In "Six's Prefixes", I proposed *ruk as one of several variants of the Old Chinese word for 'six'.  This variant would have become Middle Chinese *lowk, but no such MC reading is recorded in the rhyme dictionary tradition.  This does not mean that *lowk did not exist. On the contrary, the obscure standard Mandarin reading pronunciation lu (used in names) looks like it came from MC *lowk rather than the MC *luk (< nonemphatic OC *ruk) that is the only reading in the rhyme dictionary tradition.  (It is MC *luk that is the ancestor of Md liu 'six'.*)

In the rhyme dictionary Guangyun, there is only one graph with the MC reading *lowk: 濼.  Guangyun lists six more MC readings, and Jiyun has two more.  The Taiwanese character variant dictionary lists four Mandarin readings and the CUHK Cantonese dictionary lists four Cantonese readings.  I found a fifth Mandarin reading in Xinhua zidian (1971 ed.).  These readings have a wide variety of initials.  Yet presumably the fact that they all correspond to a single graph indicates that they must have shared some common phonetic core in the past.  I believe that core was OC *l-k(w):

*Cʌ-luk*lowkname of a riverprefix shielded *l- from becoming *d- before disappearing; OC *luk could be a zero-grade form of root *l-kw
*Cʌ-ləkw or *Cʌ-lok*lokprefix shielded *l- from becoming *d- before disappearing
*Cɯ-lakw*jɨakprefix deemphasized root before disappearing; prefix was gone by the time *l- > *j-
*Cʌ-r-lVkw*lœkprefix shielded *r-l- from becoming *ɖ- before disappearing; *V could have been *u, *o, *ə, or *a
*p-hləkw or *p-hlok*phokthe aspiration could also have come from a *k (cf. the next word below) but there is no evidence for a *k in the other variants of this river name
*p-k-lak*phakpok7po1, po4lakecognate to 澤 OC *r-lak 'pool, pond'?; *-k- and final *-k are implied by alternate spellings 氵+革 and 氵+雨+革 with the phonetic 革 OC *r-kək > later *krək; the alternate spelling 泊 with the phonetic 白 *brak also points to final *-k
*N-p-(k)-lak(*bak)bok8bo2not in the rhyme dictionary tradition but implied by the Ct and Md readings
*r-p-lak(*pæk)baak7not in the rhyme dictionary tradition but implied by the Ct reading
*lakw*jɨaka kind of medicinal grasscognate to 藥 OC *lakw 'medicine'?

All of the above forms can be traced to three roots:

*l-kw 'name of a river'

why so much vowel variation?

I wrote*-kw as *-k after rounded vowels:

*l-u-kw or *l-zero-kw > *luk

*l-o-kw or *l-ə-kw > *lok

was the unknown consonant in the prefixes *p-?

*lak 'lake'

cannot be conflated with 'name of a river' due to weak semantics and a final *-k; I know of no other roots with *-k ~ *-kw alternation after *a

*l-kw 'medicine'

the function of the *a ~*e vowel alternation is unknown

*It may seem strange to say that MC *luk became Md liu [ljow] instead of lu, but in late MC, *u became *iw, and in very late MC, *o(w) became *u:

*ow > *u > *iw

Hence *lowk became Md lu.

Last night, I though that late Ming Md loʔ was from MC *lowk, but I have since discovered that MC *-uk lowered to -oʔ in this dialect.  Hence loʔ may be from MC *luk after all. 六 SIX'S PREFIXES

The descendants of Proto-Sino-Tibetan *rVk 'six' have various prefixes of unknown function*:

A labial stop prefix: Lushai paL-rukL

A dental stop prefix: Written Tibetan drug, Lepcha tărók, Mikir thrókdrok2

A velar stop prefix: Written Burmese khrok, Jingpo kruʔ55, Takpa grok

gDong-brgyad rGyalrong kɯ-tʂɤɣ*kə-trɔk (Jacques 2004: 294) has both a velar and a dental prefix.

Tangut tɕhjiw may be from *k-t-rik, with *k- leaving a trace as aspiration.

is there any other language in this area with -i- in 'six'?

