In "Ba-k-formation", I pointed out that Gong's reconstruction of Tangut had no rhyme -ɨw. -w did not occur after any achromatic ('colorless': i.e., nonpalatal, nonlabial) vowel:
|j-like palatal front vowels||ɰ-like achromatic central vowels||w-like labial back vowels|
|high||-iw||(no -ɨw)||(no -uw)|
The absence of -uw is understandable, since -uw might have merged with -u.
(Is there any language that distinguishes [uw] from [uu]? Or [ij] from [ii]? I assume that inscriptional Burmese uw and iy were something like [ɯw] and [ɨj], and that Gong's Tangut -jij was [jej].)
Is there any language with -iw, -ew, -ow but no -aw?
Tangut -w is partly from earlier *-k. Surely pre-Tangut must have had the rhyme *-ak, since similar rhymes can be found in other early Sino-Tibetan languages. Perhaps *-k became a velar fricative and then a velar glide that was absorbed into preceding achromatic vowels (symbolized as *A):
pre-Tangut *-Ak > *-Aɣ > *-Aɰ > Tangut -A
12.9.00:45: Cf. Jacques (2004: 212):
Proto-rGyalrong *-ak > *-aɣ > gDong-brgyad -a
but note that PGR *-k-loss also occurred after the vowels at the other points of the 'triangle": PGR *-ik and *-uk became DB -i and -ɯ.
Pre-Tangut *-Vw may have merged with *-Vɰ.Hence there is no way to tell from Tangut evidence alone whether a Tangut syllable ending in -w once ended in *-k or *-w.
Next: Could Tangut -w also come from *-p?
07.12.7.1:03: SINOPHONE ALLOPHONES
The linguistic term allophone has entered lay usage with a different meaning (emphasis mine):
For the first time ever, fully one-fifth of Canada's population has a mother tongue that's neither English nor French, Statistics Canada reported Tuesday.
The 2006 census report reaffirmed the Chinese languages as Canada's third most common mother tongue group, with more than one million people - a growth rate of 18.5% since 2001 - reporting their mother tongues as Mandarin, Cantonese, Taiwanese or four other Chinese languages.
Growth in the number of allophones - people whose mother tongue is other than French or English - was the most significant in Quebec, where their numbers expanded by 25% between 2001 and 2006, said Jean-Pierre Corbeil, chief specialist on language statistics.
I wonder if allophone was coined by analogy with francophone, etc. without any knowledge of the preexisting linguistic term.
I'm glad to see the term languages used to describe the varieties of Chinese. If Catalan can be a 'language' within the borders of Spain, why can't Cantonese be a 'language' within the borders of China?
I have been hesitant to link TT0100 ONE lew 1.43 < pre-Tangut *CV-tek to various Tangut t-words meaning 'one', etc. because those words lack a final -w < *-k.
Some of those t-words may be related to gDong-brgyad rGyalrong ɯ sti 'alone'. I had hoped that the -i of sti was from Proto-rGyalrong*-ik. Somang preserves *-k in PGR *-ik, but its word for 'alone' is tə-ɕtɕî.
Jacques (2003: 13) considered sjɨ 2.28, sjwɨ 1.30, sjiw 1.46 'new' to be related. The third form has a -w corresponding to -k elsewhere: e.g., Somang kə ɕə́k. Why don't the other two forms have -w? I briefly wondered if
- the unrelated forms for 'one' were reanalyzed as being derived from a single root with or without *-k:
pre-Tangut*CV-tek > thought of as *C-te-k
as if it were related to the *tV-words
- pre-Tangut *sjik 'new' were similarly reanalyzed as if it were root *sji + -k, resulting in -w-less words for 'new'
There are two problems with this hypothesis. First, the rhymes of the *-k-less forms do not have the same vowels as the *-k forms. I know of no evidence indicating 'k-coloring' of vowels: i.e., a vowel changing in quality before *-k. Second, Tangut has no rhyme -ɨw, < *-ɨk, possibly because *-ɨk merged with some other rhyme. Perhaps sjɨ 2.28 and sjwɨ 1.30 once did have a *-k after all:
pre-Tangut sjɨk-H > sjɨw > sjɨ 2.28
pre-Tangut sjwɨk > sjwɨw > sjwɨ 1.30
(I would like to discuss the i ~ ɨ and zero ~ -w- alternations in 'new' elsewhere.)
Sofronov (1968 I: 137) reconstructed codas for many early Tangut rhymes, but not for rhyme 31 (1.30/2.28): *-I.
