Did GSR 658 have a velar phonotype? Matisoff (2003: 309) wrote:

This GSR #658 has members with OC palatal, dental, and velar initials (for an example of the latter see 'die / kill'), perhaps pointing to a Proto-Chinese *ky- initial for most of them.

On the previous page, two members of GSR 658 were compared to Proto-Tibeto-Burman reconstructions. (The Late Old Chinese reconstructions are in my modification of Schuessler's [2007] system.)

PTB *m-kum ~ *m-kim 'block/pillow' : 658f 椹 LOC *Tïm 'chopping block'

Matisoff (2003) also listed PTB *'um 'block/pillow' though not in this comparison

PTB *gum ~ *kum 'die/kill' : 658q 戡 'vanquish, kill' and 651v 戈+今 'kill', both LOC *khəm

Schuessler (2007: 330) compared 658p 堪 LOC *khəm 'able to bear' with

Written Burmese kham 'receive, endure'

Mru kham 'to bear, sustain' ("Burmese loan?")

Jingpho kham31 'endure'

The rhymes are not a problem. OC *ə regularly corresponds to a in Tibeto-Burman languages. LOC *-ïm and *-əm could have sometimes come from an earlier OC *-um via dissimilation. Although Sagart (1999: 52-54) argued against reconstructing OC *-um, perhaps such a rhyme existed in a pre-OC stage (i.e., in Proto-Sino-Tibetan). Hence in PST, 'block' and 'kill' could have ended in *-um, whereas 'to bear' could have ended in *-əm. OC merged the two rhymes, unlike PTB.

Although 658f had a retroflex initial in LOC, it could have come from an OC cluster with a velars:

LOC *Tïm < OC *r-t-kəm'

However, if the above external comparisons are correct, this internal comparison might have to be rejected:

658q 戡 EOC *khəm 'vanquish, kill'

扌+ 林 LOC *ləm' ~ *lïm' < EOC *Cə-ləm' ~ *Cə-ləm-' 'kill'

Even if the unknown EOC initial consonant of 扌+林 were *k-, its *-l- corresponds to nothing in EOC *khəm or PTB *gum ~ *kum. There is no infix *-l- in OC.

But there was an *-r- prefix and/or infix. Since 扌+林 dates from the 1st century AD, there is no guarantee that an EOC-speaking scribe would have chosen an *l-phonetic. LOC *l- could come from both EOC *Cə-l- and *r- (Sagart 1999), so the word might have had *r in EOC. Moreover, Schuessler (2007: 358-359) regarded 林 as an *r-phonetic. Therefore 扌+林 might have orignated as an *r-affixed form of 'kill':

*r-kəm-' > *krəm' with variant *kərəm' > *rəm' (loss of presyllable) > LOC *lïm'

*r-kəm-' > *krəm' with variant *kərəm' > *rəm' (loss of presyllable) > LOC *ləm'

The aspiration of 戡 EOC *khəm could reflect an alternate development of earlier *r-k-:

*r-k- [ʀq] > *ʁk- > *χk- > *hk- (preaspiration) > *kh- (postaspiration)

(Schuessler 2007: 58 proposed *rk- > *kh-, citing a comparison of 磡 'cliff, bank, step' [not in Karlgren;*khə̂ms in his EOC reconstruction] to PTB *r-ka[a]m, 'id.' [though Matisoff defined this as '‘edge/bank/precipice/lip/mouth' [no 'cliff']. He explained the aspiration of 戡'kill' as reflecting "forceful action" [2007: 60].)

645a 貪 LOC *thəm, used to write a nonstandard word for 'kill', could have come from EOC *t-khəm, so it might still share a root with 658q EOC *khəm and PTB *gum ~ *kum.

(11.18.00:45: The EOC source of 658e 揕 LOC *Tïm 'to strike' might have combined the prefixes *r- and *t- found in 戡磡 and 貪 with a root *kəm: *r-t-kəm or *t-r-kəm.)

So far, not so bad. However, reconstructing all of series 658 as velar entails a lot of EOC clusters because most of that series is nonvelar in LOC:

velars (phonetically uvulars?):

LOC *kh- < EOC *kh-

LOC *ng- < EOC *ng- < *n-k-


LOC *ch- < EOC *ky-

LOC *j- < EOC *gy-

LOC *sh- < EOC *sk-

LOC *zh- < EOC *Nsk-

LOC *y- < EOC *kəl-


LOC *T- < EOC *rtk-

LOC *Th- < EOC *rtkh-

LOC *D- < EOC *rNtk-

LOC *N- < EOC *rntk-

LOC *J- < EOC *rNtsk-


LOC *ts- < EOC *sk- (Sagart 1999: 69 allowed *sk- > *ts- as well as *sh-)

LOC*s- < EOC *skh-


LOC *t- < EOC *tk-

LOC *th- < EOC *tkh-

LOC *d- < EOC *Ntk-

The only simple EOC initial in this reconstruction is *kh- in only three words (or four if 磡 'cliff' is included), and even that may have originated from an earlier *rk- cluster.