For years, I reconstructed the Old Chinese word for 'six' as *mə-ruk (now *mɯ-ruk in my current system) because its graph GSR1032a  六 is the phonetic of GSR1032i 睦 *mruk 'concord', though this is not obvious from the modern forms of the graphs.  However, a voiced nasal prefix *m- conflicts with

- probable borrowings from Chinese: Proto-Tai *xrok (not *hr-) and Proto-Hmong-Mien *kruk (unless these are loans from a *k-prefixing Tibeto-Burman language, or there was prefixing variation within Chinese)

- Proto-Min *hlok; *hl- may originate "from an earlier cluster consisting of a voiceless stop plus *l" (Norman 1973: 233)

was that voiceless stop a velar, matching the velars in the southern proto-languages that borrowed the word from Chinese?

The "Maitreya" hypothesis allows me to derive Middle Chinese *luk 'six' from OC *ruk*truk, which conveniently resembles Written Tibetan drug.  Thus the voiceless stop prefix that devoiced Proto-Min *hl- might have come from *t-r-.  However, *t-r- still does not match the PT and PHM forms.  I wonder if

- PM and PT reflect late OC *hrouk from an unprefixed emphatic *hruk [χʀʊʕq]

voiceless *hr- could be from an earlier *s-r-

- most (?) non-Min varieties of Chinese reflect a prefixed nonemphatic *Cɯ-ruk with a voiced-initial root which lost its prefix(es) (?*mɯ-, ?*kɯ-, others?)

2.2.1:48: Two other OC forms are needed to account for

Proto-Hakka *hliuk < OC *hruk < *Cɯ-ruk

a voiceless prefix that dropped in the mainstream fused with the root initial and devoiced it

Mandarin -i-less forms: e.g., late Ming Md loʔ and the obsolete standard Mandarin reading pronunciation lu which reflect OC *ruk [ʀʊʕq]

Cantonese luk (why does Schuessler [2007: 362] write this as lok?) could be from OC *ruk or *ruk, since OC *-uk and *-uk have merged in Cantonese.  However, if Sino-Vietnamese lục was borrowed from early Cantonese, then Ct luk is probably from OC *ruk. OC *ruk would have become early Cantonese *lok, which would have been borrowed into Vietnamese as lộc.

OC *hruk [χʀʊʕq] and OC *ruk [ʀʊʕq] may go back to a single prototype:

*hruk*sruk*sʌ-ruk*-ruk (the prefix fused with the root initial)

*ruk < *sʌ-ruk*-ruk (the prefix dropped after triggering emphatic harmony)

*I cite forms from Schuessler (2007: 362) except for Li Fang-kuei's Proto-Tai from Proto-Tai'o'Matic, rGyalrong, and Tangut.  Proto-Hakka (O'Connor 1976) and late Ming Mandarin (Coblin mss.) are quoted from Schuessler's 2001 ms. "Later Han Chinese".  See Starostin's database for more Sino-Tibetan words for 'six'. 知 (K)NO(W) EXCEPTIONS?

According to my "Maitreya" proposal, Old Chinese *(N-)t-r- became Middle Chinese *l- rather than the MC retroflex stops *ʈ- and *ɖ- as others have proposed. Since MC *ʈ-graphs usually have OC *t-/*t-phonetics, I assumed that all MC *ʈ- came from *r-t-sequences: e.g.,

哲 MC *ʈɨet < OC *r-tat 'wise'

with 折 MC *tɕiet < OC *tet 'break off; decide' as phonetic (and its root?)

(is 'wise' < 'capable of analysis' < 'breaking things apart' and/or 'decisive'?)

注 MC *ʈuəh, *tɕuəh < OC *r-to(ʔ)-s, *to(ʔ)-s 'pour'

with 主 MC *tɕuəʔ < OC *toʔ 'master' as phonetic

著 MC *ʈɨəʔ/h < OC *r-tak(s) 'to place'

with 者 MC *tɕjaʔ < OC *tjaʔ 'that which' as phonetic

OC *t- in these words corresponds to foreign t-:

哲 OC *r-tat 'wise' : Written Burmese tat 'know' (Schuessler 2007: 609)

注 OC *(r-)to(ʔ)-s 'pour' : Written Tibetan chu 'water', H-chu-ba 'ladle water' (< ?*tj-; semantically questionable; Bodman 1980: 628)

