Korean Romanization

June 26, 2021 @ 5:25 am · Filed by under Alphabets, Romanization

« previous post | next post »

I can't think of another language in the world where the Romanization situation is more chaotic than it is for Korean.  There are seven schemes in common use: 

  • Revised Romanization of Korean (RR, also called South Korean or Ministry of Culture (MC) 2000): This is the most commonly used and widely accepted system of romanization for Korean. It includes rules both for transcription and for transliteration. South Korea now officially uses this system which was approved in 2000. Road signs and textbooks were required to follow these rules as soon as possible, at a cost estimated by the government to be at least US$500–600 million. Almost all road signs, names of railway and subway stations on line maps and signs etc. have been changed. Romanization of surnames and existing companies' names (e.g. Hyundai) has been left untouched; the government encourages using the new system for given names and new companies.
    RR is similar to MR [see next item], but uses neither diacritics nor apostrophes, which has helped it to gain widespread acceptance on the Internet. In cases of ambiguity, orthographic syllable boundaries may be indicated with a hyphen, although state institutions never seemed to make use of this option until recently. Hyphenation on street and address signs is used to separate proper names and numbers from their assigned function. As of 2014, under mandate from the Roadname Address Act, Korea Post officially changed the older address system from lot-based district subdivisions to a street-based system that regularly utilizes hyphenation in order to disambiguate. The Ministry of the Interior also provided the public with various service announcements and websites forewarning of the change toward a clear and complete signage system classifying all streets and individual addresses with romanization (of which hyphenation is a systematic part).
  • McCune–Reischauer (MR; 1937?): the first transcription to gain some acceptance. A slightly modified version of MR was the official system for Korean in South Korea from 1984 to 2000, and yet a different modification is still the official system in North Korea. MR uses breves, apostrophes and diereses, the latter two indicating orthographic syllable boundaries in cases that would otherwise be ambiguous.
    Several variants of MR, often also called "McCune's and Reischauer's", differ from the original mostly in whether word endings are separated from the stem by a space, by a hyphen or not at all; and if a hyphen or space is used, whether sound change is reflected in a stem's last and an ending's first consonant letter (e.g. pur-i vs. pul-i). Although mostly irrelevant when transcribing uninflected words, these variants are so widespread that any mention of "McCune–Reischauer romanization" may not necessarily refer to the original system as published in the 1930s. MR-based romanizations have been common in popular literature until 2000.

  • The ALA-LC / U.S. Library of Congress system is based on but deviates from MR. Unlike in MR, it addresses word division in seven pages of detail. Syllables of given names are always separated with a hyphen, which is expressly never done by MR. Sound changes are ignored more often than in MR. ALA-LC also distinguishes between ‘ and ’.
  • Yale (1942): This system has become the established standard romanization for Korean among linguists. Vowel length in old or dialectal pronunciation is indicated by a macron. In cases that would otherwise be ambiguous, orthographic syllable boundaries are indicated with a period. This system also indicates consonants that have disappeared from a word's South Korean orthography and standard pronunciation.
  • ISO/TR 11941 (1996): This actually is two different standards under one name: one for North Korea (DPRK) and the other for South Korea (ROK). The initial submission to the ISO was based heavily on Yale and was a joint effort between both states, but they could not agree on the final draft. A superficial comparison between the two is available here:
  • Lukoff romanization, developed 1945–47 for Fred Lukoff’s Spoken Korean coursebooks
  • Romanization of Korean (1992): the official romanization in North Korea.

(source)

Because kanji have multiple pronunciations, I have often bemoaned the difficulty of Romanizing Japanese words and names.  Even though contemporary Korean texts are written in Hangul / Hangeul / Han'gŭl alphabetic script, it is hard to Romanize them — not just because of the different schemes in existence, but also because of the different conventions for how to apply them.

Here are comments by Korean language specialists:

 

Ross King

Korean Studies is cursed when it comes to romanization.

The short answer is yes: McCune-Reischauer has traditionally been the preferred romanization. Unless the article is in linguistics, in which case Yale romanization is preferred.

MR is clumsy, is terrible for connected text, and ends up having lots of localized variants when it is actually deployed by different scholars and publications (e.g., some can't be bothered with the diacritics for the vowels ㅡ, ㅢ and ㅓ).

