I wanted to continue my series on nonstandard Korean e (part 1 / part 2 / part 3), but I'm not feeling well right now, and this Tangut character is on my mind:


1tsəuʳ 'to suffer from an illness, to fall ill, to be ill' =

left and center of 1tsəuʳ 'to stain, smear' +

right of 1ʔieu 'disease'

At first I thought that 1tsəuʳ 'to suffer from an illness' could be a metaphorical use of 1tsəuʳ 'to stain': to be stained is to be sick.

But then I found out that 1tsəuʳ 'to suffer from an illness' might only be attested as the first half of

1tsəuʳ 1bəi

translated by Li Fanwen (2008: 674) as a verb 患病 'to suffer from an illness' and Kychanov and Arakawa (2006: 674) as a noun 'disease, epidemic'. Unfortunately K&A's dictionary has no examples from texts, and all the examples in Li are from dictionaries: e.g., two Tangraphic Sea entries (13.111, 54.222) list 1tsəuʳ 1bəi as one of the


1lɨəəʳ 2liẹ 1mi (1dzõ) 2ŋwe

'four great not (harmony) harmonious'

'Four Great Disharmonies'

K&A (2006: 673) defined 1bəi by itself as 'illness, disease, suffer, disharmony of four elements - earth, water, fire and air'. Unfortunately I cannot confirm that 1bəi is a word because it too is apparently only attested as half of 1tsəuʳ 1bəi.

1bəi superficially resembles Early Middle Chinese 病 *bɨaŋʰ 'disease', but the rhyme does not match. A hypothetical Tangut borrowing of *bɨaŋʰ would be *bie or *biẽ, not 1bəi.

1tsəuʳ 1bəi could be

a. a verb + noun construction: 'smeared disease'? But is (or was) 1bəi really an independent word 'disease'? (In theory 1bəi could have been a noun that went out of use except in 1tsəuʳ 1bəi.

b. a verb + verb construction: 'to be smeared and ... (with disease)'? Again, this assumes 1bəi is (or was) an independent word.

c. an unanalyzable disyllabic word that may be part of the so-called 'ritual language': a colloquial, substratal non-Sino-Tibetan word?

If 1tsəuʳ 1bəi is (c), that would explain why it only occurs in dictionaries. It would have only occurred outside dictionaries as a gloss for a literary word having to do with illness, as there are no known entirely colloquial texts (if my interpretation of the 'ritual language' is correct). NONSTANDARD KOREAN E: REVERSAL OR RETENTION?

Martin (1992: 39) referred to nonstandard Korean e corresponding to standard (e.g., in chhene for 處女 chhŏnyŏ 'maiden') as a product of "reversal". In his romanization of Korean, [e] is ey and [jə] (my yŏ) is ye, so he described the phenomenon as ye becoming ey, "especially in the north". Ey was originally [əj]. I assume Martin thought [jə] reversed to [əj] before [əj] monophthongized to [e] because [jə] becoming [e] is not reversal (except in terms of his romanization):

Stage 1 [əj] [jə]
Stage 2: reversal (and merger) [əj]
Stage 3: monophthongization [e]

Standard Middle Korean is at stage 1. Is there any evidence for stage 2?

Martin (1992: 47) mentioned that

King (1990) has examples that argue for ye [jə] > yey [je] > ey [e] with loss of the initial glide after the fronting took place, rather than metathesis of the glide.

Unfortunately I have not seen this work and I cannot identify it as it is not in Martin's bibliography.

If I am correct about nonstandard [e] partly  retentions, the process in King would have happened in reverse in standard Korean

[e] > [je] > [jə]

while original [e] merged with secondary [əj] in some dialects:

Early Korean *əj *e *jə
Nonstandard Korean [e] [jə]?
Standard Korean [e] [jə]

According to Wikipedia, in the northwestern Phyŏngan dialect (emphasis mine),

'야, 여, 요, 유' 등은 자음 뒤에서 '아, 어, 오, 우'로 실현되기도 한다(차포[차표])

[the letters] <ya, yŏ, yo, yu> etc. are realized as [a ə o u] after consonants: [tsʰapʰo] for standard [tɕʰapʰjo]

(Northern alveolars corresponding to standard and southern palatals are not specified in the article. See the reference to Kim Yŏng-bae 1984 in Martin [1992: 48] for Phyŏngan [ts].)

