Two Languages in Korea?

I’ve been following the discussion at the Log with great interest, and this long comment by Jongseong Park (all of whose contributions to the thread are well worth reading) is so informative I had to share it here:

The literary language in Korea was Classical Chinese until the end of the pre-Modern Era, continuing long after the invention of the Korean alphabet in the 15th century. The fact of having Classical Chinese as the literary language created a huge class of words in Korean taken from it, i.e. Sino-Korean. Even in the early 20th century, after Classical Chinese ceased to be the literary language, “neo-Sinitic” terms coined predominantly in Japan came into Korean via writing and were absorbed as Sino-Korean vocabulary.

Sino-Korean words refer specifically to those that can be written with Chinese characters and pronounced with the canonical readings of those characters in Korean, standardized through the long use of Classical Chinese as the literary language. So we exclude oral loanwords from Chinese languages past and present that differ from these canonical readings, like the older 붓 but “brush” from 筆 (canonical reading 필 pil) or the newer 쿵후 kunghu “kung fu” from 功夫 (canonical reading 공부 gongbu). We also exclude originally Sino-Korean words that have gone through sound changes to diverge from the canonical readings, like 성냥 seongnyang “matchstick” from 석류황 石硫黃 seongnyuhwang. On the other hand, we include “neo-Sinitic” words of non-Chinese (Japanese) origin like 전화 電話 jeonhwa “telephone”, and even faddish internet-age Korean neologisms like 역대급 歷代級 yeokdaegeup “historic level”.

Sino-Korean syllables are more phonotactically restricted than Korean vocabulary at large. For example, ㅃ, ㄸ don’t appear as onsets, ㅒ doesn’t appear as a medial, consonant codas are restricted to ㄱ, ㄴ, ㄹ, ㅁ, ㅂ, and ㅇ. It is often easy to tell if a word is Sino-Korean or not, especially if you recognize characters the way English speakers can recognize Latin roots like trans- or super-.

All this is to say that they are felt to be very different from more recent loanwords from English, German, or even Chinese (요우커 youkeo from 遊客 yóukè to give another example). Sino-Korean words are usually not considered 외래어 oerae-eo “loanwords” in Korean, but they’re not 고유어 goyu-eo “native words” either and form a separate, third class, 한자어 hanja-eo.

As you can imagine, the existence of these three broad categories means that there are lots of doublets or even triplets of words broadly referring to the same thing. The native and Sino-Korean vocabulary base is shared between the North and South. Yes, North Korea tends to prefer native forms more than the South, but that doesn’t mean that they have eliminated or even significantly reduced Sino-Korean words from their vocabulary. North Korean speech and texts will be peppered with Sino-Korean words to a degree comparable to their Southern counterparts. The most iconic North-Korea-related words are mostly Sino-Korean, like 주체 Juche or 로동 Rodong. Leaf through a North Korean dictionary and you will find the same preponderance of Sino-Korean words as in its Southern counterpart.

The North-South vocabulary differences are often highlighted and if you focus on those, it can give an impression that the North uses far less Sino-Korean words than the South, but I highly doubt that this is the case. The proportion of Sino-Korean words is so great that the relatively small number of divergent usage makes barely a dent. Also, Sino-Korean words also face competition in the South from native words and especially (other) loanwords.

Oh, and to clear up any misunderstanding, North Korea officially eliminated the use of Chinese characters to write Sino-Korean words from the beginning, using only the Korean alphabet to write Korean. South Korea didn’t take such a drastic measure as to forbid using Chinese characters to write Sino-Korean words, but in all the textbooks only the Korean alphabet was used to write Korean. This meant that the practice of “mixed-script” disappeared gradually and naturally as generations grew up educated only in using the Korean alphabet to write Korean—even in the most conservative South Korean newspapers the use of Chinese characters virtually disappeared by the 1990s. So after a half-century, Northern and Southern policies had the same end effect, of eliminating Chinese characters from general usage. And contrary to what one might think, North Korean students still learn Chinese characters in school, just like their Southern counterparts. So today there is negligible North-South difference on this issue, which is one of orthography, not vocabulary in any case.

Incidentally, I remember how surprised and pleased I was many years ago to learn that many Sinitic words, like 電話 “telephone,” were created in Japanese (denwa) and borrowed into Chinese (diànhuà); of course it makes perfect sense when you consider the respective curves of modernization, but we tend to take for granted the reverse order, in which Japan traditionally borrowed culture and vocabulary from China.