But the waters were muddied when the ROK government decided to 'revise' MR and came out with an aesthetically problematic new government system in 2000 or so (Gari Ledyard shared a brilliant critique of it on the Korean Studies listserve that I have never been able to find again) that can't decide whether it's a transliteration system or a transcription system, and eschews diacritics at the cost of digraphs for some vowels (eu, eo). Some scholars (mostly younger ones) have now started using it (and apparently some journals published with ROK government support now require the use of this romanization). I personally dislike the ROK government system, and see no reason to use a Korean government system until and unless the ROK and DPRK agree on a unified system. MR was (in theory) what South Koreans were using anyway, but they never taught it in school. The DPRK system is also quite close to MR.

One of the original motivations behind the new ROK government system was the claim by some Koreans that MR (which dates to the colonial period) did not have Korean input (and presumably resentment that it had two white guys' names on it), but there is good research showing that Korean linguists were heavily involved; rebranding would have been a perfectly good solution. The other claim was that MR did not "match Koreans' linguistic affect/emotions)", whatever that means.

I would also note that MR is what the Library of Congress and all other libraries use.

In any event, romanization is one of the curses of Korean Studies. (Funding and critical mass being two others).

 

Min Jung You

I agree with Professor Ross King's comments.

Though MR is clumsy, I prefer to use it, because it seems more scientific and I write English papers for non-Korean readers.

Some journals of philosophy, in the US and Europe, however, still seem to use romanization supported by the Korean government.

 

Bob Ramsey

I'm not sure I'm the person to ask this Romanization question; I see by sheer, energetic persistence the RR people seem to be having their way, but I'm not a fan. For one thing, the North Koreans are still using a modified version of McCune-Reischauer, and MR is also the system the Library of Congress is still using, I believe. I myself use MR in the body of texts, too, but for linguistic work I use Yale.

Yale is difficult for non-specialists to tolerate, because the conventions cut across the grain of the phonetic values (non-linguist) English speakers associate with the letters. But Yale most consistently mirrors Hangul spellings without the diacritics of MR, and really, Yale is the only Romanization system that works for representing Middle Korean transcriptions in a straightforward way. (I'm speaking here, of course, of the modified version of Yale seen, e.g., in Martin's A Reference Grammar of Korean and Professor Lee's and my Cambridge book, A History of the Korean Language.) For anyone trying to do historical linguistics, RR just complicates and obfuscates the historical processes.

I think you may have gotten the message that I find RR annoying and frustrating. Well, I confess that's true. The creation of RR is an example of South Koreans asserting a go-it-alone nationalist attitude eschewing input from any outsiders. (You may know that the model they looked up to and tried to emulate was the PRC's creation of Pinyin Romanization, done without pushy and smug Westerners telling them what to do, the Koreans thought!) For one thing, again, as I suspect you already know, the RR transcription of the Hangul vowel <ㅓ> is <eo>, a convention based on the spelling of the familiar name "Seoul." But the spelling <Seoul> was the creation of French missionaries, and since they were French speakers!, the syllable juncture occurred between Se- and -oul. The result is that the RR convention represents a kind of embarrassing ignorance. Never mind. Even though the mistake was pointed out to them, the RR team back in 2000 went on, by analogy, to represent the vowel <ㅡ> as <eu>!

And so, because of that second decision about vowel spellings, the RR team decided that 한글 should be spelled Hangeul–even though most of the world was already writing it as Hangul–without that superfluous <e>. Ridiculous. There's no Korean word *한굴 with which, in context, it might be confused! I mean, the authorities were already allowing for a couple of exceptions to their rules; e.g., the name of the South Korean currency is Won [not Weon!] and that cultural object most revered in Korea is spelled <kimchi> (and not gimchi!).  So… what's wrong with just leaving the spelling Hangul alone, too?

All in all, it seems to me that the problem of Korean Romanization is as much political and emotional — actually more so — as it is linguistic.

 

Selected readings

June 26, 2021 @ 5:25 am · Filed by under Alphabets, Romanization


15 Comments

  1. Min Jung You said,

    June 26, 2021 @ 9:15 am

    Glad to see my name. Thanks for your post. :)

  2. J.W. Brewer said,

    June 26, 2021 @ 9:41 am

    I find it difficult to believe that the situation with Korean is more chaotic than the situation with Arabic, as famously illustrated by the alleged 112 different extant Latin-alphabet renditions of the name of the former dictator of Libya. https://www.csmonitor.com/World/2011/0222/Gaddafi-Kadafi-Qaddafi-What-s-the-correct-spelling

    Maybe one relevant difference is that most Arabic-speaking countries are reasonably content to have romanization done on an ad hoc basis by different folks using different approaches in different contexts, rather than trying to put the authority of the government behind any single uniform approach?