Minimal pairs of [Ce] : [Cə] corresponding to (premodern) standard [Cjə] may reflect an earlier *e : *jə distinction: e.g.,

Gloss Early Korean Phyŏngan (Lee and Ramsey 2000: 322) Standard
that *te [te] (if not a typo for [tə]) [tɕə] < [tjə]
temple *tjər [təl] [tɕəl] < [tjər]

Japanese tera 'temple' is probably a borrowing from a Paekche *ter(a) with a very early secondary *e from original *jə. Vovin (2007: 76) reconstructed Proto-Korean *tjara whose *-ja- underwent metathesis when borrowed into Jurchen as

<tai.ra.an> = tairan 'temple'.

Lee and Ramsey (2000: 324) listed a similar minimal pair in North Kyŏngsang to the south:

Gloss Early Korean North Kyŏngsang Standard
older brother *hjə(ŋ) [sə] < *hjə [hjəŋ]
tongue *het [se] < *hje (if not a typo for [sə]) [hjə]

'Older brother' is usually assumed to be a Sino-Korean reading of 兄 'older brother', though Vovin (2013: 225) proposed that it is actually a native Korean word. I think standard [hjəŋ] is Sino-Korean, but the North Kyŏngsang form may be from an unrelated native soundalike *hjə sans final nasal that is related to (1) the Koguryŏ word for 'older brother' transcribed in Chinese as 奢 *ɕæ and (2) Western Old Japanese se 'older brother'.

Both Western Old Japanese se and sita < *seta 'tongue' (see previous entry) have s- corresponding to early Korean *hj-. Perhaps they are both loanwords from Paekche which had *ɕ- from *hj-. Koguryŏ to the north also apparently had *ɕ- from *hj-.

I am not entirely confident that nonstandard [e] is a retention, but in any case I think we should be careful not to assume that more prestigious forms also happen to be more conservative. P_YANG CHHENE

Googling for 피양  Phiyang, a short variant of 'Pyongyang', led me to the 1969 song 피 양체네 Phiyang chhene sung by 宋海 Song Hae. The title corresponds to standard Korean 平壤處女 Phyŏngyang chhŏnyŏ 'Pyongyang Maiden'). The e in chhene caught my eye because for years I have agreed with Leon Serafim's 1999 proposal to derive at least some standard Korean from earlier *e. Nonstandard e corresponds to standard in

Pheyang (the other short variant) : Phngyang 'Pyongyang'

chhene : chhŏn < Middle Korean [tsʰn] 'maiden'

meniri (Phyŏngan dialect from Lee and Ramsey 2000: 330) : mnŭri 'daughter-in-law'

se (North Kyŏngsang dialect from Lee and Ramsey 2000: 324) : h 'tongue'

I assume that chhene is a northern form like Pheyang and meniri. Se is southern; see Vovin (2010: 188) for other se-forms from Choy (1978: 426).

Vovin proposed that Japanese shita 'tongue' may be a loan from Korean. The -t- of shita matches the *-t of Early Middle Korean *hjət 'tongue' (transcribed in Chinese as 蝎). I do not know why the final *-t was lost in later Korean.

Serafim (1999: 7) reconstructed Proto-Japonic *seta 'tongue' and derived it and the Korean forms from a Proto-Koreo-Japonic *xé.

Combining Vovin and Serafim's views, I think Proto-Japonic *seta may be a loan from an early Koreanic *het (phonetically [hʲet] or ]çet]?). Is the e of se-forms for 'tongue' a retention of early Koreanic *e, or has *e gone full circle in that word?

*e > *yŏ > e

Could e in Pheyang, chhene, and/or meniri also be retentions?

This page lists many variants of 'daughter-in-law'. Unfortunately they lack specific locations other than North and South Korea. The North Korean forms have ae, e, i, or in the first syllable whereas the South Korean forms only have i or yŏ.

The nonstandard and early Koreanic e in this post is not to be confused with the e of standard Korean which is from Middle Korean əj. P_YANG

According to McCune and Reischauer's 1939 article that established the Korean romanization that I use here with modifications, 平壤 'Pyongyang' [pʰjəŋjaŋ] "is colloquially pronounced P'iyang or P'eyang" (p. 52): i.e., as [pʰijaŋ] or [pʰejaŋ]. Who pronounced 'Pyongyang' that way, and is either pronunciation still current somewhere? How many other closed syllables could also be reduced to open syllables?

11.6.23:54: According to 李雲源 Yi Un-wŏn, [pʰijaŋ] is an attempt to imitate the dialectal pronunciation of 'Pyongyang', whereas actual northerners say [pʰejaŋ]. DID EARLY OLD CHINESE HAVE PALATAL STOPS INSTEAD OF ALVEOLAR AFFRICATES?