If I used 'majority rule' to determine how to reconstruct the phonotype of a phonetic series, I would vote for a dental phonotype for 658, since five of its LOC initials could have been derived from EOC simple dental initials:

LOC *ch- < EOC *t-

LOC *j- < EOC *d-

LOC *t- < EOC *t-

LOC *th- < EOC *th-

LOC *d- < EOC *d-

658a 甚 had LOC *j-, which usually came from EOC *d-, though *gy- would also be possible.

(Baxter [1992: 550] reconstructed 658a as *Gyum [gjum] with a capital *G for a voiced velar stop that unexpectedly palatalized in Middle Chinese. I believe that LOC and MC *j- is the normal reflex of EOC *gy-, so I use a lowercase *g-.)

It is far easier to derive LOC retroflexes from dentals (just add *r) than from velars or laterals which would both require at least two additional consonants: a *t to change the place of articulation, and an *r to add rhoticity:


< *rt- or *tr-

< *rtl- or *tlr-

< *rtk- or tlk-

Only four of the LOC initials of 658 could be derived from simple EOC lateral initials:

LOC *sh- < EOC *hl- (voiceless lateral, not an h + l cluster)

LOC *y- < EOC *l-

LOC *th- < EOC *hl- (voiceless lateral, not an h + l cluster)

LOC *d- < EOC *l-

LOC *zh- is usually a strong sign of an EOC root-initial lateral, but it might also originate from an EOC *md-.

11.18.00:54: The velar phonotype hypothesis could account for the remaining words with lateral cognates:

658h 斟 EOC *t-k-ləm 'ladle out'

first half of 斟酌 *t-k-ləm t-lewk 'ladle out'; the alliteration no longer works unless this expression was coined in a dialect which had simplified *t-k-l- to *t-l- or which had a version of the word without the *k- prefix

658l-m 湛 EOC *t-k-ləm 'deep' (among many other readings)

658n 黮 'dark' EOC *t-k-hləm', *kə-ləm'

658o 糂 EOC *s-k-ləm 'rice gruel with meat'

With clusters and presyllables, seemingly anything is possible. The LOC initials of 658 are compatible with a velar, dental, or lateral phonotype. I have tried to choose between the three phonotypes using two other criteria which point in opposite directions:

- simplicity: dentals win, since more LOC initials in 658 can be derived from simple EOC dentals than from simple EOC velars or laterals

- etymology: internal and external cognates have velars and laterals

Next: How could this conflict be resolved? GSR 658

is a very strange series from a Late Old Chinese standpoint because it contains both grave and acute initials

velars (phonetically uvulars?): *kh- *ng-

palatals: *ch- *j- *sh- *zh- *y-

retroflexes: *T- *Th- *D- ?*N- *J-

alveolars: *ts- *s-

dentals: *t- *th- *d-

Only glottals and labials are missing!

Did the Early Old Chinese phonotype of 658 have a velar or a dental initial? Sagart reconstructed lateral root initials for the EOC readings of 658h, 658l-m, and 658q. If Sagart is correct, then

658n 黮 'dark'

LOC *thəm' < EOC *hləm' < pre-EOC *ql-?

LOC *dəm' < EOC *(qə-)ləm'

may be cognate to

671k 黬 EOC *r-(N/'-)qləm(-') or *(N/'-)qrəm(-') 'black blot'

606k 紺 EOC *qləm(-')-s 'dark purple'

614e 晻 EOC *(r-)'-qləm-' ,*r-'-qlam-',*rɯ-'-qlam-' 'dark'

653h 暗 EOC *'-qləm(-')-s 'dark'

653i 闇 EOC *'-qləm'(-s) 'dark, hidden'

(and at least one more word I'll save for next time)

However, the LOC readings of 658 lacks three initials that are commonly associated with lateral series:

LOC *l- < EOC *Cə-l-

LOC *z- < EOC *sl-

All of the LOC initials of 658 could have come from nonlateral sources if I adopt Schuessler's (2007: 459) proposal of two sources for LOC *zh-: nonlateral *md- as well as *ml-.