著 OC *r-tak(s) 'to place' : Written Burmese thah 'to place' (Pejros and Starostin 1996 2: 5)

However, both internal and external evidence points toward *t(V)-r- as an another source of MC *ʈ-:

OC-internal doublets (Schuessler 2007: 82):


糧 MC *lɨɑŋ < OC *raŋ

粻 MC *ʈɨɑŋ < OC *t-raŋ


壟 or 隴 MC *luoŋ < OC *roŋʔ

冢 or 塚 MC *ʈuoŋ < OC *t-roŋʔ


豬 MC *ʈɨə < OC *t-ra

豭 MC *kæ < OC *k-ra

MC *ʈ- corresponding to foreign (C)r-:

知 MC *ʈiə < OC *tre 'know' : Written Tibetan rig-pa 'know' (Schuessler 2007: 614)

張 MC *ʈɨɑŋ < OC *t-raŋ 'stretch' : Proto-Lolo-Burmese *raŋ 'draw, pull, drag', Written Burmese kraŋʔ 'tense, tight' (Schuessler 2007: 606)

Are MC *ʈ-words like 知 'know' exceptions to the OC *t-r- > MC *l-rule? On the contrary, I think their retroflex stops are regular reflexes of OC *tV-r- with late prefix-root initial fusion:

Root initialDevelopmental pathStage 1 OCStage 2 OCStage 3 OCMCExamples
*r-Prefix loss*tV-r-*r- or *tV-r-*r-*l-糧, 壟, 隴, 勒?, 力??, 離?, 蓮??
Early prefix-root initial fusion*t-r-
Late prefix-root initial fusion*tV-r-*tr-*ʈ-粻, 冢, 塚, 豬, 知, 張
*t-Early prefix-root initial fusion*rV-t-*r-t-*rt- > *tr-哲, 注, 著
Late prefix-root initial fusion*rV-t-
Prefix loss*t- or *rV-t-*t-*t- (< *t-); *tɕ- (< *t-) any graph that has an MC retroflex-initial and a MC dental or palatal-initial reading with no apparent semantic difference: e.g., 注?

(For simplicity, I have ignored emphasis in the OC columns: e.g., *tV-r- represents both emphatic *tʌ-r- and nonemphatic *tɯ-r-.)

In stage 1, all prefixes had vowels.

In stage 2, some prefixes were dropped without a trace:

*tV-r- > *r-

*rV-t- > *t-

Some of the remaining prefixes lost their vowels:

*tV-r- > *t-r-

*rV-t- > *r-t-

In stage 3, *tr- became *r-, and a new *tr- developed from earlier *tV-r- and *r(V)t-:

*tV-r-, *r(V)-t- > *tr- > *r-

In MC, *r- became *l-, and the new *tr- became *ʈ-. MAITREYA IN ALEXANDRIA

In "What's Hr-ong with Ṭh-is?", I hypothesized that the MC reflex of OC *hr- was *ʂ- instead of *ʈh-. If this hypothesis is correct, 离/魑 MC *ʈhiə 'demon' could not have come from OC *hraj, which would have become MC *ʂiə. I think 'demon' was originally OC *t-hraj. 离 is phonetic in sinographs for various syllables beginning with MC *l- < OC *r-.

If OC *t-hr- became MC *ʈh-, I would expect OC *t-r- to become MC *ʈ-. But I don't know of any MC *ʈ-words written with *l- (< OC *r-) phonetics. This implies that OC *t-r- became something other than *ʈ-. I think it may have merged with OC *r- to become MC*l-:

OC *r- > MC *l-

OC *(N-)t-r- > *(n)dr- > *ɖ- > *r- > MC *l-

Two transcriptions imply that some MC *l- may be from an earlier *(N-)t-r-:

彌 勒 'Maitreya'

Late OC *mie rʌk

Coblin (1983: 242) compared this to Agnean metrak

I cannot understand why -tr- would have been borrowed as *-r-. Hence I would reconstruct 勒 with some sort of rhotic cluster or retroflex stop (*tr-, *dr-, *ɖ-) in late OC going back to early OC *t-r-.