So far I have only been able to reconstruct five palatal series in Early Old Chinese, excluding my candidates for -series:

亦夜 *Cak




Middle Chinese *ts(ʰ)- ~ *s- alternations might go back to Early Old Chinese *c(ʰ)- ~ *sc(ʰ)- alternations: e.g.,

GSR series Sinographs Middle Chinese Early Old Chinese
401 桼漆 *tsʰit *cʰit (or *tsʰit?)
*sit *scʰit (or *stsʰit?)

I have no examples of *C-series with the vowels or *o or codas other than *-k, *-r, and *-t. This restricted distribution is suspicious. It contrasts with the wider distribution of the alveolar series with *ts-, *tsʰ-, and *dz-. Was the Chinese script invented at a time when the mergers of the palatal stops with alveolars and *j (below) were nearly complete?

Early Old Chinese *ts- *c- *tsʰ- *cʰ- *dz- *Nts- *Ntsʰ- *Nc- *Ncʰ- *Nɟ- *ɟ- *j-
Middle Chinese *ts- *tsʰ- *dz- *j-

Or are a lot of palatal series undetected because they lack Middle Chinese *ts- ~ *j- (< Early Old Chinese  *c- ~ *ɟ-) alternations? Until recently I would have reconstructed GSR series 399/923 with *ts- in Early Old Chinese, but now I could reconstruct it with *c-:

GSR series Sinographs Middle Chinese Early Old Chinese
Until recently Now
923 *tsɨək *tsək *cək
*tsɨək, *tsiek *tsək, *Cɯ-tsek *cək, *cek
399 *tset, *dzet *Cʌ-tsik, *Nʌ-tsik or *dzik *Cʌ-cik, *Nʌ-cik or *Nʌ-ɟik
楖櫛 *tʂɛt *rʌ-tsik *rʌ-cik
923 *tsit *tsik *cik
*dzit *N-tsik or *dzik *N-cik or *N-ɟik

I could go even further and reconstruct palatals instead of *TS-type initials:

Early Old Chinese *c- *cʰ- *Nc- *Ncʰ- *Nɟ- *ɟ-
Middle Chinese *ts- *tsʰ- *dz- *j-

I am not sure a *j- distinct from *ɟ- is necessary for Early Old Chinese, and I am uncomfortable about *c(ʰ)- and *ɟ- developing in different directions. If *c(ʰ)- became an affricate *ts(ʰ)-, why didn't *ɟ- also become an affricate *dz-?

Maybe Early Old Chinese only had a glide *j- instead of a stop *ɟ-, but an alternation between stops *c- and *ɟ- is more likely than an alternation between a stop *c- and a glide *j-.

11.5.22:35: Another possibility is that Early Old Chinese *ɟ- became Middle Chinese *dz-, whereas Early Old Chinese *j- remained intact in Middle Chinese, and Middle Chinese *ts- ~ *j-alternations are rare because the phonetic match between Early Old Chinese *c- and *j- was weak. DID OLD CHINESE PALATAL INITIALS ALWAYS CONDITION HIGHER VOWELS?

Early Old Chinese had at least two vowels in unstressed presyllables and six vowels in stressed syllables:

Height Unstressed Stressed
Higher *i *u
Lower *e *a *o

The vowels above are arranged according to phonological categories (higher vs. lower) rather than their places of articulation (*e was actually higher than *a, etc.).

Early Old Chinese permitted mismatches in presyllabic and syllabic vowel heights: e.g.,

*mʌ-rə(ʔ) 'to come' (lower + higher; see "Did Old Chinese Have *Q-odas?" for the final glottal stop)

*kɯ-raʔ 'spine' (higher + lower)

The degree of mismatch was reduced when syllabic vowels 'bent' to partly match the heights of presyllabic vowels in Middle Old Chinese:

*mʌ-rə(ʔ) > *mʌ-rəɨ (higher vowels > falling diphthongs)

*kɯ-raʔ > *kɯ-rɨaʔ  (lower vowels > rising diphthongs)

The presyllables were gone by Late Old Chinese, but the bent vowels remained:

*mʌ-rəɨ > *ləɨ

*kɯ-rɨaʔ > *lɨaʔ

Vowels in syllables without presyllables remained stable unless preceded by initials of two classes:

Uvulars caused higher vowels to become falling diphthongs:
*Qi > *Qei

*Qə > *Qəɨ

*Qu > *Qou

Palatals (see my last post) caused lower vowels to become rising diphthongs:

*Ce > *Cie

*Ca > *Cɨa

*Co > *Cuo

When writing my first draft of yesterday's post, I hadn't fully worked out how presyllabic vowels interacted with vowels after palatals, and I left out presyllables in my reconstructions with the exception of 蹟 OC *Cɯ-cek or *Cɯ-tɕek 'footprint'. At the time I thought I needed the *Cɯ- to condition the 'bending' of *e to *ie, but if that were the case, then I'd need *Cɯ- to condition the 'bending' of *a in the root variant *cak 'footprint' and other words. Earlier today I added *Cɯ- all over the place for consistency. Then I realized that it would be simpler to assume that palatals always trigger 'bending' of lower vowels unless blocked by a low-vowel presyllable (e.g., 責):

Sinograph Gloss Middle Chinese Early Old Chinese
Earlier today Now
亦, 腋 armpit *jiek *Cɯ-ɟak *ɟak
footprint *tsiek *Cɯ-cak *cak
*Cɯ-cek *cek
night *jæʰ *Cɯ-ɟaks *ɟaks
to blame *tʂɛk (not in last post) *rʌ-cek
thorn *tsʰie *cʰeks
tablet *tʂʰɛk *rʌ-cʰek
to soak *dzieʰ *N-ɟeks
recondite *dʐɛk *N-rʌ-ɟek, *Nʌ-r-ɟek, or *rʌ-N-ɟek

(For simplicity I've written all Old Chinese palatals as stops, though they might have been affricates.)

If 責 were *rʌ-cek rather than *s-trek or *r-s-tek or *s-r-tek, it could not be cognate with 謫 *trek or *(N)-r-tek 'to blame'.

In my last post, I was worried about the absence of palatal aspirates, but I now think they are needed for palatal series words like 朿 and 策 with Middle Chinese aspirates.

I propose *N-ɟ- as a source of Middle Chinese *dz- in Old Chinese palatal series: e.g., 漬.

Middle Chinese *dʐ- may similarly come from *N-ɟ- with an *r- prefix: e.g., 賾.

If there is any *n-series with low vowels that mostly become rising diphthongs, it might be a -series. Here are two I quickly found:

GSR series Sinographs Middle Chinese Old Chinese
Palatal scenario Dental scenario
1223 辱溽縟蓐 *ɲuok *ɲok *Cɯ-nok
槈耨 *nouk, *nəuʰ *Cʌ-ɲok(-s) *nok(-s)
622 *ɲiemˀ *ɲamʔ *Cɯ-namʔ
*nəm *Cʌ-ɲəm *Cʌ-nəm
*tʰəm *sʌ-ɲəm *sʌ-nəm

染, the only member of series 623, might have been *ɲomʔ or *ɲamʔ in Old Chinese (see this post on its vowel).

The one thing I don't like about reconstructing *ɲ- is that it either went full circle if it wasn't preceded by *-ʌ -

*ɲ- > *n- > *ɲ-

- or it remained intact all the way into Middle Chinese even though the rest of the palatals didn't:

Old Chinese *c- *cʰ- *NC- *ɟ- *ɲ-
Middle Chinese *ts- *tsʰ- *dz- *j- *ɲ- (not *n-!)

(The table above is simplified and excludes Old Chinese *r-clusters that became Middle Chinese retroflexes.)

11.4.22:52: Old Chinese could have been like Japanese which has palatal ch, j, and y but no phonemic ɲ. ([ɲ] is an allophone of /n/ and /N/ in Japanese. See Vance [1987: 35].)

11.4.23:24: Another parallel may be Sanskrit which has root-initial c-, ch-, j-, jh-, and y- but no root-initial ñ-. *Tɕ-RA-C-ES OF OLD CHINESE PALATAL INITIALS?

Baxter and Sagart's (2011) reconstruction of Old Chinese has a wealth of exotic pharygealized initials (e.g., *qʷʰˤ- which is in only one language in UPSID) but no palatal initials.