I can reconstruct 658 with a lateral or a stop-inital phonotype:

GSR number



EOC phonotype *ləm

EOC phonotype *dəm




*dləm'(-s) or *Ntləm'(-s)





*dləm or *Ntləm


small furnace



*dləm or *Ntləm


reliable, to trust



*r-t-ləm(-'/s) or *t-r-ləm(-'/s)

*r-t-ləm(-'/s) or *t-r-ləm(-'/s)

strike; stab


*Tïm, *zhïm', *shïm'

*r-t-ləm or *t-r-ləm; *m-ləm-', *hləm-'

*r-t-ləm or *t-r-ləm, *m-dəm-', *s-t-ləm-'

chopping-block; mulberry (according to Jiyun; another spelling of 658i)



*rhləm(-'/s) or *hlrəm(-'/s)

*rthəm(-'/s) or *thrəm(-'/s)

walk hesitatingly





ladle out


*zhïm', jïmh

*m-ləm', *N-t-ləm'-s

*m-dəm', *dəm-s








*təm, *DEm', *tsiem, *Dïm(h), *shïm, *jïm, *yïm('), *Thïm', *dəm', *Diem', *Tïmh, *tsimh, ?*Nïm'

*t-ləm, *r-ləm-', *s-t-lam, *r-ləm(-s), *hləm, *N-t-ləm, *ləm(-'), *r-hləm-', *ləm-', *r-lam-', *r-t-ləm-s, *t-s-hləm-s, ?*r-n-ləm-'

*t-ləm, *r-N-t-ləm-' or *N-r-t-ləm-', *s-t-lam, *r-N-t-ləm(-s) or *N-r-t-ləm(-s), *s-t-ləm, *N-t-ləm, *tə-ləm(-'), *r-t-hləm, *N-t-ləm, *r-N-t-lam-' or *N-r-t-lam-', *r-t-ləm-s, *s-təm-s, ?*r-n-t-ləm-'

sunk in / deep / to soak (each meaning corresponds to each of the first three readings; I can't confirm Karlgren's third reading in a Chinese source)


*thəm', *dəm'

*hləm' < pre-EOC ?*ql-, *(qə-)ləm'

*t-hləm', *N-t-hləm'

dark; also used to write 658i mulberry ('dark fruit'?)





rice gruel with meat





able to bear





vanquish, kill


*khəm, *ngəm, *khEm', *ngəm', *JEm'

*(r-/N-)k-hləm('), *r-N-s-t-ləm

*(r-/N-)k-thəm('), *r-N-s-thəm


In the EOC phonotype *dəm column, I have only reconstructed *-l- whenever Sagart and I have identified Chinese cognates with laterals*.

Working mostly within Sagart's (1999) model of prefixation, I was forced to reconstruct a four-prefix sequence to account for why 嵁 LOC *JEm' [dʐɛmʔ]< *dzrəm < EOC *r-N-s-t-ləm has a lateral-initial phonetic.

Conversely, I had to reconstruct a three-prefix sequence to account for why 嵁 LOC *JEm' [dʐɛmʔ] < *dzrəm < *r-N-s-thəm has a dental stop-initial phonetic.

And those are not the only cases requiring multiple prefixes to reconcile LOC initials with the EOC initials (*l- or *d-) of 甚. Neither solution satisfies me.

Next: Gumming up the works.

*11.17.12:50: The following words probably have roots with laterals:

658e 揕 EOC*r-t-ləm-s or *t-r-ləm-s 'to strike'

658f 椹 EOC *r-t-ləm or *t-r-ləm 'chopping block' (< 'striking place'?)

658q 戡 EOC *k-(t)-hləm 'vanquish, kill'

cognate to

645a 貪 LOC *thəm < EOC *(?kə-)hləm

扌+ 林 LOC *ləm' ~ *lïm' < EOC *Cə-ləm' ~ *Cə-ləm-'

not in Karlgren; its phonetic is 林 655a-d

did its prefixes have initial *k- [q] and *k-?

both dialect words for 'kill' in Fangyan (1st c. AD)

658h 斟 EOC *t-ləm 'ladle out'

first half of 斟酌 *t-ləm t-lewk 'ladle out', assumed to be an alliterative expression (Sagart 1999: 93)

could the -m of 斟 *t-ləm be from *-wng, the nasal counterpart of *-wk?

658l-m 湛 'deep' (see above for a list of readings)

cognate to 646b 潭 EOC *ləm 'deep, abyss'

658o 糂 *s-hləm or *s-t-ləm 'rice gruel with meat'

also written with lateral phonetics:

*w-lam: 米+炎

*s-hləm: 糝 (Sagart 1999: 151 explained why this had a lateral initial)

See above for 658n 黮 'dark'.