  1. Very interesting comment. It sounds very similar to the situation in Japanese. Even this:

    붓 but “brush” from 筆 (canonical reading 필 pil) or the newer 쿵후 kunghu “kung fu” from 功夫 (canonical reading 공부 gongbu)

    could be transposed to Japanese as:

    ふで fude “brush” from 筆 (canonical reading ひつ hitsu) or the newer カンフー kanfū “kung fu” from 功夫 (canonical reading くふう kufū)

    Of course, kanfū is from English, not from Chinese, so the parallel isn’t totally valid, but the two really are similar in many ways.

    I assume the main differences are that Japanese didn’t use Classical Chinese as the literary language in the same way or to the same extent that the Koreans did (although they did use it and still learn it in schools), and the Japanese kept the characters (although they came close to getting rid of them). I suspect the reason for the difference is the use of Chinese characters to write Japanese words (kun’yomi). If I remember rightly even this was borrowed from the Koreans, but the Japanese took it much further, with the result that characters became more deeply insinuated in the vernacular.

  2. An additional difference is that in Korean, the vast majority of Chinese characters have only a single reading (pronunciation). Following a South Korean character standard, Unicode provides optional redundant encodings for hanja that have more than one reading, almost as if there were three distinct encodings for the “a” in trap, the “a” in face, and the “a” in palm. However, only 258 characters with two readings and only 4 characters with three readings are provided, just a tiny fraction of the 3000 or so hanja that have been in recent use. This contrasts sharply with the Japanese situation, where a common character such as 生 has ten readings, two in Sino-Japanese words and the rest in native words.

  3. David Eddyshaw says:

    Japan *did* use Classical Chinese as its literary language, but with a very remarkable twist. It was pronounced as Classical Japanese. For example, my favourite Confucianism

    君子不器 lord child not utensil

    “A gentleman should not be a specialist”


    Kunshi wa ki narazu

    where “kunshi” is taken over en bloc as a Confucian untranslatable technical term, the topic particle “wa” is inserted though there is nothing corresponding to it in the Chinese, and the Chinese verbalisation of “utensil” is unpacked into noun + inflected negative copula in Japanese word order.

    All this is not arbitrary, but done according to fixed rules; diacritics may be added to the Chinese text to help.

    The result is called Kanbun 漢文

    Not content with reading Classical Chinese like this, the Japanese also ran the process in reverse, composing a Classical Japanese sentence and then turning it into (sorta) Classical Chinese, to be subsequently read back in Classical Japanese. This was the standard way of writing technical stuff until Meiji times, and even now the peculiar sort of Japanese that results from the process has had a major effect on the more technical styles of Japanese writing.

    There is even a genre of poetry in which Classical Chinese poems, with Chinese scansion and rhymes, is sung in Japanese (in which it neither scans nor rhymes)

    I believe there was once something like the Kanbun tradition in Korea too, but it became defunct long before the Japanese one.

  4. Thanks, David, for that explanation. It’s clear that the penetration of Chinese into Japanese wasn’t simply a matter of kun-readings; it was also due to the use of kanbun as a kind of ‘hybridised between-language’. At any rate, there are differences between the relationship of Korean to Chinese and Japanese to Chinese that make the latter considerably more complicated.

    Jongseong Park’s division between 고유어 goyu-eo “native words”, 한자어 hanja-eo “hanja words”, and 외래어 oerae-eo “loanwords” is exactly paralleled in Japanese with 和語 wago, 漢語, kango and 外来語 gairaigo. And it was the Japanese custom of writing both kango and wago in Chinese characters that led the Chinese to unwittingly borrow a good number of Japanese words into their vocabulary in the modern era.

  5. David Eddyshaw says:

    The Kanbun tradition is one of the linguistic wonders of the world, though one sees George Sansom’s point

    “There is no doubt that [the Japanese writing system] provides for some a fascinating field of study, but as a practical instrument it is surely without inferiors.”

    I suppose the true diehard response would be that there is more to life than practicality.

  6. 붓 but “brush” from 筆 (canonical reading 필 pil) / ふで fude “brush” from 筆 (canonical reading ひつ hitsu)

    Japanese “fude” is usually explained as derived from “fumite” (“fumi”=writing, “te”=hand), which is pure native Japanese and thus not analogous to Korean 붓 “but”, which is an early loan from Chinese into Korean (and thus doesn’t follow the standard phonological patterns for later loans). Korean 붓 “but” is thus more closely analogous to Japanese words like uma 馬 (horse) or ume 梅, which are etymologically Chinese loanwords but are not usually considered kango because of their peculiar history and phonological form.

