear-cut answers about the romanization even from specialists on Korean Buddhism and language. Somehow, I managed to get through the task, which took the better part of a month, but was not completely satisfied with the results.
All of this fits with the conception of Hangul as an alphasyllabary, in that it is neither an alphabet nor a syllabary, but somewhere in between. How did that happen?
To the extent that it possesses vowels and consonants, Hangul had / has the potential to develop into a true alphabet, but the desire to make Hangul compatible with Sinitic tetraglyphs caused its creators to squeeze the letters of the Hangul alphabet into square-shaped blocks, like the strokes / components of sinographs, which they are not.
This quadratic imperative of Hangul tends to emphasize the syllable in Korean phonology, and one can hear that when spoken at normal or slow speed. But when speech is rapid, the syllable boundaries tend to get slurred or blended. This happens even with sequences of sinographs, which I have often written about on Language Log (see, for example, the long series of posts on "When intonation overrides tone", but there are many others).
In one or two forthcoming posts, I will give specific examples of such blurring / blending at syllable boundaries of spoken Korean phrases. In each case, they confused me greatly because I was not able quickly to disentangle the constituent phonemes. Indeed, they often disappeared (got swallowed up by the resultant whole).
Selected readings
- "Grids galore" (11/19/23)
- "Hangul: Joseon subservience to Ming China" (5/14/22)
- "Hangul for Cantonese" (11/18/24)
- Coblin, W. South (2006). A Handbook of ʼPhags-pa Chinese. ABC Dictionary Series. Honolulu: University of Hawai'i Press. ISBN 978-0-8248-3000-7 — 'Phags-pa played a key role in the creation of Hangul
Philip Taylor said,
May 14, 2025 @ 6:37 am
Could I trouble you to explain "the Hunminjeongeum 훈민정음 / 訓民正音 ("The Correct / Proper Sounds for the Instruction of the People), a 15th-century manuscript that introduced the Korean script Hangul, one for the populace and one for the literati", Victor ? To what does "one for the populace and one for the literati" apply ? To the manuscript, to the script, to "The Correct Proper Sounds", or to something else ?
Victor Mair said,
May 14, 2025 @ 6:41 am
"two facsimile versions"
Philip Taylor said,
May 14, 2025 @ 7:14 am
Aha — how did I miss that at a first reading ?!
Jongseong Park said,
May 14, 2025 @ 8:38 am
The romanization of Korean is a huge and complex topic, and I think the official guidelines leave a lot to be desired. I'm not talking about the choice of how to write individual sounds (even if not everyone is happy with it, that ship has sailed). I mean things like word division and how to mark ambiguous syllable boundaries.
The thirteenth-century tome 삼국유사(三國遺事), or Memorabilia of the Three Kingdoms, should be written as Samgungnyusa according to the official guidelines. I can guarantee that no one in Korean studies is using that romanization. Everyone treats it as two words, writing Samguk Yusa or Samguk yusa depending on whether to capitalize the second element. The detailed "Korean Romanization and Word Division" document from the Library of Congress would favour the latter, as it capitalizes only the first word in titles of books.
One only has to compare the 66-page guidelines from the Library of Congress to South Korea's official romanization guidelines which would take up a couple of pages at most in large print to see that the latter barely gives any consideration to the issue of word division. But this is a huge issue because Korean tends to be more flexible with word spacing than languages such as English. Spaces between words that make up a compound are optional, and in practice you end up with long compounds that would be unwieldy to romanize as single words.
Korean also has a number of phonological processes that work across syllable boundaries, so the word divisions matter quite a bit as you can see from the example of Samgungnyusa vs Samguk yusa.
I myself tend to combine the official romanization with the Library of Congress word division rules, roughly speaking. I also prefer to use the apostrophe to disambiguate syllable boundaries, and think that they should always be used when a sequence is created that could be interpreted as a single sound. For example, 가을 should be written ga'eul not to be confused with 개울 gaeul. The official rules use hyphens instead and only on an optional basis. I prefer to reserve hyphens for other uses such as separating particles.
Philip Taylor said,
May 14, 2025 @ 1:26 pm
JP — "I also prefer to use the apostrophe to disambiguate syllable boundaries, and think that they should always be used when a sequence is created that could be interpreted as a single sound" I also prefer to use the apostrophe to disambiguate syllable boundaries, and think that they should always be used when a sequence is created that could be interpreted as a single sound" — is the diaresis never used for this purpose in romanized Korean, as in, for example, French Citroën or Noël ?
David Marjanović said,
May 14, 2025 @ 3:05 pm
Maybe as part of the quadratic imperative, the modern orthography of Korean is also quite strictly morphemic, so that a lot of morphophonemic processes are left as an exercise for the reader.
