No Tangraphic Sea analysis is known for

TT3777 giaa R24 2.21 'to rejoice; to enjoy'

I think it might be a combination of the left side (Li Fanwen radical 118 'person') and center (Li Fanwen radical 005; meaning unknown) of

TT3769 niee R40 1.39 'heart'

plus Li Fanwen radical 022 志 'will':

'Will' is also on the right side of

TT5398 ŋwəi R8 1.8 'delight; rejoice; pleasure; camp; fort'

which combines with TT3777 to form the word

giaa ŋwəi 'rejoice; enjoy'

The Tangraphic Sea analysis of TT5398 is


TT5398 ŋwəi R8 1.8 'delight; rejoice; pleasure; camp; fort' =

right of TT0464 khiõ R58 1.56 'to give' +

loanword from Tangut period NW Chinese 供 ?*kjõ 'to present; to offer'?

but initials don't match!

right of TT3777 giaa R24 2.21 'to rejoice; to enjoy'

The right side of 'to give' is Li Fanwen radical 088

whose meaning is unknown.

The left side of 'to give' is Li Fanwen radical 315 which is also an independent tangraph

TT0036 gəəu R5 2.5 'five' (TT1632 'eight' with a horizontal line on top!)

borrowed from Tangut period NW Chinese ?*ŋgu 'five'

whose function in TT0464 is unknown.

The Tangraphic Sea analyzed 'to give' as


TT0464 khiõ R58 1.56 'to give' =

left of TT0460 ŋwəʳ R90 1.84 'heaven'

left of TT5398 ŋwəi R8 1.8 'delight; rejoice; pleasure; camp; fort'

gəəu 'five' could be phonetic in ŋwəʳ 'heaven' but it has no phonetic resemblance to khiõ 'give'. And although there may be joy in giving, the semantic link between 'heaven's delight' and 'give' is tenuous at best. TO RETURN IS TO HAVE AND ENJOY

Sarah's husband is back from Iraq!

The Tangut character translated by Grinstead (1972: 117) and Li Fanwen (1997: 594) as 'return'*

tị (the subscript dot indicates tenseness)

consists of the left and center of

giaa 'to rejoice; to enjoy'

plus the top and bottom right of

diu 'to have'

To have him back again is to rejoice.

According to Li Fanwen (1997: 594), tị can also mean 'lucky' and 'sincere' - which he and Sarah are.

*This word has several other translations:

Guillaume Jacques (2007: 142) translated it as se ranger au côté 'to line up with' as well as retourner.

The Tangraphic Sea defined it as 'submit'.

Kychanov (2006: 693) defined it as

покоряться 'to submit'

сдаваться 'to surrender'

прибегать с покровительству 'resort to protection'

'Submit' and 'surrender' are the last words that come to mind when I think of Sarah's husband! GSR49: TEN MOUTHS

The phonetic of Old Chinese 苦 *qhaʔ 'bitter' looks like 'ten' + 'mouth':

*qaʔ 'old' = 十 *gip 'ten' + 口 *qhoʔ 'mouth'

Although one may be tempted to view 'mouth' as phonetic in 'old', *o-graphs are not normally phonetic in *a-graphs. There is no Chinese-internal evidence indicating that 口 'mouth' ever had a reading similar to Written Tibetan kha 'mouth' or Dayang Pumi khwa-ra 'mouth' (which may have a different OC cognate).

The graphs containing 古 as phonetic mostly have type A (*emphatic) readings:

GSR number sinograph type OC MC
49g-h, k-l, o, p, b', g' 姑沽蛄辜酤橭 A *qa *qo
49a-e,m-n, q, r-s, h' 古罟盬䀇 *qaʔ *qoʔ
49f, i-j, b', e' 固故酤錮 *qas *qoh
49f' (none)
49g' *s-qa *qho
49t *qha
49u *qhaʔ *qhoʔ
49a', i', j', k', l', m' 胡瑚湖葫餬鶘 *ɢa *ʁo
49v, x, y-z, h' 岵怙祜楛 *ɢaʔ *ʁoʔ
49b' *N-qaʔ
49c'-d', n', o', p', q', t', u' 居倨据琚裾椐腒 B *Cɯ-qa *kɨə
49r', s', t' 踞鋸椐 *Cɯ-qas *kɨəh
49t' *sɯ-qa *khɨə
49u' *Nɯ-qa *gɨə, *kɨə
49c'-d' ?*Cɯ- or*qjə *kɨ

Hence I reconstruct GSR49 as a mostly emphatic uvular series with a few deemphasized members (c'-d', n'-u'). If GSR49 were an originally nonemphatic velar series, I would have to reconstruct a lot of emphasizing presyllables. Moreover, I would have to reconstruct the root of 'bitter' as *kaʔ with a velar initial which would not match external cognates with uvulars.