    [(myl) See also this and this… As Lane Greene observes in the second of those links, Arabic has the special property of unusually strong diglossia, roughly as if Latin were still being used by media all over Europe, while ordinary life went ahead in French, Spanish, Italian, Portuguese, Romanian, etc.]

  3. F said,

    June 26, 2021 @ 10:14 am

    As a linguistically interested layperson, I appreciate RR because it tells me the actual pronunciation of a word in contexts where no one bothers with diacritics. And the "Seoul"-based convention is great because the first letter represents (un)roundedness and the second represents front/backness, which is also how several other digraphs work both in Korean and other languages. So who cares if it's historically inaccurate.

  4. Thomas said,

    June 26, 2021 @ 10:36 am

    Nowadays, a good romanization needs to be easy to type. In this sense RR is great. You can just type it down without any problems. It even beats Pinyin at this. With Pinyin, most people will just omit the tones because there is no way to type some of them in a sensible way on a regular keyboard. And to make things worse, without something like a German keyboard, the umlauts in words like nüren are impossible, too.

  5. KWillets said,

    June 26, 2021 @ 11:01 am

    I've often referred back to criticisms posted here about RR. The one that sticks out most in my mind is the use of Roman consonants with Korean voicing rules; there's no way for an untrained speaker to know that "Busan" is pronounced most closely to "Pusan".

    Even more experienced speakers tend to look at RR and change their pronunciation unconsciously.

  6. Xerîb said,

    June 26, 2021 @ 12:15 pm

    I wonder if someone has preserved Ledyard’s critique of RR or can find it preserved somewhere online. I am sure many LL readers would be delighted to read it.

  7. Ronnie said,

    June 26, 2021 @ 1:03 pm

    I was lucky enough to get a mild flu my first week in South Korea, which gave me the opportunity to really lock down my Hangul and largely ignore Korean romanization for the rest of my life. The only time it proved to be a problem was when I bought a copy of "A New Practical Primer of Literary Chinese" and found that the Korean readings of the lessons and hanzi were romanized in some form or another which was disappointing.

  8. Jongseong Park said,

    June 27, 2021 @ 2:13 am

    A chaotic coexistence of multiple romanization schemes is actually the norm for most modern languages not usually written in a Roman alphabet, whether it is Amharic, Arabic, Armenian, Bengali, Burmese, Cantonese, Georgian, Greek, Hebrew, Khmer, Lao, Persian, Russian, Tamil, or Ukrainian.

    Russian for example has at least half a dozen systems in common use, partly thanks to the ever changing rules for use in passports. In linguistic literature, a Czech-style system using letters such as č, š, and ž is popular. And this is without even going into the different romanization practices used for Russian in languages such as German and Spanish.

    You can find plenty of criticism of RR here, much of which I agree with and could add to, but I also have to point out that MR is non-intuitive for Korean speakers which was a major challenge for its widespread adoption in Korea even when the official system was MR-based.

    MR uses different letters for the stops and affricates based on when the sound represented is phonetically voiced or voiceless. This is clearly oriented towards speakers of languages such as English, who can hear that ⟨Busan⟩ sounds more like "Pusan" to them.

    But as this difference is merely allophonic in Korean, most Korean speakers can't hear it. Try explaining to Koreans who aren't trained in linguistics that the sound written ⟨b⟩ in ⟨Busan⟩ 부산 and ⟨sanbu⟩ 산부 are phonetically different and should be written ⟨Pusan⟩ and ⟨sanbu⟩ respectively in MR. You may as well ask the average English speaker to transcribe phonetically all the different allophones of ⟨t⟩ they are using.

    MR's insistence on transcribing the voice distinction leads to more complications. In MR, 안다 should be written ⟨anda⟩ when it means "know(s)" but ⟨anta⟩ when it means "to hug". So in order to use MR, Korean speakers are forced to pay attention to pronunciation differences not reflected in the orthography. RR simply ignores the fortition in the latter and writes ⟨anda⟩ in both cases, reflecting the Korean spelling.

    You can present tables of when to use different transcriptions, but it is simply not practical to expect laypeople to apply the transcriptions correctly. If you want a romanization system to be the basis for road signs and the names on passports in the entire country, you need something that is at least moderately intuitive for non-specialists. You don't want to have to consult experts every time you have to romanize something.

