Nothing, at least as first glance. I can't find any Tangut descendants of Proto-Tibeto-Burman PTB hwang 'come, enter' (< ?Proto-Sino-Tibetan √w-ng 'go'? see "Hgro-w-ng phyin-s"). Here are all the Tangut verbs for 'go' listed in Grinstead (1972: 221):

TT0045 GO (Gong wyiy 1.35; anal. as WALK + HELP)

TT1052 GO (Gong shyï 1.29; anal. as VERY-FEW [phonetic] + EMPLOY)

TT1957 GO (Gong shywï 1.29; anal. as REGION + CROSS)

TT2942 GO, ACTION (Gong jyiy 1.35; anal. as STEP + WALK)

TT2977 GO (Gong jyịy 1.61; anal. as STILL/YET + EMPLOY)

TT3071 GO, transcription char for Skt pa* (Gong pyiy 1.36; no anal. available)

All of those tangraphs are more complex than Chinese characters for 'go' such as 行 and 去 (OC 'leave' but Md 'go') or even 往. Notice there is no single grapheme common to all six (or even to a majority of the six). This is unusual if one assumes that tangraphy is a semantocentric writing system in which similar meanings are indicated by shared graphemes.

TT1052 GO (Gong shyï 1.29)

has the 'motion' grapheme on top found in graphs for other motion verbs: e.g.,

TT1031 GO-FAST (Gong zhyo 1.51)

TT1036 ESCORT (Gong byu 1.3)

TT1043 FOLLOW (Gong shywịw 2.61)

TT1046 RUN-AWAY (Gong ta 1.17)

But why don't the other GO-graphs have this grapheme? (And what is that grapheme doing in, say, TT1025 ENEMY or TT1042 EXPLAIN which have nothing [obvious] to do with motion. Its presence in TT1026 DIRECTION, TT1028 LOOK-AROUND [look while moving around?], TT1041 ?WAX/?WANE [written as MOTION + MOON; the moon 'going' in/out of phases?]; is more understandable.)

There seem to be four different roots for 'go' represented among the Tangut verbs:

√wyiy: TT0045 wyiy 1.35

√shyï: TT1052 shyï 1.29, TT1957 shywï 1.29 (is the -w- the trace of an earlier affix?)

√jyiy: TT2942 jyiy 1.35, TT2977 jyịy 1.61 (is the tense [underdotted] vowel the trace of an earlier affix? There are other cases of lax and tense vowel alternations in Tangut.)

√pyiy: TT3071 pyiy 1.36

Words with shared roots do not have similar graphs. (This may tell us that the graphs also represented different-sounding Tangut B words. Even if Tangut B did not exist, dissimilar graphs for cognates are normal in Chinese: e.g., OC 于 wa 'go' and OC 往 wang' 'go' with different phonetic elements [于 itself and OC 王 wang 'king' [beneath a dot that was once a 止 foot; 止 + 王 = 主 is not to be confused with OC 主 to' 'master'].)

Even though Tangut is thought to be a Qiangic language, none resemble Taoping or Mawo Qiang kə 'go' (Sun 1981: 212).

The first root wyiy may come from a pre-Tangut wa, since -a is one source of Tangut -yi(y) (see Gong's data in Matisoff 2003: 172). Hence TT0045 wyiy 1.35 is probably the Tangut descendant of Proto-Tibeto-Burman s-wa 'go'. If the √wa of s-wa ultimately came from Proto-Sino-Tibetan √w-ng 'go', then there would be something w-ng with Tangut.

The shift of a to i could have occurred in gradual steps: e.g., a > ïa > iə > i.

The last root pyiy happens to look like pay 'go' (in the Kamarupan subgroup of Tibeto-Burman) or pay 'go' in Proto-Tai, completely outside TB (Matisoff 2003: 209).

Next: A literary movement.

*The rhyme of pyiy 1.36 doesn't match that of Skt pa at all, though in some cases Tangut yi < a see Gong's examples in Matisoff 2003: 172) but that presumably occurred long before the Tangut adopted Buddhism, so perhaps this graph had a second reading ending in -a. (But why wouldn't that reading be recorded in Tangraphic Sea?)

06.11.11.14:40: HGRO-W-NG PHYIN-S (what are they?*)

In the first part of "Failing to Make the Grade", I said that there was no Tibeto-Burman support for the zero grade hypothesis. For example, if there were a pre-Old Chinese (OC) root √w-ng 'go':

POC w-Ø-ng >OC 于 wa 'go' (zero grade)

POC w-a-ng-' > OC 往 wang' 'go' (a-grade)

we might expect Tibeto-Burman (TB) cognates with -ng. However, Matisoff (2003: 618) reconstructed the Proto-Tibeto-Burman (PTB) word for 'go' as s-wa without -ng.