There is a remote chance that 658b 煁 - if reconstructed with a prefixed lateral root as EOC *N-t-ləm - is cognate with 617a 炎 EOC *w-lam 'blaze, blazing'. THE PURPLE PROBLEM

All sinographic phonetic series can be divided into three categories:

1. Nonrhotic: Series whose Late Old Chinese readings have none of the characteristics pointing toward an Early Old Chinese *r: e.g., low front (retroflex?) vowels and/or retroflex initials

2. Mixed rhotic: Series whose Late Old Chinese readings sometimes have the aforementioned rhotic characteristics

3. Pure rhotic: Series whose Late Old Chinese readings always have rhotic characteristics

For a while now, I've been hypothesizing that EOC *r- could either precede or follow another consonant:

*(r)C-series: mixed or pure rhotic: *r could be anywhere in the syllable:

a prefix: *r-CV

a root initial: *rCV

a root medial: *CrV

*(C)r-series: pure rhotic; *r is part of the root *(C)rV

GSR 671 was a mixed rhotic series. Therefore I reconstructed its phonotype as *qəm and regarded its rhotic members as descending from earlier *rqəm-type syllables.

If this hypothesis were correct, GSR 671 words should have cognates in Chinese and in other languages with nonrhotic roots. This pair initially fit my expectations:

671k 黬 EOC *r-('-)qəm' 'black blot'

606k 紺 EOC *qəm(-')-s 'dark purple'

I would also add

614e 晻 EOC *'-qəm-' 'dark'

653h 暗 EOC *'-qəm(-')-s 'dark'

653i 闇 EOC *'-qəm-'(-s) 'dark, hidden'

I would have reconstructed a root *qəm 'dark' until I noticed that Schuessler (2007: 250) linked 紺 to

Proto-Tai *klam 'dark red; purple'

Proto-Austronesian *kəlam 'dark'

which in turn could be connected to

Proto-Tai *'dlam or *'dram 'black' (with a medial *-r-!)

(but the Proto-Kam-Sui form is *'nam with a nasal, so maybe PT *'dl-/*'dr- is from *'dn- < *'n-, or PKS *'n- < *'nr- < 'dr?)

Since the PT forms have *-l/r-, I was going to revise my reconstruction of 黬 'black blot' as EOC *('-)qrəm(-'), with an *-l- variant:

606k 紺 EOC *qləm(-')-s 'dark purple'

614e 晻 EOC *'-qləm-' 'dark'

653h 暗 EOC *'-qləm(-')-s 'dark'

653i 闇 EOC *'-qləm-'(-s) 'dark, hidden'

I would then change the *rq- and *rɢ- in my GSR 671 reconstructions to *qr- and *ɢr-.

However, now I wonder if I can still reconstruct 671k 黬 'black blot' with initial *r- as *r-('-)qləm(-').

The Chinese root *qləm for 'dark' may ultimately be borrowed from Proto-Austronesian *kəlam, though the vowels don't match:

EOC : PAN *a

I can't explain this as the result of Chinese ablaut because I don't know of any EOC 'dark' words like *qlam with *a.

11.16.1:06: EOC *lam (no *q-) is ironically associated with flames and brightness:

617a 炎 EOC *w-lam 'blaze, blazing', *lam 'brilliant'

for *w- cf. its Sino-Vietnamese reading viêm implying Middle Chinese *wiem

phonetic in 熋 (now written as 674a 熊) EOC *wəm 'bear'; cf. Proto-Tibeto-Burman *d-wam 'bear' (Matisoff 2003: 618)

Tibeto-Burman forms with w-: Jingpho wam31 'flash', Lushai vaamL ~ vamF 'red-hot glowing of fire, etc.' (Schuessler 2007: 180)

borrowed into Siamese: wOOm-wEEM 'brilliant, glowing (of fire) (Manomaivibool 1975 in Schuessler 2007: 180)

Proto-Tibeto-Burman *s-lyam 'tongue/flame' (Matisoff 2003: 601)

617c 燄 EOC *lam-'(-s) 'flame up'

617d 剡 EOC *lam-' 'brilliant'

617e 掞 EOC *lam-' 'brilliant'

617f 琰 EOC *lam-' 'jade tessera with pointed top' (< 'something bright'?)