  7. Thanks for the info! I had assumed that ふで was from the same root as 붓, similar to the 馬 and 梅 examples.

    Regarding kanbun, I recently had the experience of trying to read a piece written in Classical Chinese and not being able to make much sense of it (since I have no training in the language), which a kanbun-trained Japanese was then able to elucidate on the spot without any trouble at all.

  8. David Eddyshaw says:

    The closest analogy to the Kanbun way (of treating one language as code for another) I can think of is pre-Islamic Persian; in Achaemenid times the author of a letter dictated it in Persian, the scribe wrote his message in Imperial Aramaic, and at the receiving end the recipient’s scribe read out the Aramaic as Persian.

    In the Parthian era, this was superseded by writing Pahlavi itself, in a highly defective cursive Aramaic script, not only unvowelled but with no differentiation between many consonants (like Arabic but without all the diacritic dots) and with common words written as (often rather mangled) Aramaic, so that “mard” “man” appears as gbr’, “zan” “woman” as nsh, for example.

    lhyk may represent du:r “far” (via Aramaic) or rahi:g “child”

    A scribal monopoly is well worth defending, I suppose. Can’t have those common folk learning to read and write …

  9. David Eddyshaw says:

    The question “what language is this document in?”, which seems so straightforward to us, has not at all times and in all places had a simple answer, or even been a meaningful question.

    The cases where a quite unrelated language underlies writing seem particularly strange, but there are have been a great many more where the written form of a language is different enough from the spoken to require separate grammatical treatment, even though the two are conventionally regarded as the same.

  10. The cases where a quite unrelated language underlies writing seem particularly strange

    Another case being the Hittite use of Akkadian and Sumerian to write their own language (one result being that the Hittite words for most numbers is unknown).

  11. pre-Islamic Persian

    That doesn’t seem to me one bit as weird as Kanbun. The analogue today would be one German executive for a multinational company writing a memo in English to another German executive, because English is the working language of the company. The executives would normally serve as their own scribes, but otherwise the situation seems quite parallel.

  12. marie-lucie says:

    In early and medieval Europe when most writing was in Latin, documents must have been dictated in the local language and written down by scribes in Latin, with the reverse at the recipient’s end. The same procedure would have been followed whether the language at each scribe’s end was Italian, German or whatever.

    Diglossia : When I was a child, in primary school we did writing exercises in which we contrasted Je dirais ‘I would say’ and J’écrirais ‘I would write’, as for instance on est allé à la mer and Nous sommes allés à la mer ‘We went to the seaside’ (eg for our vacations). The written version was always something that no normal child would ever have uttered. In newspapers or magazines, ordinary people’s oral comments were transposed into the written style, even when the quoted speakers were children.

  13. In newspapers or magazines, ordinary people’s oral comments were transposed into the written style, even when the quoted speakers were children.

    Oh, how I hate that! It turns the varied and healthy living language into a bland gruel that nourishes no one and reinforces an artificial conformity that poisons people’s minds.

  14. David Eddyshaw says:

    In premodern Japanese, the Epistolary Style, sōrōbun 候文, a highly specialised variant of Classical Japanese notable inter alia for using the auxiliary -sōrō (cognate with “samurai”!) everywhere, was sometimes conventionally used to represent actual speech; from which it is far removed, being one of the types of writing heavily infleunced by Kanbun.

  15. It turns the varied and healthy living language into a bland gruel that nourishes no one and reinforces an artificial conformity that poisons people’s minds.

    As in the works of Shakespeare, whose refined characters are so refined they actually talk in blank verse. Down with artificiality!

  16. You’re joking, I hope. Blank verse has zero to do with what I’m talking about.

  17. Of course I’m joking. My only point was that written dialogue isn’t necessarily the same as spoken dialogue.

  18. David Marjanović says:

    and even faddish internet-age Korean neologisms like 역대급 歷代級 yeokdaegeup “historic level”

    Surely that’s “epic-level”?

  19. @DMT:

    This source might not be reliable, but at the very least it seems that the theory that ふで derives from Chinese 筆 is not original.

    The interesting thing is that even if ふで derives from ふみ+て, this source also lists ふみ as derived from Chinese 文.