I will explain the reasoning behind individual reconstructions later.

*Matisoff (2003: 173) regarded 戶 OC *ɢaʔ 'door; opening' as cognate to his Proto-Tibeto-Burman *m-ka 'mouth'. Perhaps the *ɢ- of 戶 is from *m-q-.

He linked 口 OC *qhoʔ 'mouth' to his PTB *ku(w) 'mouth'.


I have written a lot on this blog about where uvulars should be reconstructed in Old Chinese. My proposal only partly overlaps with those of EG Pulleyblank (1982) and Laurent Sagart (2007). Which of the three proposals, if any, best matches the non-Chinese ('Tibeto-Burman') evidence in Sino-Tibetan? If Old Chinese had uvulars, were those uvulars retentions or innovations? And if they were retentions, was Old Chinese the only Sino-Tibetan language preserving uvulars, or did other ST languages have reflexes of proto-uvulars distinct from reflexes of proto-velars? I can begin to answer these questions by looking at Sino-Tibetan words for 'bitter'.

Old Chinese 苦 *qhaʔ 'bitter' is written with the phonetic 古 *qaʔ 'old'. The phonetic series of 古 (GSR49) contains no graphs with Middle Chinese *ʔ- or *x- which Sagart believed were characteristic of Old Chinese uvular-initial series. Thus Sagart would probably reconstruct GSR49 as velar-initial.

However, the Middle Chinese reading of 古 had the same back allophone of */k/ as readings for graphs which were deemed inappropriate for the transcription of Indic k-syllables. I suspect this back allophone was uvular *[q], a sound absent from Indic. Hence I reconstruct 古 as MC *qoʔ and 苦 as MC *qhoʔ.

In "Loanwords as Evidence for Old Chinese Uvular Initials", Pulleyblank (1982: 209-210) wrote,

Though the loanword evidence [see this post] is thus inconclusive in itself as far as I tell at present, one can argue on general grounds that the Middle Chinese allophonic contrast is more likely to have arisen from the fronting of uvulars before high vowels than from the backing of velars before nonhigh vowels. Fronting processes are much more widely attested than backing processes in diachronic phonology. The fact that Middle Chinese back allophones of /k/ and /kh/ have not been preserved in any modern dialect, so far as I am aware, is in conformity with this general trend.

Is it true that uvular fronting is more common than velar backing? I don't know, but I'm sure that uvulars are more marked than velars, and sound changes tend to go in the direction of decreasing markedness. k occurs in 89.8% of languages in the UPSID database, whereas q occurs in only 11.75% of UPSID languages. (I'm surprised the figure for k isn't nearly 100%. Which languages lack k?) If k > q were a frequent shift, the percentage for q might be higher.

In modern Chinese languages, 'bitter' generally has a velar initial kh-. (Cantonese fu has a labiodental initial from earlier *x- < *kh-.)

However, elsewhere in Sino-Tibetan, 'bitter' may appear with uvular as well as velar initials:

With uvulars

(data from Sun Hongkai 1981, Jackson Sun 2003, Huang and LaPolla 1996)

Zhongu and Yanyun Tibetan qhɐ-nde

Shibazi Tibetan qhaa

Ren'entang Tibetan qhʌ

Taoping Qiang qhɑ

Mawo Qiang qhɑ

Ronghong Qiang qhɑ(q)

With velars

Written Tibetan kha

Written Burmese khaaḥ

Tangut khɪ R9 1.9

I can account for this data with four different scenarios:

1. Since 'bitter' has or had an initial uvular in three different branches of Sino-Tibetan (Chinese, Tibetan, Qiang), it may have had an initial uvular in Proto-Sino-Tibetan. However, are there are any other etyma with uvulars in all three branches?