617h 燅 EOC *s-lam 'to heat, to warm'

617k 惔 EOC *lam 'aflame; burning with grief'

(no EOC attestation; form projected back from MC) 閃 *hlam 'flash/lightning' (also written as 617i 覢 'time of a short glance; a moment' < ?'flash'; Schuessler 2007: 553)

It would be confusing if *lam 'bright' had derivatives with which could be confused with *ləm 'dark', and vice versa. Yet there are probable derivatives of *lam 'bright' with a schwa and a final *-p differentiating them from members of the *ləm 'dark' word family:

682a 燁 EOC *w-lap ~ *w-ləp 'shining'

690f 熠 EOC *w-ləp ~ *s-ləp ~ *ləp 'gleaming'

cf. Proto-Tibeto-Burman *s-lyap 'glitter/flash/lightning' and Proto-Tibeto-Burman *s-lyam 'tongue/flame'

The above forms may be related to words ending in (?labio)velars: e.g.,

(no EOC attestation; forms projected back from MC) 煜 EOC *w-ləp ~ *lukw < ?*ləkw 'illuminate; shine'

(ditto) 昱 EOC *lukw < ?*ləkw 'sunlight'

and via *a-ablaut:

1119f 爚 EOC *lakw 'burn; brilliant; illuminate; brightly, clearly' and its relatives

More starting here.

Next: Extreme complications. GSR 671

If pre-Old Chinese - or an early Old Chinese - word for 'needle' was *Ci-qəp-NV, how would I reconstruct other words sharing the same phonetic - 咸 'all' (numbered 671a-d in Karlgren's Grammatica Serica Recensa)?

Let me start by reconstructing their Late Old Chinese forms using Schuessler's system in my notation. These are far less controversial than what is to follow.

671a -d 咸 *gEm 'all'

671e 諴 *gEm 'harmony' (prob. rel. to 671a-d)

671f 鹹 *gEm 'salt' (cf. Written Tibetan rgyam 'sea salt')

671g-h 減 *kEm' ~ *gEm' 'abridge; moderate'

671i-j 緘 *kEm 'rope; to tie'

671k 黬 *kEm ~ *'Em' 'black blot'

671l 感 *kəm' 'to move; to touch'; also used to write 671p *gəmh

671m 顑 *xəmh ~ *khəm' 'emaciated'

671n 箴 *kim 'needle'

671o 鍼 *kim 'needle'

671p 憾 *gəmh 'dissatistifed; resent'

Almost all of them have *E which can point to an earlier OC *e or preceded by an *r in emphatic syllables. Since most of the exceptions (671l-m, 671p) had LOC *ə, I assume that was the original shared vowel in the series.

The absence of high vowels in all but one word (671n/671o 'needle') implies that the series was originally emphatic.

Most of the words begin with velars, so I will reconstruct 'emphatic' velars (phonetically uvulars?) in Early OC.

Summing up, the shared archetype (i.e., phonotype) of the series was *qəm, and all variation consists of voicing or aspiration of its initial and the presence of other consonants in preinitial, medial, and coda positions:

671a-d 咸 EOC *rɢəm > LOC *gEm 'all'

671e 諴 EOC *rɢəm > LOC *gEm 'harmony' (prob. rel. to 671a-d)

671f 鹹 EOC *rɢəm > LOC *gEm 'salt' (cf. Written Tibetan rgyam-tshwa 'rock salt' - if the *-g- is not secondary, could the first half be a loanword from Chinese coupled with the native word tshwa 'salt'?)

there could have been a *-y- in this word and others reconstructed here as *r...ə since I think *r...yə would also have become *E.

671g-h 減 'abridge; moderate'

EOC *rqəm' > LOC *kEm'

EOC*N-rqəm' > LOC *gEm'

671i-j 緘 EOC *rqəm > LOC *kEm 'rope; to tie'

671k 黬 'black blot'

EOC *r-qəm > LOC *kEm

EOC*r-'-qəm-' > LOC *'Em'

cognate to 紺 EOC *qəm(-')-s > LOC *kəmh 'dark purple'?

671l 感 EOC *qəm' > LOC *kEm' 'to move; to touch'

also used to write 671p EOC*N-qəm'-s > LOC *gEmh

671m 顑 'emaciated'

EOC *s-qhəm'-s > LOC *xəmh

EOC*qhəm' > LOC *khəm'

671n 箴 EOC *Ci-qəm > LOC *kim 'needle'

671o 鍼 EOC *Ci-qəm > LOC *kim 'needle'

from an even earlier *Ci-qəp-NV; *-p-N had already simplified to *-m by the time the graphs containing an *-m phonetic were devised

671p 憾 EOC *N-qəm'-s > LOC *gəmh 'dissatistifed; resent'

11.15.3:15: Next: The purple problem and other reasons why the above reconstructions are erroneous. AN ALL-METAL MYSTERY

When I found Zhongu khi 'needle' while writing about that language's unexpected /hk/onsonantal /kh/ontrast, it reminded me of 鍼/針 Late Old Chinese *kim 'needle'. This word is interesting for several reasons:

1. It has both -m and -p versions.

The earlier Chinese spelling 鍼 has an *-m phonetic (咸 'all'), but the newer spelling 針 has a *-p phonetic (十 'ten'), though there is no other evidence to suggest that the word had *-p in Chinese. Both spellings have the semantic element 金 'metal'.