    In my completely layman opinion, brushes and writing seem like obvious candidates of words that would have been borrowed by the Japanese at first contact with Chinese civilization.

  20. @dainichi: Thanks for linking to that page – it is good to know of the alternative explanation, even without details. I had initially hesitated before posting my comment, since the derivation of Japanese fude from Middle Chinese byut 筆 seemed prima facie plausible (for the historical reason you mention), but I couldn’t immediately find any sources proposing it. Another example on the page you linked to would provide further support, since the derivation of Japanese take “bamboo” from Middle Chinese tyak 竹 (assuming this is correct) would provide an analogue for the unusual epenthetic /e/ corresponding to a final stop in Middle Chinese. (Middle Chinese final stops were normally transcribed in phonetic Japanese using kana syllables rhyming in /u/ or /i/; however, these final vowels in the kana transcriptions of kango may not be accurate reflections of pronunciation. Sixteenth-century Jesuit transcriptions of the spoken language generally omit them. I have no idea what this would imply about the Old Japanese pronunciations of the proposed loanwords for “brush” and “bamboo.”)

    There are a couple of other examples on that page that seem dubious and perhaps indicative of excessive eagerness to trace foreign origins for Japanese words.

    (1) Japanese iwashi “sardine” clearly cannot come from Russian ivasi, since the Japanese word is attested in eighth-century sources.

    (2) Spanish olla “pot” has been proposed as an origin for Japanese ojiya “flavoured rice gruel”, but this etymology isn’t well attested and doesn’t seem especially probable.

    (3) Dutch kabas as an origin for Japanese kaban seems suspect: (i) the phonetic correspondence would be very strange; (ii) the word first appeared in 1877, slightly too late to make a Dutch origin seem likely; (iii) the Dutch word seems to have been rare or archaic by the late nineteenth century (I don’t know Dutch well enough to say this definitively), and thus is unlikely to have served as a source for a Japanese loanword.

    (4) Finally, there is the proposal that Japanese fumi is from Middle Chinese bun 文 “writing”, which is a bit more ambiguous. As you say, this seems like an attractive etymology that would have a convincing historical explanation, but are the phonetic forms convincing? How does Middle Chinese syllable-final /n/ give rise to Japanese /mi/? The explanation offered on the linked webpage (bun>bum>pumu>pumi) seems pretty ad hoc and speculative, especially the initial bun>bum. Are there any parallel examples for this change?

  21. Jongseong Park says:

    Wow, I’m much honoured that my comment was shared here. Thanks!

    I wish I knew more about Japanese because there are so many interesting historical parallels with Korean. Basically, both Koreans and Japanese were faced with the problem of using a literary language (Classical Chinese) in a logographic writing system that didn’t fit their word order.

    Many strategies were tried out in Korea, and if I recall correctly, this included writing Chinese in Korean word order, seen in writings from the Three Kingdoms period (up to the 7th century).

    As David Eddyshaw says, something like the Japanese kanbun was also practised in Korea. A Classical Chinese text would be translated into Korean as it was read, and it was annotated with Korean particles and reading marks, such as those telling you to go back in a sentence so that it comes out in the right word order in Korean. This is attested from around the 12th century during the Goryeo period, but probably dates from earlier.

    In the Joseon period, though, instead of translating a Classical Chinese text into Korean, the Chinese characters would simply be given the canonical readings and treated as content words, with Korean particles added in and word order unchanged. This system is called 구결 口訣 gugyeol after the shorthand characters used to write the particles.

    This is completely beyond my area of knowledge, and I haven’t seen any actual examples of texts with gugyeol, but I imagine that it would be something like writing 君子는 不器라 in the “A gentleman should not be a specialist” example above (except with the subject marker 는 and the copula 라 written in shorthand Chinese characters instead of the Korean alphabet). The original word order of 君子不器 is kept intact.

    This divergent evolution in Korea and Japan with respect to the established way of reading Classical Chinese texts must explain why Chinese characters are only read in their canonical readings (corresponding to original Chinese pronunciations) in Korea while in Japan they can also stand for native words.

    About diglossia: in Korean, when people are quoted in newspaper reports and such, their speech is rendered in the neutral, literary form. Where one might have said “있어요/있습니다/있어” with various degrees of formality and politeness, it is flattened to the citation form “있다” stripped of all such markers. It is such an established practice that it would be distracting to see the actual spoken verb endings. Knock on diglossia all you want, but if I’m only interested in the content of the quote, I don’t want to be reminded about the question of hierarchy and formality that comes into play with the relationship between the interviewer, the interviewee, and the audience.

