Why am I so interested in aberrant readings like Sino-Korean 묘 myo instead of 모 mo for 墓 'grave'? Such readings disturb conversion formulae that make learning easier. Soon after I got my copy of Nelson's Japanese-English Character Dictionary about 23 years ago, I wrote Sino-Korean readings in hangul beside the characters. My ultimate goal was to make the dictionary a Sino-Japanese-Korean-English dictionary, but I never got there.

Adding SK readings made me notice correspondences between them and SJ readings: e.g.,

- if a character had SJ readings with initial b- and m-, its SK reading would have initial m-

- if a character had SJ readings with final -o, its SK reading would have final -o

Learning these correspondences cut down on memorization. Since 募 'recruit', 慕 'yearn for', 暮 'dusk', and 墓 'grave' have SJ readings bo ~ mo, I would guess that their SK reading would be mo. I would only be three-fourths correct, since the SK reading of 墓 'grave' is myo with an extra -y-.

Of course, such predictive correspondences do not work at all with unrelated words: e.g., I cannot predict the native Korean word mudŏm 'grave' (< mut- 'bury') on the basis of the native Japanese word haka 'grave'. Nevertheless, being able to guess Sino-Korean readings gave me a useful key to half of the vocabulary of Korean.

I'm still on the lookout for correspondences and exceptions today. While looking at Chinese readings for 墓 'grave', I noticed that two readings ended in an unexpected -ŋ:

Hankou moŋ

Xiamen bɔŋ

Starostin's online database also lists Fuzhou muoŋ but Hanyu fangyan zihui lists Fuzhou muɔ without a final nasal.

Hankou generally has moŋ corresponding to mu in standard Mandarin:

Sinograph Gloss Old Chinese Middle Chinese Hphags-pa ChineseStandard Mandarin Hankou Xiamen (colloquial if available)
grave *maks *moh mumu moŋ ŋ
yearn mɔ̃
tent *mak *makmaw
to herd *mək *mukvu bɔk
tree*mok*mok mubak
eye *muk *mukvu
mother *moʔ *məwʔmuw bo
Chinese acre mou
model *ma *momu mo
sink *mət *mot mo moŋ but

(The reading for 模 is as in 模子 'mold'. 模 is generally read mo in Mandarin.)

The correspondence Md -u : Hankou -oŋ : Xiamen -ɔŋ is unique to 墓 'grave'.

I hypothesize pre-Hankou *mu (?) regularly became *mũ, *muŋ, and then moŋ. (There is no rhyme -uŋ in Hankou.)

mou and 模 mo may be loans or archaisms that were never *mu in pre-Hankou.

A similar change did not affect pre-Hankou *nu which became nou, not *noŋ.

I don't know how to explain Xiamen 墓 bɔŋ which may truly be irregular. 墓碑國 MYOPI-A: MYSTERY OF THE GRAVESTONE GLIDE

Yesterday, I looked at Sino-Korean readings which had unexpected front vowels from an earlier unetymological final -y. There's at least one Sino-Korean reading with an unexpected medial -y-:myo for 墓 'grave'. No Chinese language has a -y- in this morpheme: e.g., Mandarin mu, Cantonese mou Sino-Japanese bo has no -y-. Vietnamese borrowed this morpheme three times:

mả < Late Old Chinese ?*mhah (or *mah plus a LOC or early VIetnamese devoicing prefix?)

mồ < Early Middle Chinese *mo

mộ < Late Middle Chinese *mo

(The different tones of the last two forms imply a tonal change between the dialects of EMC and LMC spoken in Vietnam.)

None of the three borrowings have an initial d- from an earlier *my-.

Did SK myo preserve a *-y- lost elsewhere? I would normally reconstruct the earliest Old Chinese reading of 墓 as *maks (cf. its phonetic 莫 *mak) without a *-y-. I am hesitant to insert a *-y- because I think the glide is a Korean innovation.