烏 弋山離 'Alexandria'

OC *ʔa lək ksan r(j)aj

Although it's possible that the third graph in 'Alexandria' began with *r-, an initial *(N)dr- (< *N-t-r-) matches the original -(n)dr- and is not far from the *t-hr- of its phonetic 离 'demon'.

The phonetic of 勒 OC *t-rək 'bridle' is 力 'strength' which was borrowed twice into Vietnamese:

lực (< MC *lɨk)

sức (< late OC *C-rɨk < OC *C-rək; *C- was voiceless)

力 may have had initial *tr- like its derivative 勒 OC *t-rək.

Here are two examples of *tr- becoming s- in native Vietnamese words from Pejros' database:

'sycamore': Viet sung : Ruc taruuŋ

'kingfisher': Viet sáo : Ruc traawtraaw

(but Muong khawkhaw implies *kr-)

Note that *pr- and *kr- also became Viet s- in native words, so 'strength' may have been late OC *prɨk or *krɨk from an earlier OC *pɯ-rək or *kɯ-rək. However, a *pɯ-rək or *kɯ-rək would have been a poor choice for the phonetic of OC *t-rək 'bridle'.

(1.31.00:52: Baxter [1992: 473] compared 'strength' to 逼 OC *prək 'urge' [i.e., apply force]. Could this be a compressed derivative of *pɯ-rək? If so, then Viet sức may be from a late OC *prɨk.

It is also possible that Viet s- in 'strength' reflects a native Vietic prefix added to a borrowed OC root with *r-.)

Some phonetic series consisting solely of MC *l-graphs may originally have been *t-r-series. Besides GSR23 离 and GSR928 力, a third possible *t-r-series is GSR213 連. All of its members had MC *l-, but an Old Sino-Vietnamese borrowing implies an original cluster:

213d 蓮 'lotus':

liên (< late MC *ljen)

sen (< late OC *C-ren; *C- was voiceless)

The initial of the OC word for 'lotus' may have been *t-, though there is no way to confirm that it was not some other voiceless stop such as *p- or *k-.

(1.31.1:01: The possibility that Viet sen is from a native Vietic prefix added to a late OC *ren cannot be ruled out.)

Next: What is the ʈ-rue origin of MC *ʈ?

1.31.1:39: Comparative Vietic evidence in Pejros' database points toward *kr- in 'strength':

Viet sức < *C-rɨk

Phong (a dialect of Hung?) khlɨk

Ruc khrɨɨk

This *kr- conflicts with my *t-r- proposal and Baxter's proposal to link 'strength' to 逼 OC *prək 'urge'.

It is still not clear whether the *k- is a Chinese prefix or a Proto-Vietic prefix. In any case, it cannot be part of the original Chinese root. It would make no sense to write 勒 OC *(t)rək 'bridle' with a *kr-phonetic as opposed to a *r-phonetic.

One unlikely possibility is that the *kr- of 'strength' reflects a *k- prefix (of whatever origin) added to a late OC *r- (or *dr- or *ɖ-) from early OC *t-r-. The 'Maitreya' transcription is based on a late 2nd century northern OC dialect. Did *t-r- became *r- earlier in the southern late OC dialect that Proto-Vietic borrowed 'strength' from, or did the borrowing of 'strength' postdate the 2nd century? WHAT'S HR-ONG WITH ṬH-IS?

in "Ryhor u Ryme", I expressed doubt about the sound change

Old Chinese hr- > Middle Chinese *ʈh-

I have never seen a similar change in any other language. Moreover, it is unlike the changes that affected the other OC nonemphatic liquids:

OC *r- > MC *l-

OC *l- > MC *j-

OC *hl- > *hj- > *ç- > MC *ɕ-

Fortition was a characteristic of the changes that affected OC nonemphatic liquid clusters and most OC emphatic liquids:

OC *rl- > MC *ɖ-

OC *rhl- > MC *ʈh-

OC *r- > MC *l-

OC *l- > MC *d-

OC *hl- > MC *th-

OC *rl- > MC *ɖ-

OC *rhl- > MC *ʈh-

Could *hr- have developed into a fricative like *hl-?