    @David Marjanović: Surely that’s “epic-level”?

    That’s a much better translation!

  22. David Marjanović says:

    Middle Chinese final stops were normally transcribed in phonetic Japanese using kana syllables rhyming in /u/ or /i/; however, these final vowels in the kana transcriptions of kango may not be accurate reflections of pronunciation. Sixteenth-century Jesuit transcriptions of the spoken language generally omit them.

    That’s fascinating!

    How does Middle Chinese syllable-final /n/ give rise to Japanese /mi/?

    That’s strange indeed. Final /n/ > /m/ is regular in Portuguese, but it’s clearly not regular in Japanese, is it? Also, the opposite change is much more common.

  23. Eli Nelson says:

    Maybe the Japanese moraic nasal used to be pronounced more like an m sound; apparently the hiragana used to write it used to also be used for the syllable “mu”. That would naturally lead to an “m” sound in an imported one-syllable word that ended in a nasal.

  24. Eli Nelson says:

    Clarification: obviously the spelling doesn’t itself lead to the m pronunciation, but the spelling suggests it might have existed as an important allophone of the moraic nasal.

  25. Of course I’m joking.

    Sorry! I guess my sensayuma is temporarily deranged by having two separate editing deadlines impending. Gotta stop falling victim to Freelancer’s Disease (never turn down a job! you may not get another!).

  26. David Eddyshaw says:

    @Jongseong Park:

    Thanks for the fascinating stuff about Korean and Kanbun-type strategies. As in so many other respects relating to Chinese culture, Korea was evidently the teacher of Japan.

  27. @ David Marjanović: Final /n/ > /m/ is regular in Portuguese, but it’s clearly not regular in Japanese, is it? Also, the opposite change is much more common.
    Precisely. (While we’re on the subject of nasals in Portuguese – I recently learned that Jesuit transcriptions use “m” as a transcription of syllable-final /ŋ/ in Mandarin. I presume this was simply because they didn’t have /ŋ/ in their own language and “m” was the closest nasal they could think of – conveniently, there was no syllable-final /m/ in Mandarin by the time the Jesuits arrived and thus no ambiguity, since Middle Chinese syllable-final /m/ is reduced to /n/ in Mandarin. However, I would have thought that even Portuguese Jesuits would have realized that /ŋ/ isn’t all that uncommon a sound in the world’s languages and that it would be convenient to come up with a better standard transcription for this sound than “m”. I would be very happy to be enlightened about what was going on here!)

    @Eli Nelson: You are entirely right about the spelling ambiguity of mu む vs. n ん in early Japanese writing. It might be possible to come up with a story of how 文 became fumi in terms of allophones or spelling pronunciations or Japanese mishearing of Korean Buddhist missionaries who couldn’t pronounce Middle Chinese properly anyway or something like that – but the story would need to be explicit before we could even begin to evaluate its plausibility. Presumably the internet poster that danichi linked to was drawing on a source that was written with such a story in mind, but I haven’t yet seen one spelled out.

  28. 1) Japanese iwashi “sardine” clearly cannot come from Russian ivasi, since the Japanese word is attested in eighth-century sources.

    and in Russia they have no doubts that their word is borrowed from Japanese

    although nowadays people don’t associate it with herrings or sardines – the word got a completely new meaning, as the scenic name of a duo of guitar players and singers, IVASchenko & VASIl’ev.

  29. In my completely layman opinion, brushes and writing seem like obvious candidates of words that would have been borrowed by the Japanese at first contact with Chinese civilization.

    According to Schuessler’s ABC Etymological Dictionary of Old Chinese entry for 筆 (Middle Chinese pjet):

    Shuōwén jiězi gǔlín (1271) says that is a word from the region of Qín 秦 (NW), and that in Chǔ 楚 the word is ⾀ (jiuet) Old Northwest Mandarin iuit (?), in Wú 吳: bù-lǜ 不律 (pjəu-ljuet) Old Northwest Mandarin pu-luit, and in Yán 燕 (NE) 弗 (pjuət) Old Northwest Mandarin put (discussed by Sagart ICSTLL 1990:7). The word has been borrowed by Middle Korean (pwut), Old Japanese (pude), and Viet (bút) (pre-Sino-Viet.) (Miyake 1997: 189, 192). All these data point to an initial cluster *pr- or *pl- and an Old Chinese rime *-ut. [I omit etymologies for the Chinese word, which include Sino-Tibetan and Proto Austronesian].