Southeast Asian languages borrowed the -m version from Chinese:

Proto-Hmong *kɐəng (Wang Fushi in in Schuessler 2007: 610)

Proto-Mien *siim (Wang Fushi in in Schuessler 2007: 610)

Mon-Khmer languages:

Vietnamese kim (Bodman 1980 in Schuessler 2007: 610)

Khmu skam (Benedict 1992 in Schuessler 2007: 610)

Tai languages:

Siamese เข็ม khem

Saek kim (Schuessler 2007)

Proto-Tai (Li Fang-kuei 1977): *khyem

Proto-Southwestern Tai (Jonsson 1991): *Xem ~ *Xim

(reconstructions from Proto-Tai'o'Matic)

Tibeto-Burman languages have the -p version:

Written Tibetan khab < *kap

Japhug rGyalrong ta-qaV < Proto-rGyalrong *-qap

(Matisoff [2003: 659] reconstructed Proto-Tibeto-Burman *ram with *-m as well as a number of *-p forms. Are there are any TB languages with -m in 'needle'?)

2. It has various initials:

- unaspirated *k- in Late Old Chinese (palatalizing to Middle Chinese*ch-)

- aspirated kh- or a uvular (?) fricative *X- in Tai (WT aspiration is secondary)

- uvular q- in Japhug

- a liquid (cluster) *(k-)r- in Proto-Lolo-Burmese (Matisoff 2003: 557)

- a sibilant *s- in Proto-Mien

- Matisoff (2003: 517) also reconstructed Proto-Tibeto-Burman versions with a voiced initial and a glottal stop: *gaap and *'ap. He did not list any modern forms as evidence for these reconstructions.

3. It has various vowels:

- high *i in Late OC

- mid (*)e in Siamese and Proto-(SW)Tai

- *a(a) in Proto-Tibeto-Burman

Is it possible to come up with a single root at the base of all this variation? I'll try:

Proto-Sino-Tibetan *qəp

> WT khab, Japhug -kaV

> *X- > *ʁ- > Proto-Lolo-Burmese *(k-)r- (with optional prefix)

this would imply other cases of PLB *r- corresponding to velar or uvular stops elsewhere

maybe PLB borrowed this word from a language which had lenited *q- to a fricative

> *g- or *'- in other Tibeto-Burman languages?

> Pre-OC *Ci-qəp-NV

nasal-initial suffix caused root-final stop to assimilate before the suffix was lost: *-p-N- > *-m-N- > *-m

root-initial *q- fronted to *k- due to adjacent high vowel

cf. Sagart's (2007) hypothesis of prefixes causing uvulars to front to velars

high vowel of prefix metathesized and became post-root initial glide

the spelling with the emphatic phonetic 咸 *r-gəm [ʀɢəm] may predate the fronting of *q-

> OC *kyəm

> Late OC *kim (schwa dropped)

borrowed into Vietnamese and the Tai language Saek as kim

> Middle Chinese, Proto-Min *chim (Schuessler 2007: 610)

late southern Middle Chinese *chəm was borrowed into Vietnamese as châm

was Proto-Mien *siim borrowed from a Chinese dialect which had reduced *skim to *schim, *sxim, *shim, *sim, etc.?

could the long vowel *ii reflect an OC long vowel or diphthong?

> Late OC *kiem (schwa fronted to *e)

> Proto-Min *chem (Schuessler 2007: 610)

prefixed variant *N-kiem > Middle Chinese *giem (in Guangyun and jiyun)

> doubly prefixed variant *r-N-kiem or *N-r-kiem > Middle Chinese *gïem (in Jiyun)

> OC variant *sʌ-kyəm

nonhigh vowel of prefix associated with emphasis

> *skyəm > *skem > *khem [qhem] ~*xem [χem]

borrowed into Tai; emphatic prefix lowered root vowel and fused with root initial to become an aspirate or a fricative

borrowed into Khmu as skam

> Pre-OC*C-qəp-NV (*C- = nonemphatic prefix; rapid speech variant of *Ci-)

> OC *kəm

borrowed into Proto-Hmong as *kɐəng without a front vowel

> Pre-OC *Ci-qəp (no nasal suffix)

> OC *kyəp

> Late OC *kip > no modern descendants

implied by alternate spelling 針 with the phonetic 十 *gip

(Data from Schuessler 2007, Guangyun, and Jiyun added 11.14.2:27.) AN UNEXPECTED /HK/ONSONANTAL /KH/ONTRAST

I haven't run out of Tibeto-Burman / Old Chinese comparisons for "Eight Reasons". I've been saving some since Friday. But I just learned something so surprising that I thought it was worth interrupting the series.