    My personal preference might have been closer to something like combining RR's approach on consonants with MR's approach on vowels, perhaps with the stipulation that ⟨eo, eu⟩ are acceptable alternatives to ⟨ŏ, ŭ⟩ in situations where diacritics are not desired, similar to how ⟨ae, oe, ue⟩ are acceptable alternatives to ⟨ä, ö, ü⟩ in German.

    @Bob Ramsey: I mean, the authorities were already allowing for a couple of exceptions to their rules; e.g., the name of the South Korean currency is Won [not Weon!]

    RR writes ㅝ as ⟨wo⟩, not ⟨weo⟩. So Won is the regular romanization according to RR, not an exception.

  9. KWillets said,

    June 27, 2021 @ 12:25 pm

    Would it be possible for RR to simply replace initial consonants with their unvoiced equivalents? There are only four cases (ㄱㄷㅂㅈ), and that would give us back spellings like Kim, Taegu, Pusan, Cheju, etc.

  10. J.W. Brewer said,

    June 27, 2021 @ 5:00 pm

    Jongseong Park's comment maybe provides a helpful focus on the problem that what makes one romanization system better than another may differ depending on the intended audience. If the primary intended audience (or userbase) is native speakers already fully literate in their own non-Latin script, what is most user-friendly for them may not be most user-friendly for foreign users who are habituated to using Latin script for their own language. That said, it seems a bit weird to focus on user-friendliness for native speakers of Korean who are fully literate in Hangul, because they are more or less by definition the subset of humanity with the least need to have Korean words/texts romanized for their benefit.

    And what is most user-friendly for foreigners may depend on their own L1. To follow up on the mention of Russian, that сове́т comes out as "soviet" in English but "sowjet" in German (not to mention "sobiet" in Basque and "sovyet" in Turkish etc etc) strikes me as a feature rather than a bug, and it would have been a shame if the CCCP regime had insisted on a single standard system to be used by everyone everywhere and had foreigners kow-tow to them.

  11. Jongseong Park said,

    June 27, 2021 @ 10:29 pm

    @KWillets: Would it be possible for RR to simply replace initial consonants with their unvoiced equivalents? There are only four cases (ㄱㄷㅂㅈ), and that would give us back spellings like Kim, Taegu, Pusan, Cheju, etc.

    That leads back to the problem of how to distinguish the lenis and aspirated series. You wouldn't be able to tell apart common elements like Dae and Tae, Dong and Tong, Bo and Po, Jeong and Cheong when they appear initially. It also complicates affixed and compound names. Daerim would be spelled Taerim, but Sindaerim ("New Daerim") would retain its spelling. The Song of Chunhyang could be Chunhyangga or Chunhyang ka depending on how you decide on the spacing.

    There are plenty of languages where the historically voiced letters represent voiceless stops (Danish, Icelandic), and this is also true for romanization schemes like Hanyu Pinyin and the two common scholarly romanizations for Cantonese. The difference may be that the Korean lenis stops and affricates are lightly aspirated, leading English speakers to identify them readily as equivalent to their own voiceless stops and affricates. They wouldn't hear Danish or Chinese b as "p" since they are unaspirated, but Korean b would sound very much like "p" to them (in contrast, French or Spanish speakers would likely hear Danish or Chinese b as "p"). But to insist on this single point for the benefit of English speakers would unnecessarily complicate the system.

    @J.W. Brewer: That said, it seems a bit weird to focus on user-friendliness for native speakers of Korean who are fully literate in Hangul, because they are more or less by definition the subset of humanity with the least need to have Korean words/texts romanized for their benefit.

    You may be underestimating the need that Koreans have for an intelligible romanization system, given that it would affect the spelling of their names on passports and their addresses for international deliveries. Everyone educated in South Korea learns the Roman alphabet and it is an indispensable part of the contemporary Korean language milieu. Why else do you think ordinary Korean speakers would care that much about how their language was romanized, if it was just meant for non-speakers of Korean? It is a question that affects them directly.

  12. J.W. Brewer said,

    June 28, 2021 @ 8:26 am

    Jongseong Park is correct that I may have overlooked or slighted the domestic ROK need for romanization, as a moment's reflection on how ubiquitous romaji was in the Tokyo of my 1970's childhood ought to have reminded me.