    According to Gerard Clauson (Studies in Turkic and Mongolic Linguistics 1962), the Mongolian word biçi- ‘to write’ as found in the Secret History of the Mongols is from Turkic biti:-, ultimately derived from Chinese pi ‘a writing brush’, and biçig ‘letter’ is from Turkic bitig. [This etymology was the subject of some controversy at an earlier LH thread.]

    According to Schuessler, Written Tibetan bir ‘writing brush’ is a Middle Chinese loan (Middle Chinese final -t -> Written Tibetan -r).

    I assume that this is the origin of the Mongolian word bi:r ‘writing brush’.

    ICSTLL = International Conference on Sino-Tibetan Languages and Linguistics
    Miyake, Marc H. 1997 = Pre-Sino-Korean and Pre-Sino-Japanese: reexamining an old problem from a modern perspective. In Japanese/Korean Linguistics vol. 6: 179-211 (ed. Ho-min Sohn and John Haig. Stanford: Center for the Study of Language and Information)

  30. That should have been:

    and in Yán 燕 (NE) 弗 (pjuət) Old Northwest Mandarin put

  31. While we’re on the subject of nasals in Portuguese – I recently learned that Jesuit transcriptions use “m” as a transcription of syllable-final /ŋ/ in Mandarin.

    Written -m is a nasalization diacritic rather than a nasal consonant in Portuguese, and it is not uncommon for Mandarin -ng to also appear as nasalization of the preceding vowel, particularly in Southern Mandarin, so the use of -m in Mandarin romanization strikes me as rather sensible.

  32. marie-lucie says:

    JC: Written -m is a nasalization diacritic rather than a nasal consonant in Portuguese, …., so the use of -m in Mandarin romanization strikes me as rather sensible.

    I was just about to write basically the same thing about Portuguese: written m at the end of a word is not normally pronounced as a consonant. I have not studied the history of Portuguese, but judging from the Mandarin transcriptions, this was already the case a few centuries ago.

  33. @John Cowan, marie-lucie: Thanks for all the information on Portuguese m. I believe that Nanjing Mandarin remained the standard until as late as the seventeenth century – if the Nanjing Mandarin of this period realized syllable-final /ŋ/ as a nasalization, then the Jesuit transcriptions would be entirely natural.

    @Bathrobe: Thanks for posting this – it makes the Chinese-loanword explanation look very plausible, even if details are lacking. Given that fumite and funde are also attested words for brush in Old and Middle Japanese, it may be very difficult to disentangle the possible etymologies. The Miyake paper should be worth a look.

  34. Jongseong Park says:

    Korean 붓 but (Yale Romanization pwut) is probably from Old Northwest Mandarin put or a similar form. Is this an earlier or a later loan than Sino-Korean 필 筆 pil (Yale Romanization phil)?

    Sino-Korean readings seem to represent an unbroken tradition dating back to the Unified Silla period (668–935), but periodically influenced by exposure to contemporary Chinese and Chinese rime books. Originally it must have been from Old Chinese or Middle Chinese where -p, -t, -k in entering tones were maintained.

    These are robustly maintained in Sino-Korean readings, but the original -t has been systematically converted to -l. This could reflect borrowing from a variety of Chinese (older forms of Mandarin?) that already weakened the original -t to something like [ɾ]. Alternatively, the change from -t to -l could have taken place in Korean. There are arguments for both sides in the existing literature, but the latter view appeals more to me. There are plenty of examples in Korean of alternations of /t/ and /l/, as you see in 숟가락 sutgarak = 술 sul + 가락 garak “spoon”, or in irregular conjugations such as 걷다 geotda “walk (citation)” – 걸어 georeo “walk (infinitive)” (cf. t-flapping in varieties of English; r- [ɾ] is the same phoneme as -l).

    In Sino-Korean words, the coronal obstruents d, s, and j systematically undergo fortition after -l, as in 발단 發端 baldan, 발생 發生 balsaeng, 발전 發展 baljeon which are pronounced as if written 발딴 balttan, 발쌩 balssaeng, 발쩐 baljjeon. Revised Romanization of Korean doesn’t reflect fortition, though McCune-Reischauer does use voiceless letters in the case of fortition, and would write paltan, palsaeng, palchŏn for these cases, leaving only palsaeng unchanged. No such fortition happens systematically with other consonants following -l.