OC *hl- > *hj- > *ç- > MC *ɕ-

OC *hr- > *ʂ-

If the above change occurred, then MC *l- ~ *ʂ- alternations such as

*lɨh 'clerk' ~ 史 *ʂɨʔ 'scribe; history'

could have originated as OC *r- ~ *hr- alternations (which in turn may have come from even earlier *r- ~ *s-r- alternations):

*rəʔ-s ~ 史 *hrəʔ < *s-rəʔ

After early OC *sr- fused into *hr-, a new *sr- developed from earlier presyllables:

*sʌ-r- > *sr-

*sɯ-r- > *sr-

*sr- merged with *rs-/*rs- and *hr- into MC *ʂ-.

Early OC *sr- became MC *dz- (Sagart 1999: 69).

Sagart derived the initials of MC *tsh- words with *r-roots from *s-hr-. If *hr- is from *s-r-, then MC *tsh- could go all the way back to early OC *s-s-r-.

But a reduplicated prefix seems improbable. Could MC*tsh- be from OC *s-dz-, with a new *s- prefix added onto an initial whose original *s-prefix was no longer recognizable?

*r- (root initial) > *s-r- (layer 1 prefix) > *dz- > *s-dz- (layer 2 prefix) > *tsh-

A velar or uvular prefix could have been another source of aspiration:

*k-s-r- > *x-dz- [χdzʕ] > *tsh-

cf. Mawo Qiang which has initial χts- (but has ʁdz- instead of χdz-)

Did stop prefixes fuse with a *z- to form affricates?

*s-r- > *z- > *dz- > MC *dz-

cf. OC *sl- > MC *z-

and OC *sl- > MC *dz-

*C-s-r- > *C-z- > *dz- > MC *dz-

*C- = a stop prefix (*ʔ-, *q-, *k-, *t-, *p-)

In any case, these MC *dz- and *tsh- words were written with *ts-phonetics rather than*r-phonetics, suggesting that their affricate initials developed before the invention of sinography.

Next: Maitreya in Alexandria. RYHOR U RYME (translation*)

The patronymic of Belarus' president is Ryhoravič 'son of Gregor'. In Belarusian, *g became h, even before -r-: e.g.,

?'layer': hrada (cf. Rus grjada)

'mire': hraz' (cf. Rus grjaz')

'mushroom': hryb (cf. Rus grib)

'mane': hryva (cf. Rus griva)

So why isn't 'Gregor' Hryhor in Belarussian instead of Ryhor? The medial *g lenited to h as expected, but the initial *g has vanished without a trace. The only version of 'Gregory' beginning with r- on the Wikipedia entry for the name is Finnish Reijo. (Until recently, Finnish had no consonant clusters or g.)

Belarusian r is always 'hard' (nonpalatalized) and cannot be followed by j or i. This is why *rja and *ri have become ra and ry in the hr-words above.

If Belarusian is monorhotic, Russian is birhotic, since it has both hard r and 'soft' (palatalized) r'.

In Polish, soft *r' seems to have merged with hard ż as [ʒ] and is spelled rz. That spelling and its pronunciation remind me of the reflexes of earlier Vietnamese *r which vary from [r] to [ʐ] to [ʒ] to [z].

I still suspect that the z-like consonants of Tangut may have been an r-like retroflex [ʐ] and an l-like lateral fricative [ɮ]. This would explain why the 'zetic' consonants were grouped with the liquids l lh r in chapter IX of Homophones.

In Czech, soft *r' seems to have become ř. Is this sound really unique to Czech?

Upper Sorbian also has a letter ř, though this is pronounced [ʃ] and is currently homophonous with š. (Cf. Polish rz which can be [ʃ] after a voiceless obstruent and in word-final position.) I presume the orthography reflects an earlier period in which US ř was like Czech ř and was distinct from š.

Lower Sorbian is birhotic like Russian with hard r and soft r' (written with an acute accent as ŕ).

Slovak also has the letters r and ŕ, but it is monorhotic in terms of hard and soft since ŕ is a long syllabic r, not a soft r.

Like Slovak, Serbo-Croatian has nonsyllabic and syllabic r but no hard-soft distinction for r.