2. The initial of 'bitter' independently backed in Chinese, Qiang, and some Tibetan dialects. This scenario would be improbable if backing of velars is rare.

3. The initial of 'bitter' backed in a single language (Old Chinese?), and other languages borrowed this backed version. This would imply that the ST languages with uvular-initial words for 'bitter' might have had more early contact with Old Chinese. However, uvular initials also appear consistently in 'snow', a nonstandard Tibetan word without any OC source:

Zhongu qha

Shibazi qhɑɑ

Yanyun qhaa

(but Ren'entang has khææ; cf. Written Tibetan kha-ba)

4. 'Bitter' originally had a velar initial but was sporadically backed for sound-symbolic purposes (e.g., to resemble a noise made in disgust?). But I cannot think of any onomatopoetic motivation for backing of the initials of other uvular-initial etyma (e.g., 'snow' above). SHE HATES THE GORKY TASTE OF OLD GRASS

Sarah has been waiting and waiting for her husband to come home. She's been having a bitter time.

One Tangut character for 'bitter' (Russian горький)


is composed of its homophone

khɪ 'to hate'

plus the left of

vị 'taste' (probably a loan from Tangut period northwestern Chinese 味 *vi, but the tenseness symbolized by the subscript dot i

khɪ 'to hate' is phonetic in the graph for khɪ 'bitter'. 'To hate' may even be derived from 'bitter'. Similar words for 'bitter' (but not 'to hate') are found in other Sino-Tibetan languages -

Old Chinese 苦 *qhaʔ

Written Tibetan kha

Written Burmese khaaḥ

(Tangut has raised *a to ɪ, whereas these languages preserved the original *a.)

- suggesting that 'bitter' is the older meaning, and that 'to hate' is a Tangut innovation.

The Chinese character for 'bitter' also consists of a phonetic element plus a semantic element:

phonetic semantic
Tangut khɪ 'bitter' khɪ 'to hate' vị 'taste'
Old Chinese 苦 *qhaʔ *qaʔ 'old' 艹 'grass'

I think Sarah's had enough of old grass. I hope she has some sweeter days ahead.

I have no idea how to analysis the Tangut character for 'sweet':


It has 'motion' on the left and 'tilt' on the right. Neither component has any obvious semantic relationship to 'sweet'. Perhaps they are phonetic.

Combining 'sweet' with

zəʳ 'dew' (written as jaʳ 'eight' + 'water' - why 'eight'?)

results in

zəʳ liẹ̃ 'sweet dew' (Tangut has noun-adjective order)

a translation of Chinese 甘露 'sweet dew', used as an equivalent of Sanskrit amṛtam, meaning/cognate to 'ambrosia'. amṛtam is neuter; its masculine equivalent is my name amṛtas 'immortal' - which I've anglicized as Amritas.

अमृत उवाच ।

amṛta uvaaca

'Amritas said

सेरे तव पतिः प्रत्यागमिष्यति ।

sere, tava patiḥ pratyaagamiṣyati.

'O Sarah, your husband shall return.' A GRACIOUS LINK

Today I discovered that Leofwende 'gracious' linked to me. (I discovered her blog through Sarah's comments section.) This might be the first time I've been blogrolled in years. Writing technical posts tends to scare people away. I wonder how many people are going to make it past this paragraph?

Leofwende's name contains the diphthong eo. I can't think of any East or Southeast Asian language with any of the diphthongs of Old English (written here in phonetic notation rather than OE spelling):

short long
high iy iiy
mid eo eeo
low æe ææe

(12.17.1:33: I regard Vietnamese eo as a vowel-glide sequence [ɛw], not a diphthong.)

I'd like to see a survey of diphthong frequency across languages. Knowing which diphthongs are more common will help me reconstruct the most likely diphthongs in the two languages that I work on the most: Old Chinese and Tangut. I believe that both Late Old Chinese and Tangut were full of 'bent' vowels: vowels that had 'warped' into diphthongs: e.g.,

earlier e > ie (first part of e 'bent up') or ae (first part of e 'bent down')

earlier o > uo (first part of o 'bent up') or ao (first part of o 'bent down')

'Up' and 'down' refer to vowel heights:

front central back
high i u
mid e o
low a

Schuessler (2001) proposed triphthongs and even a tetraphthong *-eiau in Late Old Chinese. (Does any known language have such a tetraphthong?)