This does not necessarily mean that Proto-Sino-Tibetan (PST) did not have -ng in its word for 'go'. It's possible that pre-OC kept PST -ng whereas PTB dropped it (but why? - PTB did have other words ending in -ng; did PTB also change syllabic -ng to -a?). Nonetheless, I am not enthusiastic about projecting an already shaky Chinese feature back into PST.

There are TB 'go' words with -ng. In Starostin's online database, I found:

Written Tibetan: Hong 'come' (< ?ã-wang-); song (< s-wang?; past and imperative of Hgro 'go')

Written Burmese: swah 'go', wang 'enter, go or come in'

Proto-Kuki-Chin: wang 'go'; cf. the Kuki-Chin language Lushai (in India). vaak, va' 'go, walk' (-k is the stop counterpart of -ng)

Proto-Kiranti (in Nepal): wòng

Proto-Gurung (in Nepal) wàŋ ˜ khòŋ 'enter'

Dhimal (in Nepal) wang 'enter'

Bunan (in India) hwangs ˜ hoang 'come out'

Starostin reconstructed PST (s-)'^wă(-ng). Matisoff (2003: 269) reconstructed PTB hwang 'come, enter' which is distinct from his aforementioned PTB s-wa 'go'. It is not possible to derive PTB hwang from PTB wa in his system since he did not reconstruct a PTB prefix h- or a PTB suffix -ng. He may consider them to be similar-sounding but unrelated roots. The zero grade hypothesis allows me to tie Matisoff's reconstructions together -

PST √w-ng:

PST w-Ø-ng > OC 于 wa 'go', PTB (s-)wa 'go' (zero grade)

PST w-a-ng-' > OC 往 wang' 'go', PTB hwang 'come' (a-grade)

- but it alone cannot account for PTB h- in PTB hwang.

(I don't reconstruct a h- in PST √w-ng because there is no Chinese evidence for an h- in 往 for 'go'. h- may have been added in PTB after it broke away from Sinitic. One [unlikely] possibility is that PTB h- is the remnant of a PST prefix [changing 'go' to 'come'? - cf. Skt gam 'go' ~ aa-gam 'come'] that was associated with the a-grade. This prefix was lost in OC but was retained in PTB. But I know of no evidence for such a prefix.)

Calling every inconvenient sound an 'affix' is a copout that historical linguists, myself included, often resort to. Although we cannot be expected to identify the functions of all affixes, ideally any proposed affix should occur repeatedly with a semantic common denominator.

The problem is that affixes can become semantically opaque: e.g., what does Eng be- mean? No English speakers today think of become as be-come; they regard become and come as independent entities. Did PTB speakers regard (s-)wa and hwang as variants of the same root (do, undo) or as different roots (come, become)? And even if they didn't believe that (s-)wa and hwang were related, that doesn't mean the two roots weren't related at a deeper level: e.g., they were both derived from a PST √w-ng, just as come and become were both derived from Proto-Germanic kuman 'come'.

Next: Can you figure out how Written Tibetan Hgro 'go' could have been derived from √w-ng?

*The Written Tibetan verb 'go' has a suppletive paradigm:

Hgro (present and future)

phyin-pa or song-ba (both past)

Verbs for 'go' have suppletive paradigms in other languages: e.g., English go, went (not goed).

Goldstein's modern Tibetan dictionary (2001: 246) lists rgyug as the imperative of Hgro 'go', but rgyug 'run' is a separate verb:

rgyug (present)

brgyugs-pa (past)

brgyug-pa (future)

rgyugs (imperative)

His entry for rgyug 'run' (2001: 269) does not state that it is the imperative of Hgro 'go'.

Instead of posting Sagart's (1999: 135-6) treatment of what I've been regarding as 'zero grade' pre-Old Chinese (OC), I'd like you to try to come up with another analysis of your own for the OC open vowel ~ vowel-nasal alternations (concrete examples are in these three posts):

(V = an unspecified vowel)

-V ~ -Vm

-V ~ -Vn

-V ~ -Vng

Try to think of a second solution if you can. Then see how you can interpret this quartet (mostly from Pulleyblank 2000: 41, but the reconstructions are mine) in terms of zero- and non-zero grade hypotheses:

OC 于 wa 'go' (cf. Proto-Tibeto-Burman s-wa 'go' [Matisoff 2003: 618])

OC 往 wang' 'go'

OC 云 wən 'say'

OC 曰 wat 'say'

(Pulleyblank links the last two to OC 于 wa 'go', though I fail to see the semantic connection.**)

Next: So wat are the alternatives to the zero grade?