I've never seen that; I suppose the model for the romanization is too strictly English and/or the goal is to stick with ASCII without any diacritics whatsoever (as opposed to the main other romanization, which uses ŏ, ŭ – Sŏul – instead of eo, eu).
murawaki said,
May 14, 2025 @ 6:48 pm
When people discuss the influence of the 'Phags-pa script on the creation of Hangul, they often focus exclusively on its use for writing Chinese. However, I suspect its application to Mongolian is also relevant, particularly with regard to syllable composition.
For example, the phrase "of soldiers" is written as "chee ri -u dun," where the syllable boundaries do not align with morpheme boundaries:
čerig -ud -un
soldier -PL -GEN
I wonder if this approach to syllable division in Mongolian may have influenced a similar treatment in the typologically similar Korean.
Mehmet Oguz Derin said,
May 14, 2025 @ 7:02 pm
The National Hangeul Museum had a fascinating presentation, esp. with illuminated blocks, when I went there in 2024 (the print of the facsimile is a great souvenir to this day). Though I was hoping to see more about origins, such as block-layout similarity with the Khitan small script probably being more than a coincidence (and also the grapheme similarity with Brahmi, as mentioned in the post through Phags-pa); nevertheless, I think the museum was a great celebration of Hangeul as it is a particularly space-efficient script once familiarized and unique in persisting to our date.
I was sad to read the news about the fire; I hope they can reopen more strongly.
Victor Mair said,
May 14, 2025 @ 7:45 pm
"Fire breaks out at Hangeul Museum", by Lim Jae-seong
The Korea Herald
Published : Feb. 1, 2025 – 12:49:34 Updated : Feb. 1, 2025 – 13:40:10
https://www.koreaherald.com/article/10409700
Jongseong Park said,
May 14, 2025 @ 11:03 pm
@Philip Taylor, I suspect that using diaereses may lead to some confusion if we write 네온 ne'on and 예우 ye'u as neön and yeü for example, as they be mistaken for umlauts. Also, where would you put a diaeresis in a case like 예외 ye'oe? If you write it as yeöe, it wouldn't be clear at all that the öe represents a single vowel (albeit one that most Korean speakers today pronounce identically to 웨 we).
Syllable division in Hangul orthography in case of enchaînement has been an issue since the creation of the Korean alphabet. Most early Hangul texts carry over the final consonant to the next syllable, as in 말〯ᄊᆞ미〮 mǎl-ssò-mí for 말〯ᄊᆞᆷ mǎl-ssòm "language" plus the subject marker 이 i. This example is seen in the Hunminjeongeum Eonhae (훈민정음 언해, 訓民正音諺解), the Korean version of the preface of the document announcing the new alphabet itself. This practice is called 이어 적기 ieo jeokgi or 연철(連綴) yeoncheol ("linking").
In today's orthography, we write 말씀이 mal-sseum-i instead of 말쓰미 mal-sseu-mi to divide 말씀 mal-sseum from the subject marker 이 i. This practice is called 끊어 적기 kkeuneo jeokgi or 분철(分綴) buncheol ("separating"). In Early Modern Korean, you even saw the consonant written in both syllables, e.g. 말ᄊᆞᆷ미 mal-ssom-mi – 거듭 적기 geodeup jeokgi or 중철(重綴) jungcheol ("doubling").
However, in the two literary works directly linked to the creator of the alphabet, King Sejong – Worin cheon'gangjigok (월인천강지곡, 月印千江之曲), composed by the king himself, and Yongbieocheonga 용비어천가(龍飛御天歌), written under his supervision – one sees a preference for "separating" as in modern orthography. These works also differ from other early Hangul texts in the spelling of final consonants, showing that Sejong preferred a morphophonemic approach to spelling. "Sejong's theory of literacy and writing" by Young-Key Kim-Reynaud (2000) goes into this in English if you'd like to read more.
So it seems that the creator of Hangul intended a more morphophonemic syllable division, but this was ignored by most users in the first couple of centuries of the alphabet. Then after a transitional period, morphophonemic syllable division became more common in Modern Korean and was finally codified with the standardization of Korean orthography.
Philip Taylor said,
May 15, 2025 @ 3:00 am
JP — understood about the possible confusion with umlauts, particular as both ablaut and umlaut phenomena are attested in Korean (or so a quick Google search tells me). As regards 예외 ye'oe, yëoe would appear to be a possibility …
Peter Cyrus said,
May 15, 2025 @ 3:58 am
Please forgive me if this is too detailed a question, but as I understand it, there are two different cases of ambiguous syllable boundaries. The easy one is caused by the lack of a romanization for the null initial ㅇ, which some say is phonetically [ɰ]. An apostrophe (or a hyphen) would solve that problem, even when there's no ambiguity.
The harder case is when a final [m n ŋ p t k l] is followed by a syllable beginning with a glide [y w ɰ]. Hangeul writes ㅇ, because the glide is written as part of the vowel, but phonetically the glide is the initial, isn't it? These cases are ambiguous with those where a syllable with no final is followed by [m n ŋ p t k l] plus a glide. But I'm told that the spoken language tends to "move" the final to the following syllable in the first case.