For years I've reconstructed a sinhgle palatal initial *j- in Old Chinese (OC). That was also the only palatal initial in Schuesser (2007, 2009) reconstruction. Schuessler (2007: 96-97) reconstructed OC *j- as a source of Middle Chinese (MC) *j- that alternated with MC *ts- in phonetic series and Tibeto-Burman* ɲ- in word families. For a long time I couldn't understand why MC *ts- and *j- would be in the same phonetic series, but last night I realized that they might be reflexes of OC palatal stops or affricates. Here's my interpretation of the phonetic and cognate sets from Schuessler (2007: 97):

酉 MC *juˀ < OC *ɟuʔ or *dʑuʔ < *N-cuʔ or *N-tɕuʔ 'tenth Earthly Branch' (but written as a drawing of a wine vessel)

phonetic in 酒 MC *tsuʔ < OC *cuʔ or *tɕuʔ 'wine'

phonetic in 庮 MC OC *ɟuʔ or *dʑuʔ 'to rot'

cf. Matisoff's Proto-Tibeto-Burman *zyaːw ~ *zyu(w) 'to rot, digest' and Written Tibetan Hju-ba 'to digest'

亦 MC *jiek < OC *Cɯ-ɟak 'armpit' (also see 腋 below)

cognate to Mru yak 'armpit'

phonetic in 跡 MC *tsiek < OC *Cɯ-cak or *Cɯ-tɕak 'footprint'

cognate to 蹟 MC *tsiek < OC *Cɯ-cek or *Cɯ-tɕek 'footprint'

and Limbu yok 'trace', Lushai hniak < *s-N-ɟ-? 'footprint'

the 責 phonetic series can be reconstructed with *c- or *tɕ-

夜 MC *jæʰ < OC *Cɯ-ɟaks 'night'

cognate to Written Burmese ña 'evening' < *N-ɟ-?

phonetic in 腋 (alternate speling of 亦 above)

允 MC *jwinˀ < OC *ɟurʔ

phonetic in 𠬍 MC *tsʰwin ~ *tswinʰ < OC *(H-)cur(-s) ~ *(H-)tɕur(-s)

*H- is a prefix conditioning aspiration; it may have been *k- or *s-

cognate to Written Burmese yun 'rabbit'

I am skeptical about the last set since OC *-r shouldn't correspond to *-n. (This isn't a problem for Schuessler whose OC *-n corresponds to Starostin's, Baxter and Sagart's, and my *-r.)

Here is a comparison of the initials of my proposal with those of Schuessler and Baxter and Sagart:

Middle Chinese Old Chinese
This site Schuessler 2007, 2009 Baxter and Sagart 2011
*ts- *c- or *tɕ- *ts- *ts- for 酒, 跡, 蹟
*j- *N-c- or *N-tɕ-
*ɟ- or *dʑ-
*j- *m.r- for 酉
*ɢ(r)- for 亦
*N.r- for 夜
*l- for 允

Alternations between *(N-)c- and *ɟ- (or *(N-)tɕ- and *dʑ-) are more clearly motivated than those between *ts- and *j- or *ts- and *N.r-, *ɢ(r)-, and *l-.

However, my proposal has at least two problems.

First, Old Chinese obstruents come in sets of three: e.g., *ts-, *tsʰ-, *dz-. But there is no evidence for a palatal *cʰ- or *tɕʰ-. Perhaps some Middle Chinese *tsʰ- go back to an Old Chinese palatal. Could Proto-Min (PM) *tšʰ- corresponding to Middle Chinese *ɕ- partly go back to *cʰ- or *tɕʰ-: e.g.,

手 'hand' PM *tšʰiuB1, MC *ɕuˀ < OC *cʰuʔ or *tɕʰuʔ?

(Cf. how Lao ຊ [s] may be romanized as x which I presume reflects an earlier *ɕ-like pronunciation of what had been *cʰ-. That aspirate goes back to an even earlier *ɟ-.)

Second, my OC *ɟ- or *dʑ- corresponds to Tibeto-Burman j-. That is phonetically plausible, but Tibeto-Burman also has voiced palatal stops or affricates distinct from the glide j-. Did OC merge Proto-Sino-Tibetan *j- with *ɟ- or *dʑ- into *ɟ-? The fortition of *j- to *ɟ- or *dʑ- is also phonetically plausible but unusual in context, as I don't know of any other fortitions prior to Old Chinese.

*11.3.21:36: I use 'Tibeto-Burman' as shorthand for 'non-Chinese Sino-Tibetan'. I do not believe in a Tibeto-Burman subgroup, so I use Proto-Tibeto-Burman reconstructions as composites of early non-Chinese Sino-Tibetan forms: e.g., I don't believe there was a Proto-Tibeto-Burman *zyaːw ~ *zyu(w) 'to rot, digest', but early non-Chinese Sino-Tibetan languages may have had similar forms.

Tangut fonts by Mojikyo.org
Tangut radical and Khitan fonts by Andrew West
Jurchen font by Jason Glavy
All other content copyright © 2002-2013 Amritavision