*06.11.11.11:27: A reference to this Clash song.

**06.11.11.00:36: Then again, in English, go can be a quotative verb: e.g., He goes (= 'says'), what's the deal with that data? or How does that line go? = 'what words are in that line?' I don't think Pulleyblank had that semantic parallel in mind, though.

06.11.8.23:58: FAILING TO MAKE THE GRADE: LABIAL NASALS

I don't have enough time to deflate my own hypothesis in one post, so I'll cut this up into three parts. The second part will be posted on 11.10 since I'll be working even later than usual tomorrow night.

Two general arguments against pre-OC (Old Chinese) syllabic nasals > OC a:

1. There is no evidence for separate rhyming categories -m, -n, -ng. I dodged this by proposing that syllabic nasals were pre-OC.

2. There is no (known) Tibeto-Burman support for this hypothesis. However, one could argue that the early Chinese script preserved a feature (zero grade syllabic nasals) lost in the rest of Sino-Tibetan*.

Why pre-OC syllabic m > OC a doesn't work:

One would expect many OC alternations of the type -a ~ -am. But offhand, I think -a ~ -an and -a ~ -ang alternations are more common. This suggests that -m roots rarely took the zero grade, or that this hypothesis is wrong: i.e., pre-OC syllabic m often became some other OC vowel, or there was no syllabic m in pre-OC.

One might even expect OC alternations like CaC (< pre-OC CmC) ~ CVmC ~ CmVC, but none are known. (For instance, given 古 POC km' > OC ka' 'old' and 久 POC k-m-ə-' > OC k^wə' 'be a long time', one would expect an OC form like kVm' meaning something like 'old', but no such words exist.)

One could try to salvage the hypothesis by saying that pre-OC syllabic m became different OC vowels in different environments. Data from Pulleyblank (1962: 234-235) could be quoted for (very weak) support:

pre-OC syllabic m > OC a after k and dentals (other than l)

古 POC km' > OC ka' 'old', thought to be phonetic in a-grade OC 敢 kam-' 'dare' (from a homophonous but unrelated root) and sharing a root with ə-grade

久 pre-OC k-m-ə-' > OC k^wə' 'be a long time'

那 POC nml > OC nay (a place name), whose phonetic may be 冉 OC nam' 'advance' (whose earlier form looks like the left side of 那)

pre-OC syllabic m > OC ə after s and ng (but not k!?)

纔 ?POC √s-m 'only just' (not attested in OC)

zero grade: ?OC n-t-sə > MC dzəy

schwa grade: ?OC r-səm > MC ShEm

雨+ 乑 ?POC √ng-m 'rain' (not a common word; not attested in OC)

zero grade: ?OC r-ngə > MC ngE

schwa grade: ?OC ngəm > MC ngïm

phonetic is 乑 ?OC ngəm 'crowd standing' (not attested in OC)

pre-OC syllabic m > OC u after ss, l, l, k^w (k?)

慘 ?POC √ss-m 'grieved':

zero grade: OC ssu > MC tshaw

schwa grade: OC ssəm > MC tshəm

冘 ?POC √l-m 'walk' (not a common word; not attested in OC)

zero grade: ?OC lu > MC yu

also cf.

'go along': 猶 OC lu > MC yu (not primary meaning of char; linked to 冘 by Gong 1996: 60)

'road': 道 OC lu' > MC daw'

schwa grade: ?OC ləm > MC yïm

九 ?POC √k^w-m-' 'nine':

zero grade: OC k^wə' > MC ku'

(or is this √km': k-m-ə-'?; same root as 'old/long' above? did POC syllabic m > OC mə after k?)

phonetic in 'dye': 染 POC ngkyam' > OC ngyam > MC ñiem

The changes proposed above appear to be arbitrary. For instance, there is no reason why POC syllabic m would become a schwa after velar ng but not equally velar k.

One might expect the vocalic descendants of POC syllabic m to vary in height depending on the 'emphasis' of the preceding consonant as 'emphatic' consonants have a lowering effect on vowels, but there is no correlation between 'emphasis' and the OC vowels that were supposedly once POC syllabic m.