    Such a rule can be explained if the consonant causing the fortition of coronal obstruents is -t. So this seems to me to be evidence supporting to hypothesis that -l actually was -t at one stage in Korean. It would be more complicated to explain how this was inherited as a systematic rule in Korean from the source pronunciation after -t had already been lenited.

    But there are plenty of native Korean words ending in -t. Why did all Sino-Korean -t turn systematically in -l but not -t in native words? Something about -t in native words being a neutralization not just former -d and -t but also -s, -j, and -ch may have something to do with it, but the crucial factor may be tone/length. Middle Korean had a tone system, and all Sino-Korean syllables in -t had the entering tone in Chinese, so were mapped to a single (short) tone in MK. In native words, -t could occur in any of the tones. This may explain the systematic change to -l in Sino-Korean that didn’t affect a large number of native -t.

    As for 붓 but, if the corresponding Chinese word had an *-ut rime already in Old Chinese which more or less continued until Old Mandarin, I think it will be very difficult to ascertain whether it was earlier or later than the Sino-Korean 필 筆 pil. It could have entered the language at any time and for whatever reason not subject to the -t to -l change in Sino-Korean (if that in fact took place in Korean and it came after 붓 but entered the language).

  35. @Jongseong Park: This is perhaps a long shot, but does orthography provide any clues? What is the history of hangul’s use of ㅅ to represent /s/ in syllable-initial position but /t/ in syllable-final position?

  36. Historical linguistics has always confused me, but in this case it looks like Korean 필 筆 pil is related to Japanese 筆 ヒツ ~ ヒチ hitsu ~ hichi (on’yomi), suggesting a source more like bit than but. Hitsu ~ hichi is a kan’on (漢音) reading, one which, according to Wikipedia, is “from the pronunciation during the Tang Dynasty in the 7th to 9th centuries, primarily from the standard speech of the capital, Chang’an (長安 or 长安)”. Hazarding a highly inexpert guess here, it looks like the Korean and Japanese was borrowed from the Chinese of the 秦 area before lost its final -t.

  37. It also seems to me that 붓 but almost certainly must be older than 필 pil because 붓 but gives the appearance of being an assimilated word whereas 필 pil fits in with the systematised canonical readings.

    My apologies for omitting the readings that Schuessler gives at the entry for 筆:

    pjet (Middle Chinese)
    pɨt (Later Han)
    *prut (Minimal Old Chinese)
    *pjrut (Old Chinese according to Baxter)

    The periodisation of Chinese is: Old Chinese — from the beginning of writing till the Han period, Later Han Chinese — ca. 2nd-3rd century AD, Old Northwest Chinese (not Mandarin as I wrote) — ca. AD 400, Middle Chinese — ca. 600 AD.

  38. Jongseong Park says:

    Orthographic ㅅ s as in 옷 ot (os) “clothes” (attested in this form in the 15th century) simply represents historic -s that neutralized to -t in pronunciation. But it surfaces as -s when followed by a vowel-initial particle, so you have 옷이 ot-i, 옷에 ot-e, 옷을 ot-eul pronounced as if written [오시] osi, [오세] ose, [오슬] oseul.

    The neutralized coda -t is really lenis ㄷ -d, though in romanization we use -t because it is a voiceless unexploded [t̚]. Followed by a vowel-initial word that is not a particle, it surfaces as -d. Hence 옷 안 ot an is pronounced [오단] odan

    It wasn’t just -s that neutralized to -d. Other consonant codas that neutralized to -t such as -t, -j, and -ch show similar patterns:
    bat (bat) “field” : 밭이 [바치] bachi (due to palatalization), 밭에 [바테] bate, 밭을 [바틀] bateul
    nat (naj) “daytime” : 낮이 [나지] naji, 낮에 [나제] naje, 낮을 [나즐] najeul
    kkot (kkoch) “flower” : 꽃이 [꼬치] kkochi, 꽃에 [꼬체] kkoche, 꽃을 [꼬츨] kkocheul

    In dialectal pronunciations, even words like 밭 bat (bat) and 꽃 kkot (kkoch) may be reanalyzed as -s words, so that you have nonstandard pronunciations like 밭을 [바슬] baseul and 꽃을 [꼬슬] kkoseul. These dialects would also tend to neutralize other codas to -b and -g, saying 무릎을 [무르블] mureubeul instead of the standard 무릎을 [무르플] mureupeul and 부엌을 [부어글] bueogeul instead of the standard 부엌을 [부어클] bueokeul. For whatever reason, where the other non-nasal, non-liquid codas are neutralized to -b and -g, -s tends to be the choice instead of the expected -d.