It is debateable whether complex vowel sequences should be reinterpreted as sequences of vowels and glides: e.g., was Late Old Chinese 標 'branch' (now 'mark; sign; label')

pieu (consonant + triphthong)

piew (consonant + diphthong + glide)

pjew (consonant + glide + vowel + glide) ZERO GRADE OR VOWEL LENGTH?

Recently, I proposed that Old Chinese

*-aŋ ~ *-a

alternations come from an earlier vowel alternation:

*-aŋ ~ *-ŋ (zero grade; no vowel)

However, I have not found any evidence for similar alternations involving other sonorant codas.

More recently, I have wondered if the loss of velar nasals was conditioned by vowel length:

*-aaŋ > *-aŋ


*-aŋ > *-ã > *-a

These changes could be regarded as different parts of a single chain shift:

*-aaŋ > *-aŋ > *-ã > *-a

Nonvelar nasals did not 'fuse' with preceding vowels:

*-a(a)n > *-an

*-a(a)m > *-am

Nasal loss occurred mostly after *a and less frequently after *ə. Was there a constraint against neutral short vowels followed by velar nasals?

This hypothesis fails to account for the fact that

*-əŋʔ ~ *-əʔ

is far more common than

*-əŋ ~ *-ə

The only example of the latter that comes to mind is

能 OC *-nəŋ 'capable' (OC *-nə 'bear' also writtten with same graph)

態 OC *-nə-s 'bearing, manner'

(purely coincidental that 'bearing' contains 'bear' in English!)

which is a xiesheng series but not a word family.

Moreover, if a small number of nasal ~ zero alternations are attributed to vowel length, can vowel length be reconstructed in morphemes that did not participate in such alternations? I have long suspected that OC *-ʔ was an automatic coda for syllables ending in short vowels:

[CVʔ] = /CV/


如 OC *Cɯ-na [naa] /naa/ 'like'

汝 OC *Cɯ-naʔ [naʔ] /na/ 'thou'

OC *-ʔ after sonorants may reflect (earlier?) short vowels or a suffix *-ʔ:

無 OC *Cɯ-ma < ?*maŋ 'not have'

亡 OC *Cɯ-maŋ < ?*maaŋ 'disappear'

罔 OC *Cɯ-maŋʔ < ?*maaŋ-ʔ 'not have'

I have also wondered if OC *-ʔ could be a trace of an earlier stop: e.g., *-ŋʔ < *-ŋk. This hypothesis could be combined with Sagart's hypothesis of final voiced stops shifting to *-Nʔ:

*-ŋk > *-ŋg > *-g > *-ŋʔ (and on to *-ʔ after a short vowel?)

cf. my normal pronunciation of can't as [kæ̃ʔ] vs. my formal pronunciation [kænt]

I suspect that velar nasal loss was an irregular, unpredictable change. *-aŋ ~ *-a sets do not constitute coherent paradigms. Here's a different interpretation of OC syllables with nasal loss and/or *-ʔ:

pre-Old Chinese Old Chinese
*-aaŋ *-aŋ (regular)
*-a (irregular)
*-aŋ [aŋʔ] *-aŋʔ (regular)
*-aʔ (irregular)
*-əəŋ *-əŋ (regular)
*-ə (irregular)
*-əŋ [əŋʔ] *-əŋʔ (irregular!)
*-əʔ (regular!)
*-VV [VV] *-V
*-V [Vʔ] *-Vʔ
*-VVN [VVN] other than *-aaŋ and *-əəŋ *-VN
*-VN [VNʔ] other than *-aŋ and *-əŋ *-VNʔ

*-N represents any sonorant.

The above table excludes *-ʔ as a suffix or as a root-final consonant (see Schuessler 2007: 30-.35)

Maybe zero ~ *-ʔ alternations originated as vowel length alternations:

無 OC *Cɯ-ma < ?*maaŋ 'not have' (irregular nasal loss)

亡 OC *Cɯ-maŋ < ?*maaŋ 'disappear'

罔 OC *Cɯ-maŋʔ < ?*maŋ 'not have' (or ?*ma(a)ŋ-C)

Tangut fonts by Mojikyo.org
Tangut radical font by Andrew West
All other content copyright © 2002-2008 Amritavision