In the previous post, I discussed cases of fronting in Korean conditioned by palatal segments in following syllables. The expected Sino-Korean reading of 墓 is 모 mo, which is the idealized Middle Sino-Korean reading in 東國正韻 Tongguk chŏng'un (1447). Martin's rule would predict that 墓 mo would become 뫼 moy (> modern Korean moe) following a syllable triggering fronting: e.g.,

'gravekeeper': 墓직이 *motsiki > *moytsiki > modern *moejigi (actually myojigi)

'tombstone': 墓碑 *mopi > *moypi > modern *moebi (actually myobi; Yale romanization myopi - hence the post title 墓碑國 myopi-a 'tombstone country', meant to sound like utopia as well as myopia)

But obviously the -y- in 墓 myo is medial, not final. And the earliest attestation of 'gravekeeper' I can find (譯語類解補, 1775) has myo-, not mo-. So why is -y- medial? I have two explanations that are both unsatisfying:

Sinification: The *moy resulting from fronting is not a Sino-Korean syllable, and was changed to myo to sound more Chinese (cf. other myo readings in Sino-Korean for 苗妙廟卯淼, etc.) Unfortunately there is no record of a *moy stage.

Taboo deformation: mo 'grave' became a taboo morpheme and -y- was arbitrarily inserted to create a replacement myo. But I am unaware of other examples of deformed Sino-Korean readings. (Nguyễn Đình-Hoà wrote an article about such deformation in Sino-Vietnamese.) FISH-E READINGS

Yesterday I stumbled on the word 제설 (除雪) chesŏl 'snow removal' (lit. 'remove-snow') in Minjung's Essence Korean-English Dictionary (4th ed., p. 1959), which is fitting considering the weather around here lately. 除 che 'remove' is one of a small set of 魚 'fish' rhyme Chinese characters pronounced with -e instead of -ŏ in Korean:

Chinese character Meaning Middle Chinese Sino-Korean: -ŏ ~ -e Expected Sino-Korean reading: -ŏ only Idealized Middle Sino-Korean Mandarin: -u [y] ~[u] Sino-Japanese: -(y)o Sino-Vietnamese:
fish *ŋɨə ŏ (has expected reading) ŋə yu gyo ngư
in advance *jɨəh ye yŏ like their Chinese homophone 轝 yo dự
fame; praise
various *tɕɨə che chŏ like its Chinese homophone 藷 tsyə zhu sho chư
remove *ɖɨə chŏ like its Chinese homophones 儲躇 ttyə chu jo trừ

The idealized Middle Sino-Korean readings from 東國正韻 Tongguk chŏng'un (Correct Rhymes of the Eastern Nation; 1447) all end in which normally developed into modern Sino-Korean -ŏ. So why aren't 豫預譽 yŏ and 諸除 chŏ? I used to think that the seemingly irregular SK readings endings in -e came from an earlier *-əy reflecting a Chinese final glide like *-ɰ (corresponding to Pulleyblank's *-ɣ and Li Fang-kuei's *-g). These *-əy readings would have regularized in Tongguk. However, I now have an internal explanation for -e instead of -ŏ.

There are many cases of vowel fronting in Korean native words. For instance, two weeks ago I mentioned yesun < yəysyun 'sixty' correspoding to yŏsŏt < yəsɯs 'six'. Martin (1992: 39) described this phenomenon:

We also find pairs of words in which one member, usually the more common form, has a front vowel either after c(h) [= ch(h) on this site] or before a syllable that contains i or y.

This could be formulated as

CVc(h)ŭ ... ~ CVc(h)i ...

e.g.., achhŭm ~ acchim 'morning'

CVCi ... ~ CVyCi ...

e.g., agi ~ aegi (spelled ayki) 'child'

CVCyV ... ~ CVyCyV ...

e.g., hakkyo ~ haekkyo (spelled haykkyo) 'school'

with y representing fronting.

'Sixty' fits this pattern since its second syllable contained y:

*yəs-syun > yəysyun > yesun

The e of 豫 ye could be from the i of the following syllable in

豫備 ?*yəpi > yəypi > yebi 'preparation'

which is the earliest Korean word with 豫 that I can find. 預 is a synonym of 豫, so naturally it too would be read the same way.

The e of 除 che could be from the y of the following syllable in

除授 *tyəsyu > tyəysyu > chesu 'be assigned to a post' (obsolete)

However, such explanations may not work for the other irregular -e readings. The earliest Korean words I can find containing the graphs for those readings are

諸王 tsyəywaŋ > chewang 'kings'

名譽 myəŋyəy > myŏngye 'glory'

There is no palatal segment in waŋ to condition the fronting of 諸 *tsyə and 譽 yəy isn't followed by a palatal-segment syllable.