    It is probably because of the paucity of nouns in -d that one could generalize from. I can’t really think of any free-standing common nouns ending in ㄷ -d. But they do occur in compounds. As I mentioned before, 숟 sut (sud) occurs in 숟가락 sutgarak “spoon (utensil)”, but by itself it is 술 sul “spoonful”. Another example is 이튿날 iteunnal (iteudnal) “the second day/the following day” and 이틀 iteul “two days”. This wouldn’t be surprising if we supposed that -d regularly changed to -l in Sino-Korean subject to certain tone constraints, but were preserved in compound forms.

    Intriguingly, in the earliest sources in the Korean alphabet dating from the 15th century, today’s 갓 gat (gas) “traditional hat” appeared as 갇 gat (gad), 못 mot (mos) “nail” as 몯 mot (mod), and 벗 beot (beos) “friend” as 벋 beot (beod). I don’t know if these historical spellings were based on actual considerations of how these codas surfaced when followed by vowel-initial particles or were just ad hoc spellings, but considering their early date they’re probably from before the neutralization of the codas. This suggests to me a possibility that there were originally more -d nouns still left over from any historical changes that turned -d to -l in many cases, but they were then reanalyzed as -s.

    When Korean orthography was standardized, ㅅ -s was used extensively for all cases of [-t] as a default unless there were clear reasons to do otherwise (i.e. it potentially surfaces as a different consonant or is clearly etymologically connected to a form where it does). In more recent loanwords, original -t (when it doesn’t take an epenthetic vowel) is systematically adapted as -s, as in 인터넷 inteonet (inteones) “internet”, with 인터넷이 [인터네시] inteonesi, 인터넷에 [인터네세] inteonese, and 인터넷을 [인터네슬] inteoneseul. Indeed, according to the standard orthography, only ㅅ -s is permitted as a coda in writing loanwords among the various codas that neutralize to ㄷ -d.

  39. @Jongseong Park: That’s a much more thorough answer than I had any right to expect! Since you didn’t mention it, I take it that you don’t know of any early cases of the word for brush being written as 붇 – so the earliest written sources for this Korean word already employ the standardized “default” orthography with final ㅅ.

  40. Jongseong Park says:

    @DMT: Good point, I should have checked for 붓 but (bus). The earliest spelling from the 15th century is indeed 붇 but (bud). This is to be expected if we’re assuming a loan from a Chinese source in -t, because at this earliest stage the Korean codas had not been neutralized.

    After the original -s and -d codas were neutralized in pronunciation (along with -t, -j, and -ch), it must have created a confusion both in spelling where original -d began to be written -s, and also in pronunciation where -d nouns began to be pronounced like -s nouns by analogy. When Korean orthography was standardized, I suspect most instances of coda [t̚] were already being spelled -s. Indeed, pre-standardization spelling often uses -s where modern spelling uses something else, like 잇다 itda (isda) for modern 있다 itda (issda) “to exist” or 밋다 mitda (misda) for modern 믿다 mitda (midda) “to believe” that you can find in early 20th century publications.

    I haven’t mentioned yet the tough-to-pin-down clitic known as 사이시옷 sai siot “in-between s”, which acts a bit like the English genitive marker ‘s. Middle Korean originally had a range of consonants that acted in this function, but after the end of the 15th century -s became the sole “in-between” consonant. In modern pronunciation, it generally surfaces just as the fortition of the following consonant, as in 시내 + ㅅ + 가 = 시냇가 [시내까] sinaekka “bank of a stream”, although it is also possible to pronounce the [t̚] as in [시낻까] sinaetkka. This sound has long been indicated with ㅅ s, sometimes standing on its own and not joining other syllables, but in modern South Korean standard orthography it is always written as a coda -s under certain conditions (with exceptions, of course). This causes endless confusion in Korean orthography, with successive spelling reforms going back and forth on certain examples. In North Korea, they have done away with writing this “in-between s” altogether, so they write 시내가 even if it is pronounced [시내까] sinaekka just as in the South.

    Anyway, it could be the prevalence of this s sound that led to people treating -s as the default for the coda [t̚] sound.


  1. […] Hat notes the long history of a Chinese-influenced literary language in […]

Speak Your Mind