I am reluctant to endorse -w-loss as the origin of what appears to be zero ~ -w alternation because I cannot find any other examples of Tangut rhyme 31 corresponding to gDong-brgyad rGyalrong -ɣ < *-k.
07.12.6.23:49: A NEW HYPOTHESIS OF LENI-SH-ON
In Jacques (2003), gDong-brgyad rGyalrong ɕ- corresponds to Tangut s- and ʑ- as well as ɕ-:
|gDong-brgyad rGyalrong||Tangut||Li Fanwen (1997) numbers of tangraphs||Notes|
|ɕ-||ɕ-||0169, 2160, 4469, 4481, 4995||́Occurs before Tangut -j(w)i, -jij ([jej]?), -jɨ, -jow|
|s-||0432, 0441, 1597, 3457, 3460, 4461||́Occurs before Tangut -jiw, -j(w)ɨ, -ow|
One might expect the pattern
|*ɕ- + any vowel||ɕ- (partly original and partly from *s- palatalizing before front vowels)||ɕ-|
|*s- + front vowel||s-|
This would predict that all GBR ɕ-words cognate to Tangut s-words must have (or have had) front vowels. But that is not always the case. 'Who' and 'clean' had nonfront vowels, and the vowel of 'new' lost its frontness in Proto-rGyalrong:
|Gloss||gDong-brgyad rGyalrong||Tangut cognates||Other evidence|
|'who'||ɕu < *-o (not *-u!)||sjwɨ 2.28, sjwɨ 1.30||Written Tibetan su|
|'new'||kɯ ɕɤɣ < *-ɯk (cf. Somang kə ɕə́k; proto-rGyalrong also had *-ik, but not in this word: PGR *-ik > GBR -i and Somang -ik/ek, not -ɤɣ and -ək; why doesn't PR have a front vowel in this word?)||sjɨ 2.28, sjwɨ 1.30, sjiw 1.46||Old Chinese 新 *sin < *siŋ < ?*sik-NV, though one could claim that OC *s- is in part from a Proto-Sino-Tibetan *ɕ- (is there any evidence for this?); Mawo Qiang khsə; Matisoff (2003: 344) cites a 'rGyalrong' (variety unspecified) kə-sik|
|'clean'||kɯ ɕo < *-aŋ||sow 1.54||Other rGyalrongic languages also have ɕ- (Jacques 2003: 229); Written Tibetan sang-ba 'cleanse' (see Schuessler 2007: 432 for more possible cognates)|
I am completely stumped. The only solution I can think of is to claim that PGR *s- became *ɕ- when preceded by a presyllable with *i -
*Ci-s- > *Ci-ɕ- > *ɕ-
- but there is no evidence for this speculation.
When I made up the title of this post, I thought I could claim ɕ- in GBR kɯ ɕɤɣ 'new' reflected palatalization before *i (as suggested by Matisoff's kə-sik), but further investigation indicated otherwise.
The rest of the title refers to the single correspondence between GBR ɕ- and Tangut ʑ-:
'juniper': GBR ɕɤɣ : Tangut ʑjiw 1.46
cf. Written Tibetan shug-pa
I hypothesize that the voicing of the Tangut form is a trace of an earlier presyllable:
pre-Tangut *CV-ɕjik > *CV-ʑjik > Tangut ʑjiw 1.46
cf. the lenition of early Korean *-s- to Middle Korean -z-
If this is correct, did other sibilants and affricates also have 'softened' forms resulting from medial voicing?
07.12.5.3:13: A SH-INGLE SH-OURCE?
Tangut had many words meaning 'one', 'alone', etc. Only one of these
TT4162 ONE ɕiə 1.28
had initial ɕ-. I'd like as few 'one'-roots as possible, so i wondered if I could somehow relate ɕiə 1.28 to the other coronal-initial 'one' words with initial t- and l- (< ?*CV-t-).
If Middle Chinese *ɕ- could partly come from Old Chinese nonemphatic *st- (Baxter 1992: 228-229, Sagart 1999: 64) -
OC *st- > *stɕ- > *ɕtɕ- > *ɕɕ- > *ɕ-
cf. OC *t- > *tɕ- and Russian щ [ɕɕ] (more conservatively [ɕtɕ]; originally a ligature of шт)
- could Tangut ɕ- also partly originate from a similar cluster? Probably not, because pre-Tangut *st- became Tangut t- + tense vowel: e.g.,
pre-Tangut ?*stjij > *ttjij > *ttjịj > TT2057 tjịj 'alone'
cf. gDong-brgyad rGyalrong sti 'alone'
pre-Tangut ?*stu > *ttu > *ttụ > TT1183 tụ 1.58 'thousand'
cf. Written Tibetan stong 'thousand'
pre-Tangut ?*stwu > *ttwu > *ttwụ > TT1152 twụ 1.58 'loyal', TT1827 twụ 1.58 'straight'
pre-Tangut ?*stwu-H (with glottal suffix) > *ttwu > *ttwụ > TT1826 twụ 2.51 'direct'
cf. gDong-brgyad rGyalrong stu 'straight'
pre-Tangut ?*sto > *tto > *ttọ > TT1055 tọ 1.70 'solidify'
cf. gDong-brgyad rGyalrong kɤ stɤm 'solidify'
Nevertheless, I am curious to find out if Tangut ɕ- could derive from a pre-Tangut non-shibilant. What gDong-brgyad rGyalrong consonants does Tangut ɕ- correspond to in Jacques (2003)?