Martin (1992: 39) lists some other puzzling frontings. I think the first two reflect attested -y-s that were lost in modern Korean, but I cannot account for the others:

tanchhu ~ taenchhu < tantshyo 'button'

wŏnsu ~ wensu < 怨讐 wənsyu 'enemy'

mandŭnda ~ maendŭnda 'makes' (earlier *-ty- would have become -j-)

onthong ~ oenthong ~ wenthong 'entirely' (earlier *-thy- would have become -chh-) BUGS ALSO WORD WELL

- mostly written 09.10.27; expanded tonight -

One of the Chinese varieties mentioned in last night's post was 泰順蠻講 Taishun Manjiang, a branch of Eastern Min. 泰順 Taishun is a 縣 county and Manjiang literally means 'southern barbarian speech'. Not very PC. The simplified characters


for 蠻 講 Manjiang consist of

yi 'also', arbitrary abbreviation of 䜌 luan? < Old Chinese *m-ron 'bells on horse's trapping' (phonetic) +

chong 'bug' (semantic)

yan, abbreviation of 言 yan 'words' (semantic) +

jing, abbreviated phonetic substitute for the phonetic 冓 gou (second half of 中冓 'inner chamber') which also happens to resemble its topmost strokes.

井 makes no sense as a phonetic of 讲 in non-Mandarin languages; e.g., in Cantonese, 井 is Ct jing and 讲 is Ct gong.

Manjiang used to have a dedicated site (mangomo.com; what is "mangomo"?*), but it's now offline and the front page stored at archive.org doesn't look informative. So my only source of Manjiang data is eastling.org. Since Manjiang and the other Chinese varieties at that site aren't northwestern dialects, I wouldn't expect them to contain any nonstandard words that were borrowed into Tangut. Nonetheless, my interest in Chinese language diversity goes far beyond searching for cognates of potential Chinese loans in Tangut. I am always on the lookout for archaisms: e.g.,

the generic classifier 个 kɔi (< Late Old Chinese *kɑjh) has a final -i lost in some other varieties: e.g, Mandarin ge [kɤ] < kɔ, Cantonese

And oddities: e.g.,

几 'how many' is ky with an unexpected rounded vowel: cf. Mandarin gei [kej], Cantonese kej

To tie Manjiang into my recent posts on Korean, a script with letters resembling hangul has been proposed for Manjiang. I am sure that hangul was the inspiration because of correlations between shapes and points of articulation: e.g.,

Manjiang ㅁ b is labial like hangul ㅁ m

but Manjiang ㅂ is an affricate c, not a labial like hangul ㅂ p!

Manjiang ㄴ d,l,n are dentals like hangul ㄴ n,th,r

Manjiang ㄱ g,k are velar like hangul ㄱ k,kh (often romanized g, k!)

Some vowel letters look like hangul but have different sound values:

Manjiang ㅅ yu (IPA [y]?) looks like hangul ㅅ s (not a vowel!)

Manjiang ㅜ e looks like hangul ㅜ u

Manjiang ㅗ a looks like hangul ㅗ o

*I suspect "mangomo" is from Manjiang for 'Manjiang net':

?miã 'southern barbarian'

kɔ̃ 'speech'

̃ 'net'

I am not sure about the Manjiang reading of 蛮 in the name Manjiang since I got the reading miã from other words written phonetically with 蛮:

蛮齐个月日 miã sei kɔi ŋ (sic) 'several months' (月日 is listed elsewhere as ŋu - is 月日 an attempt to represent a Manjiang character with 月 'moon' on the left and 日 'sun' on the right? Or is a syllable missing?)

蛮好 miã hou 'pretty good' (< hou 'good'; same 蛮 as in the previous phrase?)

miã 'naughty' (< 'barbarian-like'?)

I wonder if Old Chinese 蠻 *mron 'southern barbarian' and 閩 *mrən 'Min' are cognates - and borrowings of two variants of a pre-Chinese ethnonym. Could there be any connection with the Mien? (Names can take on lives of their own, so shared names do not necessariliy entail shared ancestry: e.g., the Huns and the Hungarians). I blogged about the Mien language two years ago. OTHER INCARNATIONS OF THE PHOENIX TREE

While looking through Lee and Ramsey's (2000) The Korean Language tonight, I found an obsolete Korean word 머귀나무 mŏgwi-namu 'paulownia / Chinese parasol tree / phoenix tree' (< Middle Korean məkuy) on p. 309 that reminded me of a sequel to "Why Isn't 梧 My Tree the 同 Same?" that I wrote on October 27th last year but never posted until now:

- start of previously unpublished entry -

Last night [09.10.26], I realized that Tangut

xə thwo 'phoenix tree'

might be borrowed from a Tangut period northwestern Chinese colloquial word like 囗桐 *xəthũ (or perhaps even xəthwo) corresponding to literary 梧 桐 *ŋgothũ. (I use 囗 to represent a syllable that has no Chinese character associated with it.) Unfortunately, I left my copy of 汉语方言词汇 Chinese Topolect Vocabulary back in Hawaii, so I looked for 梧 桐 in the very small dialect database at eastling.org and did not find anything like 囗桐 *xəthũ:

Branch of Sinitic Language 'phoenix tree'
N/A Middle Chinese 梧桐 *ŋodoŋ
Mandarin Standard Mandarin 梧桐 wutong
Yue Cantonese 梧桐 ngtung
Hakka Hakka dialects on Taiwan 梧桐 ngtung, except for 詔安 Zhaoan 梧 桐 mtung
Southern Min Taiwanese 梧桐 ngôotông
Eastern Min 泰順蠻講 Taishun Manjiang dialect 梧桐 ŋ̍təŋ
Wu 松陽 Songyang dialect 梧桐樹 nuədəndʑiəɯ (樹 dʑiəɯ = 'tree')
蒲門 Pumen dialect 梧桐 ŋɔdoŋ
龍泉 Longquan dialect 梧桐 uəɯtəŋ

I added the Taiwanese and Hakka forms from the Taiwanese government's online dictionaries. Notice that even though all the words are cognates, they don't sound much alike, especially when tones are taken into consideration. For example, although most of the Hakka forms are segmentally identical, they have different tones:

四縣 Sixian: low level tone + low level tone

海陸 Hailu and 饒平 Raoping: high level tone + high level tone 

大埔 Dabu: low rising tone + low rising tone

詔安 Zhaoan: high falling tone + high falling tone

If differences in tones could be compared to segmental differences, to Hakka ears those words would sound as alike as Dutch heet and English hot. Dutch heet sounds like English hate but is in fact cognate to hot. (Dutch god [xɔt] sounds almost like Eng hot but is 'god'.)

Middle Chinese *ŋodoŋ is the source of Vietnamese 梧 桐 ngô đồng and Korean 오동나무 (梧桐나무) odong namu. 나무 namu is the native Korean word for 'tree'.

Japanese has a native word 青桐/梧桐 aogiri, literally 'green paulownia' and unrelated to the Chinese forms despite the shared character(s) (梧) 桐 in the spelling.

- end of previously unpublished entry -

Tonight (2.28), I checked the index of 汉语方言词汇 Chinese Topolect Vocabulary which doesn't list 梧 桐 'phoenix tree'. Oh well.

The native Korean word for 'phoenix tree', 머귀나무 mŏgwi < Middle Korean məkuy, superficially resembles the Sino-Korean word 마귀 (魔鬼) 'demon', but the two cannot be related since the vowels of the first syllables don't match and the pitches of the second syllables didn't match in Middle Korean (high in 'phoenix tree', rising in 'demon'). (But some Middle Korean pitches for Chinese loans can be irregular: e.g., 신 [神] sin in 귀신 [鬼神] kuysin 'ghost' is high rather than the expected low.)

Heinz Insu Fenkl noted the similarity between the od- of the Sino-Korean word for 'phoenix tree' and oda 'come':

In Korean, the paulownia tree is called odong namu, playing as a pun on the word oda, which means “to come.” In one of the traditional mask dances, there is a scene in which one of the characters in the section called “The Yangban Dance” praises his precious son, comparing his value to that of gold, silver, and jewels. One part of his song lists several trees, punning on their names, including the line: “Oda kada odong namu,” which translates as “Coming and going paulownia tree.” The paulownia tree is also the only tree on which the mythical phoenix is said to alight. The Chinese phoenix is generally associated with the imperial family, especially the empress (which is why it is also known as the “Empress Tree”), and in iconography is often paired with a dragon. Paulownia wood was also traditionally used for coffins. Mourners at a traditional Korean funeral carried a paulownia cane if the mother had died. All in all, quite a loaded symbol.

I first read Fenkl's translations of Korean comics in 1990, never guessing I'd ever cite him in a linguistic context.

Oda 'come' is the modern pronunciation of the Old Korean *ota commonly assumed to be the reading of 來如 in poems like 風謠 Phungyo 'Wind Song':


I discussed the problem of reading the graph 如 last Friday.

Tangut fonts by Mojikyo.org
Tangut radical font by Andrew West
All other content copyright © 2002-2010 Amritavision