Pulleyblank mentions one xiesheng (phonetic compound) set that, if genuine**, undermines the syllabic nasal hypothesis even further:

'go': 于 POC w-ng (see here) or w-m? (I'd rather not propose w-ng^w!)

may be phonetic in ?'small fire': 于+火 (also 干+火) ?OC wəlam (unattested; < POC w-l-ng-m?) > MC dam

probably sharing a root with 炎 OC wlam 'fire'

alternate reading: ?OC rlam > MC Dïem

in turn phonetic in

'?' 干(< 于)+火+頁 (typo for 覝 'examine'?) ?OC wəlam (unattested; can't find gloss; < POC w-l-ng-m?) or ?OC ram (cognate to 監 kram 'inspect'?) > MC liem

If 于 ended in -ng in POC, why would it be phonetic in graphs for -m words?

Next: Dental nasals.

*06.11.9.7:40: Sanskrit is the only Indo-European language with syllabic R and L (and the latter appears only in one root, kLp 'be well ordered'). All other IE languages have lost R and L, and none have preserved syllabic M and N.

**06.11.9.7:41: 于+火 / 干+火 may be a semantic compound of 于 OC wa 'go' or 干 OC kan 'dry' with 火 OC hməy' 'fire'.

06.11.7.23:51: THE SYLLABIC NASAL SOLUTION

... to yesterday's Old Chinese (OC) problem set is probably wrong, but here it is anyway:

Pre-OC nasals in zero-grade roots became syllabic nasals that later became OC a, just as the Proto-Indo-European syllabic nasals m and n in the zero grade became Sanskrit a:

pre-OC syllabic m > OC a

古 pre-OC k-m-' > OC ka' 'old'

thought to be phonetic in a-grade OC 敢 kam-' 'dare' (from a homophonous but unrelated root)

'old' shares a root with ə-grade (but in a different interconsonantal slot!)

久 pre-OC k-m-ə-' > OC k^wə' 'be a long time'

(nonsyllabic -m-; cf. Sagart's [1999: 98] proposal of OC km > Middle Chinese kw in 袂 OC kmet ~ [kə]mets > MC kwet ~ myeyh 'sleeve')

pre-OC syllabic n > OC a

語 pre-OC ng-n > OC nga' 'talk'; cf. a-grade 言 OC ngan 'words'

拒 pre-OC g-n-' > OC ga' 'oppose; ward off'; cf. a-grade 捍 OC gans 'ward off'

圄 pre-OC ng-n-' > OC nga' '(im)prison'; cf. a-grade 犴 OC ngans 'prison'

(食+干) pre-OC (ng-)k-n(-s) > 餬 OC ngka 'eat a meal at another person's expense; be given a meal', 糊 ngka 'thick gruel', (食+固) kas 'thick gruel of rice'; cf. a-grade (食+干) kan 'thick gruel of rice'

pre-OC syllabic ng > OC a

胥 pre-OC s-ng > OC sa 'mutually'; cf. a-grade 相 OC sang 'mutually'

無 pre-OC m-ng > OC ma 'not'; cf. a-grade 亡 OC mang 'not'

于 pre-OC w-ng > OC wa 'go'; cf. a-grade 往 OC wang' 'go'

舁 pre-OC (k)-l-ng(-') > OC la 'lift', 舉 OC kla' 'lift'; cf. a-grade 揚 OC lang 'lift, raise'

女 pre-OC r-n-ng-' > OC rna' 'woman'; cf. a-grade 娘 OC rnang? 'young woman' (not attested in early texts)

Next: Why this solution is wrong. And if it were right, why would my zero-grade short vowel hypothesis be wrong?

06.11.6.23:59: FROM NASALS TO VOWELS?

Proto-Indo-European zero-grade syllables ending in sonants ended in short vowels in Sanskrit: e.g.,

PIE g^w-m- > Skt ga- 'go' (in ga-ta 'went'; guNa grade: gam)

PIE n- > Skt a- 'un-' (in a-mR-ta 'immortal' = 'un-died')

PIE m-r- > Skt mR- 'die' (R is a syllabic r; in a-mR-ta 'immortal'; guNa grade: mar)

If pre-Old Chinese (OC) once had a zero grade, can you guess how these OC sets from Pulleyblank (1962: 234-235; 1991: 73), Gong (1996: 58) and Sagart (1999: 135) might be explained?