Unlike Written Tibetan, the modern Zhongu dialect of Tibetan contrasts preaspirates (corresponding to WT s- and r-clusters) with postaspirates (corresponding to WT aspirates). I've heard of languages with one or the other before, but never both:

hki 'language' : WT skad

khi 'needle' : WT khab

hkO 'marrow' : WT rkang

khO 'leisure; have free time' : WT khom

hcha [htʃa] 'fear' : WT skrag

chha [tʃha]] 'blood' : WT khrag

htsa 'roast; burn' : WT sreg

tsha-Shu 'sieve' : WT tshags + ?

hta 'tiger' : WT stag

hta 'mark' : WT rtags (not stags)

tha 'weave' (perfective) < ?*thags : cf. WT Hthag 'weave' (present correspodning to Zhongu ntha; the WT past is btags)

hpu 'pile up' : WT spung

phu 'belly' : WT pho-ba

I wonder if some early stage of Chinese also had such a distinction with a similar origin:

original *s-consonant clusters > *preaspirates

original aspirates > *postaspirates

(I'll explore the question of whether *r- led to Chinese preaspiration in a future post.)

I suspect that the voiceless sonorants and (some) aspirated obstruents of OC originated as *s-clusters, and that OC *s-clusters originated from *s-presyllable-initial sequences: e.g.,

pre-OC *sk- > early OC *hk- > later OC *kh- (merging with the original postaspirate below)

pre-OC *kh- > OC *kh-

pre-OC *sŋ- > early OC *hŋ- > later OC *ŋ̊- (voiceless nasal) > very late OC *ŋ-

note that Zhongu has no preaspirated sonorants, though it does have voiceless sonorants: Zhongu ŋ̊- : WT sŋ-

pre-OC *sV-k(h)- > early OC *sk(h)- > later OC *ts(h)-

pre-OC *sV-ŋ- > early OC *sŋ- > later OC *s-

Could pre-OC *skh- have briefly become *hkh- before merging with *hk- and *kh-? Zhongu has no double aspirates because Old Tibetan had no sequences like *skh-.

Zhongu has some other interesting features which are potentially relevant for the reconstruction of Old Chinese and Tangut, but I'll get to them later.

(11.13.1:21: Replaced two of the original examples of velar minimal pairs [khɐ-hkO 'thirst' and khO-ntshə 'angry'] with simpler examples, and added nonvelar examples. Added nasal examples to the section on OC.) EIGHT REASONS *Y (PART 6)

This is another data-heavy, analysis-less post listing all remaining PTB *(-)y- words and their proposed OC cognates from Matisoff 2003, unless otherwise indicated:

PTB *velar-y- words:

PTB *kyang 'ginger' : 薑 OC *kang 'id.'

a loan from late OC *kïang with *-ï-?

PTB *k(y)ik/*k(y)it 'tie/bind' : borrowing from *-l-less descendant of OC *klit < ?*-ik 'tie/knot'?

繫 OC *keks 'tie' may be the true cognate

PTB *k(y)im 'pillow' : not cognate to 枕 OC*təm 'pillow'

PTB *k-yum 'house' : 宮 OC *kung < ?*-um 'dwelling house'

PTB *kyeng 'red' : 赤+巠/赬 OC ?*rtkeng 'id.' (not normal word; borrowing from TB?); not cognate to 騂 OC *sing 'red ox'

PTB *m-kyen 'know' : 見 OC *ken 'see'

is the latter really cognate to PTB *hyen 'hear/listen; look/see'?

PTB *kywal 'jackal, wild dog' : 犬 OC *khw(e/i)(n/r)' (vowel and coda uncertain) 'dog'

PTB *b-r-gyat (I would reconstruct *p-ryat) 'eight' : 八 OC *pret (< ?*pryat) or *prit 'id.'

PTB *b-r-gya (I would reconstruct *p-rya) 'hundred' : 百 OC *prak 'id.'

PTB *gyi/dzyi 'ride horse' : see part 5

PTB *g(y)ip 'ten' : 十 OC *gip 'id.'

PTB *s-ngya 'fish' : 魚 OC *nga 'id.'