    On the other hand, Taiwan demonstrates that you can have ubiquitous romanization without having standardization, even domestically. As I understand it, for ROC passport etc purposes individuals can and do choose their own romanization and people with the "same" name (if written in Chinese characters) can and do use different romanizations. Especially gIven that different romanization systems are often perceived in the Taiwanese context (and ditto in the diaspora) to have political/partisan overtones, not forcing a uniform approach seems especially sensible for personal names on official documents.

    When I was visiting Taiwan a few years ago the national gov't had recently thrown its weight behind a different romanization scheme than the one previously favored and street signs etc. had been updated to reflect the change in some parts of the country but not others (perhaps depending on which political party was in charge locally?). If I recall correctly, there were some subway stations in Taipei whose name clearly matched that of a street but the station-name romanization was sometimes slightly different from the street-sign romanization because different bureaucracies were in charge. And of course private-sector signage reflected the ad hoc and varying preferences of whoever put up the sign. And somehow the chaotic variation did not lead to unacceptably dysfunctional side effects in practice.

    For purposes of getting international mail, I am a big fan of the old Chinese postal map approach which was simply a large but not infinite set of toponym-specific romanizations that did not purport to follow a single consistent set of rules and thus could not be mechanically applied to romanize any and all Chinese texts. It was an approach specifically designed to function well in its own specific domain, and it did so.

  13. Jongseong Park said,

    June 28, 2021 @ 9:11 am

    @J.W. Brewer, I was tempted to point out that the romanization of Mandarin Chinese was just as chaotic given the various competing systems in Taiwan as well as the old postal romanization. But Hanyu Pinyin clearly has established itself as the dominant one in the English-speaking world.

    Even now in Korea, ad hoc romanizations are allowed for passport purposes as long as they correspond to the Korean names (i.e. you can't register a completely different name as your Roman-alphabet name), although RR is the default system that they will use and is strongly encouraged.

    Regarding the guideline just mentioned, there was a case where the passport office refused to let a certain 주은 Jueun to register the Roman-alphabet name "June", but the Central Administrative Appeals Tribunal decided that this should be allowed, since "June" was close enough to be seen as a romanization of the Korean name and not just an English name.

    For surnames, established ad hoc romanizations are still followed 99% of the time rather than RR, hence Kim, Lee, Park, and Choi rather than Gim, Yi, Bak, and Choe (Yi was the recommended romanization instead of the regular I since there are practical difficulties with single-letter surnames).

    When I was doing my military service, I once saw a guideline for romanizing the names of Korean soldiers for the purposes of interactions with the U.S. military. It was literally a table matching syllables used in Korean names with English-oriented romanizations rather than a system like MR or RR, so more in the spirit of the Chinese postal romanization. The romanizations actually used by the individual in question in passports and such would be ignored for these administrative purposes.

    This was after RR had become official, but the possibility of following it for the romanization of the names of Korean soldiers had evidently been rejected if it was even considered in the first place.

  14. David Morris said,

    June 29, 2021 @ 6:34 am

    I noticed relatively recently that my wife's Korean passport has her name only in 'English'. I suspect that she provided the spelling herself because she uses 'young' for 영 (part of her given name), which none of the recognised systems do.

  15. 번하드 said,

    July 4, 2021 @ 8:01 pm

    Now here's a post to my taste, thank you!

    @Ronnie: Well done! For serious language learners, romanization helps for two weeks, after that it becomes a hindrance. I remember two former fellow students in some Korean class who had learned or taught themselves to speak Korean from a book heavily using romanization. They had to effectively relearn all vocabulary.

    Re: MR: it seems to allow just leaving out the breve if not possible to render in print, which always makes me really furious, with a language that's already brimming with ambiguity as is. Compared to that, I'd even prefer using one of two not-yet-mentioned transliteration (not transcription) systems, either https://en.wikipedia.org/wiki/SKATS (yeah, call me a nerd) or, on more practically inclined days, the assignment roman letters jamo from one of the computer keyboard standards, https://en.wikipedia.org/wiki/File:KB_South_Korea.svg, I've seen the latter method in regular use for mail account or IRC user names when forced to use the roman alphabet and not wanting to romanize.
    Example: 김치볶음밥, LUM CU WALL KDM WEW (SKATS), rlaclqhRdmaqkq (dubeolsik).
    Last: I got the impression that part of Korea's political right wing sees RR as a devilish left-wing device and would prefer to return to the old standard.

    @David Morris: the systems may not do it that way, but I know maybe ten Koreans whose names contain 영, and all of them romanized it as "young" when getting a passport.

RSS feed for comments on this post