|Tangut||gDong-brgyad rGyalrong||Number of instances||Notes|
|ɕn-||1||in 'seven': Tangut ɕjạ, GBR kɯ ɕnɯs (the rhymes are hard to reconcile, and I don't want to posit an *-a suffix to explain the low vowel in Tangut)|
|ɕr-||1||Tangut lacks the expected retroflex vowel|
|ntɕh-||2||Does GBR reflect a nasal prefix absent in Tangut? For the development of a stop between a nasal and a sibilant, cf. Eng prince [prɪnts].|
|zr- < *sr-||1||Tangut lacks the expected retroflex vowel|
This implies that Tangut ɕ is (mostly) from pre-Tangut *ɕ- which may have been followed by sonorants.
Tangut ɕ + tense vowel (e.g., 'seven') is presumably from pre-Tangut *C-ɕ-.
Tangut did not have any syllables like sjir from *srji. Did *srji become ɕji without any retroflexion?
pre-Tangut ?*srjik-H > TT2276 ɕjiw 2.40 'louse'
cf. Old Chinese 虱 *srit < *srik 'louse'
Perhaps *ɕr- also became ɕ- without any retroflexion before noncentral vowels:
pre-Tangut *ɕrjow-H or *srjow-H 'otter' > TT2865 ɕjow 2.48 'otter'
the -w is unexpected
cf. gDong-brgyad rGyalrong tɕhɯ ɕrɤm
borrowed from Written Tibetan chu sram
(chu = 'water'; cf. Eng sea otter)
All Tangut syllables of the type ɕVr have central vowels (I am skeptical about Gong's medial -i-):
ɕiar 1.81 (3 instances)
ɕjar 1.82 (3 instances)
ɕiər 1.85 (1 instance)
ɕjɨr 1.86 (1 instance)
ɕjwɨr 1.86 (1 instance)
ɕjwar 1.97 (1 instance)
I don't know how -jwar 1.97 differed from jwar 1.82.
Sofronov (1968 I: 138) reconstructed them as -jọ̃ 1.97 and -jạ 1.82.
The only external evidence I know for 1.97 is the transcription 月 (Sofronov 1968 II: 54) which tells us that a 1.97 syllable sounded like Tangut period NW Chn ?*ɥe. So was 1.97 -jwe?
Next: Tangut ɕ- almost always corresponds to gDong-brgyad rGyalrong ɕ-. But is the reverse also true?
07.12.4.1:02: HA-NDY WATERS
According to Tangraphic Sea,
TT4162 ONE ɕiə 1.28
contains not only TT0100 lew 1.43 but also WATER, the left side of
TT4227 xa 1.17
used to write a Tangut surname, the Sanskrit syllable ha, and the Tangut period northwestern Chinese equivalent of modern standard Mandarin he (i.e., ?*xɔ).
In turn, WATER in TT4227 was derived from the left side of
TT4295 xa 2.14
another surname tangraph whose analysis is unknown.
I can think of two explanations for TT4227 in TT4162:
- Phonetic: The Tangut B reading of TT4227 was XY, and the Tangut B reading of 4162 was XZ.
TT4227 consists of WATER + HAND
HAND seems to be a phonetic symbol for tji-syllables
there is a tji-root for ONE
ergo, WATER was supposed to evoke the other half of TT4227, HAND, which sounded like ONE (in Tangut B?)
The phonetic explanation cannot be tested at present. However, if the cryptosemantic explanation were correct, I would expect the right side of WATER to originate from a tangraph representing a tji-like syllable. Unfortunately, the Tangraphic Sea listed
TT2785 STRETCH ʔjiir 2.86 (no analysis available)
as the source of HAND in TT4227. HAND is presumably semantic in STRETCH which does not sound like tji or TT1545 HAND lạ 1.63. The right side of STRETCH is LONG (Nishida radical 162) which is unquestionably semantic.