I would add

PTB *kyap 'stick into/insert' : 夾 OC *rkep (< ?*rkyap) 'press between'

Is PTB *tsap 'stick into/insert' borrowed from an *s-prefixed 夾 late OC *tsep < OC *s-kep 'press between'?

PTB *gyap 'narrow' : loan from 狹 late OC *gep < OC *N-rkep (< ?*N-rkyap) 'narrow'?

PTB *gyar/*hyar 'run/ride/go by vehicle' : 騎 OC *gay < ?*gal 'ride a horse'

PTB *gyi and *dzyi 'ride horse' may be later borrowings from late OC and an Austroasiatic word cognate to Old Khmer jih; see part 5

PTB *hy-words have OC cognates with emphatic velars (uvulars?):

PTB *hyam 'salt(y)' : 鹹 OC *rgem (< ?*r-[g]yam) 'salt(y)'

is this a loan from late OC *GEm [ɣɛm] 'salt(y)'?

also cf. PTB *g-ryum 'salt' : 鹽 OC ?*rə-yam 'salt' (< *rə-hyam?)

PTB *s-hywəy 'blood' : 血 OC *xwit 'id.'

I could only find three PTB *labial-y- words with Chinese cognates in Matisoff (2003):

PTB *pyar/*byar 'affix/plait/sew' : 編 OC *pen 'weave', 辮 OC *pens 'braid'; 編 OC *ben' (< ?*N-pen') 'arrange in a series) may not be related

PTB *byer 'fly' : 翩 OC *phen (< ?*phyan) 'fly about'

PTB *s-myak/*s-mik 'eye' : 目 OC *muk or ?*mikw 'id.' (but Sagart [1999: 154] reconstructed *m-r-liwk)

I would add

PTB *pyaw 'swim/float' : 漂 OC *phew < OC ?*phyaw (found in Schuessler [2007: 414], who reconstructed OC *phiau)

is 浮 OC *bu 'float' < *N-phyu, a zero-grade cognate of *phyaw sharing a root *py-w?

PTB *pyen/*pyet 'fart' : 屁 MC *phih < OC ?*phi(t)s 'pass gas' (not attested)

could be independently developed onomatopoetic words

Until recently, I would have agreeed with Sagart (1999: 29): PTB *y-words with apparent Chinese cognates were actually borrowed from late OC forms with *y- < earlier OC *l-. However, I am no longer sure that was always the case, since I now think that words written with *l-phonetics representing only nonemphatic syllables may have been *y-initial. Hence I provide both *l- and *y- reconstructions below whenever possible:

PTB *ya 'night' : 夜 OC *laks or *yaks 'id.'

PTB *yang/*g-yak 'sheep' : 羊 OC *lang or *yang or 'id.'

PTB *yaar 'spread/extend/sail' : loan from 延 OC *lan 'extend', 筵 OC *lan 'mat', 演 OC *lan/r' 'flow out, extend' (Schuessler 2007: 554 added 衍 OC *lan'[s] 'be overflowing'), 引 OC *lin' 'pull' are all doubtful, as most of them are not written with OC *y-r phonetics

the 延 phonetic series cannot be reconstructed with *y- because it contains graphs for emphatic syllables with MC retroflex and dental stops:

梴 MC *Thien < OC *r-hlan 'long (beams)' < ?*'extended'

誕 MC *dan' < OC*lan' 'boast' < ?*'to extend the truth'

although 演 OC *lan/r' 'flow out, extend' belongs to a pure nonemphatic series implying the possibility of *yan/r, I reconstruct a lateral initial since I assume the word is cognate to '延 OC *lan 'extend'

PTB *yəw/*yu(w) 'liquor' (latter form on p. 185) : 酒 OC *s-t-lu' (suggested to me by Sagart) or *tsyu' 'wine' (Schuessler 2007: 321)

Schuessler (2007: 568) listed

PTB *(g-)yak 'armpit' : 腋掖亦 OC *lyak or *yak 'id.'

I was tempted to add

PTB *yay/*ay 'mother/grandmother/maternal aunt' : 姨 OC *ləy 'sister-in-law'

but the semantic match is poor and I cannot reconstruct *y for 姨 because its phonetic appears in

洟 'mucus from the nose'

MC *yi < OC *ləy

MC *theyh < OC*hləy-s

which is cognate to 涕 OC *hləy-s 'tears', written with a definite *l-phonetic. Moreover, it is more likely that MC dental *th is a fortition of a dental *hl than a palatal *y.

I will examine Sagart's (2007) recent proposal of a voiced uvular stop *ɢ- corresponding to my *y- in a future installement of this series.

Tangut fonts by Mojikyo.org
All other content copyright © 2002-2007 Amritavision