If that's true, then is there still ambiguity?
Jongseong Park said,
May 15, 2025 @ 5:36 am
@Peter Cyrus: The orthographic null initial ㅇ is definitely silent in Modern Korean. It might occasionally get a non-phonemic hard attack as the glottal stop [ʔ], but this only occurs word-initially.
However, many scholars think initial ㅇ once had a sound value in Middle Korean, at least in some instances. The late Ki-Moon Lee proposed [ɦ], and something like [ɰ] could very well have been possible.
Korean has enchaînement, meaning that a coda consonant becomes the initial consonant of the following null-initial syllable, and this also applies when the nucleus of the latter contains a glide.
So you have 참여 cham-yeo [ʦʲʰa.mjʌ], 학원 hag-won [ha.ɡwʌn], and 합의 hab-ui [ha.b(ɰ)i] for example, with the /m/, /ɡ/, and /b/ phonetically becoming the initial consonant of the following syllable.
숙영 sug-yeong and 수경 su-gyeong are thus pronounced identically as [z̥ʰu.ɡjʌŋ], just as 목이 mog-i and 모기 mo-gi have the identical pronunciation [mo.ɡi] (ignoring phonemic vowel length, which is lost in most Korean speakers today).
Please note that [ŋ] only occurs as a coda in Korean, so it does not participate in enchaînement (e.g. 공연 gong-yeon [ɡ̊oŋ.jʌn], 병원 byeong-won [b̥jʌŋ.wʌn], 정의 jeong-ui [ʣ̥ʲʌŋ.(ɰ)i]).
Chris Button said,
May 15, 2025 @ 6:10 am
It reminds me a lot of the Late Middle Chinese ɰ- onset (generally just left un-noted) that came out of Early Middle Chinese w-.
Victor Mair said,
May 15, 2025 @ 8:56 am
From South Coblin*:
I hesitate to say anything about Mongolian ‘Phags-pa, since I know virtually nothing about Mongolian. When the Koreans were designing the Hangul system, they often talked about the Chinese ‘Phags-pa system, using traditional Chinese terminology for the initial classes, etc., as those are applied and represented in the Hongwu zhengyun. Quite understandably, this is frequently the case when they discussed the use of of the specially modified Hangul system used to transcribe Chinese. However, exactly how they arrived at the original or basic Hangul system as applied specifically to Korean, and whether or not they drew directly on Mongolian during their work, is a matter for the experts in Korean and Mongolian to determine. I think they are the ones you should consult here. For Korean, I would recommend that you consult Bob Ramsey, whom I suspect you know well. For Mongolian I am uncertain. If Jerry Norman were still with us, he would be the one to ask; but, alas, that is no longer an option. And his friend Jim Bosson, who was also an expert in this area, is gone now too. I’m afraid I can’t help you in this area. It is known that when the Koreans devised the Hangul system, they first consulted and considered all the foreign alphabets they knew about, which of course included Mongolian. So they surely must have given some thought to the way Mongolian was spelled. But what they actually did or did not do with Mongolian, I am not qualified to say.
———–
*Author of A Handbook of ʼPhags-pa Chinese. ABC Dictionary Series (Honolulu: University of Hawai'i Press, 2006).
Richard Futrell said,
May 15, 2025 @ 9:42 am
The Hangul museum is great, but when I went in 2022 they didn't actually have a copy of the Hunminjeongeum.
KWillets said,
May 15, 2025 @ 12:12 pm
Sorry you missed the museum — Ichondong is my wife's neighborhood, and I feel a bit of (irrational) embarrassment that you couldn't get in. If you make it back I recommend a few cafes nearby (Coffee Class is top notch), and a walk out to the Han is always worthwhile.
Penglin Wang said,
May 15, 2025 @ 3:36 pm
This is an interesting discussion. I appreciate Professor Mair’s great and fruitful effort to organize such discussions. But I have no expertise in the field of hPhags-pa script and orthography of classical Mongolian. What is reasonable in murawaki's post is that of a similar typology in Mongolian and Korean. I don't agree with their implication that the syllable boundaries align with morpheme boundaries. Syllable and morphemes are very different things. It is natural in phonographic scripts that a syllable division cannot converge with a morphemic boundary, e.g., the four-syllable word 'American' consists of two morphemes–'america' having four syllables and '-n' not standing as a syllable in pronunciation. From a cognitive perspective, a syllable may be perceived and even conceived by speakers, whereas not every speaker can recognize how many morphemes in words. I think that classical Mongolian syllable division is a continuation of its preceding Old Uyghur script and not affected by hPhags-pa.
Chris Button said,
May 15, 2025 @ 4:27 pm
I think it was Pulleyblank who first showed that 'Phags-pa showed the distinctions between all four grades in the rime tables for Late Middle Chinese too.