Next: Could TT4162 ONE ɕiə 1.28 have a dental stop-initial root?
07.12.2.23:59: DOUBLE NEGATIVE: -1 * -1 = 1
The various tangraphs for 'one' can be grouped into two categories:
1.Those that incorporate TT0100:
TT0100 ONE lew 1.43 < pre-Tangut *CV-tek
TT4162 ONE ɕiə 1.28 (TT0100 with WATER [!?] on left)
TT4601 ONE gjɨ 2.28 (TT0100 with SPEECH on left)
2. Those that don't incorporate TT0100:
TT2369 ONE-OF-TWO dzjij 2.33
TT2422 ONE ljɨ 1.29 < pre-Tangut *CV-tjɨ (with PERSON on the right)
TT2485 ONE/A gji 2.10
TT2507 ONE/ALONE tjịj 1.61 < pre-Tangut *C-tjij (?*k-tjej)
TT2704 ONE tjɨ 1.30 (looks like HAND + NOT; see below for analysis)
TT3603 ONE/ONLY lew 1.43 < pre-Tangut *CV-tek (with PERSON on the left)
TT5479 ONE ʔa (rhyme unknown)
These groupings have nothing to do with word families. lew-and gjV-words were written with and without TT0100.
No Tangraphic Sea analyses are available for any of the graphs for rising tone words.
The title refers to the TS analysis for the level tone word
TT2704 ONE tjɨ 1.30 =
HAND = right of TT4743 NOT tji 1.11 +
NOT = left of TT1491 NOT mji 1.11
HAND may also be phonetic in (examples added 126.96.36.199):
TT1365 CEREMONY tjɨ̣j 2.55
TT2656 破 BREAK/割 CUT tjɨ̣ 2.61 (HAND could also be semantic)
TT2705 BLOCK-UP tjɨj 1.62 (HAND could also be semantic)
TT2730 KIND-OF-DISEASE tjɨ̣j 2.55
TT2762 (part of name of an animal) tjɨ 1.30 (BEAST on right)
TT2795 CAST-ASIDE thjɨ 1.30 (HAND could also be semantic)
TT4379 ARROGANCE thjɨ 1.30
TT4806 COFFIN thjɨ̣ 2.61
TT2624 ?檢 INSPECT djɨɨ 1.32
TT3037 BREAK djɨ̣ 1.69 (HAND could also be semantic)
TT4925 PREPARE/OFFER djɨ 2.28 (HAND could also be semantic)
also cf. the structure of Chn 受 'receive', 授 'give' which have a 又 'hand' on the bottom
TT4927 PROPERTY djɨ 2.28
TT4930 悴 DISTRESSED djɨ 2.28
TT5806 STRONG djɨr 2.77 (HAND could also be semantic)
TT1981 GO-SOUR tjị 1.67 (HORNED-HAT + PLACE [below])
TT2445 REALIZE tjị 2.60
TT2655 除 REMOVE tjị 1.67 (HAND could also be semantic)
TT2778 POSSIBILITY tjị 2.60
TT2779 PLACE tjị 1.67 (POSSIBILITY + the mysterious tail element ヒ)
TT5603 握 HOLD/GRASP tjị 2.60 (HAND could also be semantic)
(no examples of thjV 'handy' graphs)
TT0819 FURNITURE djii 2.12 (with WOOD on top)
TT2667 CUT-UP djịj 1.61 (HAND could also be semantic)
TT2741 DIVIDE djii 1.14 (HAND could also be semantic)
Does all this imply that the Tangut B word for 'hand' or 'arm' was something like *ti, as opposed to the Tangut A word
TT1545 HAND/ARM lạ < pre-Tangut *C-la(q)?
cf. Japhug tɯ jaʁ < Proto-rGyalrong *ljaq, Written Tibetan lag, Written Burmese lak; 翼 Old Chinese *lək 'wing'
does Qiang have a cognate? LaPolla and Huang (2003: 357) did not list the following Ronghong Qiang words as descending from Proto-Tibeto-Burman *lak:
ji-pi < ?*lj- 'shoulder', jɑpɑ < ?*lj- 'hand', lo-pu 'arm', la-ʁuəs 'elbow', la-xʂ 'palm'
also cf. Taoping lɑχɑpi, Mawo dʒɑχu 'shoulder', Taoping i 'hand' (and other similar Taoping and Mawo forms)
But what is TT1491 NOT doing in TT2704 ONE? Why not write TT2704 ONE as HAND (phonetic) + TT0100 ONE (semantic)?