The Old Chinese distinction between 'emphatic' and 'nonemphatic' has some parallels to the hard/soft distinction in Slavic. This was first pointed out by Jerry Norman (1994), the originator of the emphatic hypothesis.

As far as I know, no Chinese language has distinct reflexes for voiced emphatic *r- and nonemphatic *r-: they generally merged as l-. (Baxter [1992] proposed that OC *rj- [equivalent to my nonemphatic *r-] sometimes became j-, but see Sagart's [1999: 41-42] arguments for reconstructing these exceptional cases with *l-.)

On the other hand, Old Chinese voiceless emphatic *hr- and nonemphatic *hr- may have had different reflexes.

Sagart (1999: 41) proposed that *hr- merged with *xr- as Middle Chinese *x. I think this is likely if *hr- were phonetically a voiced uvular approximant [ʁ] which would be close to a voiceless uvular fricative *x [χ].

According to Sagart, OC *hr- hardened to a MC retroflex stop *ʈh. This fortition has no parallels among the other nonemphatics which (a) remained the same, (b) palatalized, or (c) lenited. I suspect that these instances of *ʈh orignated from *C-hr-clusters. This hypothesis needs further investigation.

*My attempt to write 'Gregor in Rome' in Belarusian. U is cognate to Russian v 'in'. I don't know why the Slavic names for 'Rome' have a high vowel instead of a mid vowel:

Be Rym (with *i > y to match the hardened initial; see the discussion of Hryhor above)

Uk Rym (*i > y also occurred after initials other than *r, unlike Be)

Ru Rim

Po Rzym

Cz Řím

Slovak Rím

Srb/Crt, Slovenian, Mac, Bu Rim

(What's the Church Slavonic name, and why wasn't this name borrowed as Roma?)

1.29.00:51: Slovak Rím [riim] with its long vowel reminds me of (modern standard?) Arabic Riim 'Rome', mentioned by Kaye (1987: 671) as a variant of Ruum. (The Arabic Wikipedia entry spells the name as روما Ruumaa.) The -ii- of the Arabic form originated from -uu- via an intermediate stage *-üü-. I doubt that the Slavic forms have anything to do with the Arabic -ii- form. -L-KSAND-R

In my last post, I hypothesized that Czech final -e in ulice came from an earlier *-ja.

The name of Belarus' president exemplifies the opposite change: Aljaksandar corresponds to Russian Aleksandr and Ukrainian Oleksandr. It seems that unstressed *e has broken to ja in Belarusian. Here are examples from pravapis.org's Swadesh list:

'split'raz'dzjalic' razdelit'

Stressed e remained unchanged:

'small': malen'ki (cf. Rus malen'kij)

'person': chalavek (cf. Rus chelovek)

cha- can be regarded as ch-ja-; its a corresponds to Russian unstressed e

The *e > *ja shift doesn't account for jany 'they', which looks like the Belarusian cognate of Russian oni. (I would have expected ani.) Moreover, 'he' is jon rather than the expected on (cf. Russian on). I suspect that this prothetic j- was added so that all third person pronouns began with j- or i-. Compare the Belarusian pronouns with their Russian cognates:

CaseBe 'he'/'it'Ru 'he/it'Be 'she'Ru 'she'Be 'they'Ru 'they'
Nom.jon 'he', jano 'it'on 'he', ono 'it'janaonajanyoni

(Russian adds n- if a third person pronoun is preceded by a preposition: jego > nego. The prepositional case is always preceded by a preposition, so its n-form is its citation form.)

(For ease of comparison, I have used the same transliteration system for both languages. Hence my Belarusian transliterations do not always match the Łacinka spellings: e.g., my ix is ich in Łacinka.)

I found one case of Belarusian unstressed ja corresponding to Russian unstressed i:

'sit': Be sjadzec', Ru sidet'

Ukrainian siditi also has i. Vasmer listed the Belarusian form as sidzec' which is what I would have expected. However, I could not find sidzec' in the dictionaries at slounik.org. Be -ja- may not be entirely unexpected since other Slavic languages have -e- in the first syllable: e.g., Polish siedzieć (si- = [ɕ], not s + i), Czech sedět, Bulgarian sedja.

Tangut fonts by Mojikyo.org
All other content copyright © 2002-2008 Amritavision