古 OC ka' 'old', possibly phonetic in OC 敢 kam' 'dare' according to Shuowen (written by a late OC speaker; probably not accurate since neither early nor later forms of 敢 'dare' contain 古 'old') and possibly cognate to 久 OC k^wə' 'be a long time'

語 OC nga' 'talk', 言 OC ngan 'words'

拒 OC ga' 'oppose; ward off', 捍 OC gans 'ward off'

圄 OC nga' '(im)prison', 犴 OC ngans 'prison'

both occur in the late OC disyllabic word 圄犴 ngïa'nganh 'prison'

餬 OC ngka 'eat a meal at another person's expense; be given a meal' [I'm not sure this really belongs, and it can be omitted if desired]; 糊 ngka 'thick gruel', (食+固) kas 'thick gruel of rice', (食+干) kan 'thick gruel of rice'

胥 OC sa 'mutually', 相 OC sang 'mutually'

無 OC ma 'not', 亡 OC mang 'not'

于 OC wa 'go', 往 OC wang' 'go'

舁 OC la 'lift', 舉 OC kla' 'lift', 揚 OC lang 'lift, raise'

女 OC rna' 'woman', 娘 OC rnang? 'young woman' (not attested in early texts)

06.11.7.7:57: There are, of course, explanations that do not involve a zero grade.

06.11.5.23:59: ZERO GRADE OR LONG VOWELS?

David Boxenhorn suggested that my 'zero grade' Old Chinese (OC) forms were really i- and u-grade roots:

I cannot rule out his interpretation for two reasons:

1. As an adherent of a six-vowel system for OC like Starostin, Baxter, and others, I think there could have been six or seven possible grades: one for each vowel plus a zero grade:

Not all of these forms necessarily existed. I have no constraint against similar vowel-glide combinations: i.e., DB's -u-w and -i-y. I am, however, uncertain as to whether -uw and -iy were distinct from -u and -i, which brings us to my second reason:

2. For many years I have assumed that there was a partial correlation between OC vowel length and phonation / tones in later stages of Chinese. In Burmese, open syllables written with Indic long vowel symbols have 'low tone' (the default category) and open syllables written with Indic short vowel symbols have 'creaky tone'. I believe there were similar correlations in OC:

(I write creaky phonation and 'rising' tone with final -' for typing convenience, even though a final glottal stop was disappearing or already gone in those later stages of Chinese.)

I suspect that glottal stop was the default coda for all OC (and earlier Burmese) syllables ending in short vowels.

If the words that DB interpreted as kuw and hliy were really zero grade forms, they should have ended in short vowels and acquired creaky phonation / 'rising' tones. But they didn't; they were 'level' tone words in Middle Chinese, implying that they had long vowels in OC - which might have originated from the sequences -uw and -iy:

-uw = -uu

-iy = -ii

Unaffixed zero-grade forms of √k-w and √hl-y do not seem to exist; if they did, they might have been ku' and hli', with nonetymological (i.e., non-root) final glottal stops.

There is, however, another possibility: the vowels of zero-grade forms ending in -w and -y could have been automatically lengthened, resulting in MC 'level' tones:

k-w > ku [quu] > MC kaw ('level' tone)

hl-y > hli [hlii] > MC shi ('level' tone)

So how can we choose between DB's analysis, a zero-grade analysis, or a Pulleyblank-like two-vowel analysis with ə (as opposed to a, the only other vowel he permits in OC)?

kə́w > MC kaw ('level' tone)

hlə̀y > MC shi ('level' tone)

(Pulleyblank's acute accent corresponds to 'emphasis' in my reconstruction and his grave accent corresponds to non-emphasis in my reconstruction. They do not correspond to later Chinese tones, though those are often written with accents.)

Each analysis has different implications:

DB analysis: If ku is really derived from kuw, it should have something in common with other u-grade forms (and/or other high-vowel-grade forms like hli < hliy? and/or other back-vowel-grade forms with -o-?) and not with other zero-grade forms (if they exist).

AMR zero-grade analysis: If ku is really derived from k-w, it should have something in common with other zero-grade forms like hli < hl-y (assuming that is not really the i-grade form hliy).

Pulleyblank-like two-vowel analysis: kə́w should have something in common with other ə-grade forms like hlə̀y.

The latter two analyses don't look very good, since I don't see what my ku / Pulleyblank's kə́w (皐 'high', a stative, intransitive verb) has in common with my hli / Pulleyblank's hlə̀y (尸 'set forth, display', a transitive verb).

Without systematically and comprehensively going through the entire OC lexicon, it is impossible to make statements which pin down the functions of the grades like these:

In Sanskrit,

- the base grade (-a-, -i-, -u-, -R-, -L-) of a root is found in past participles: e.g.,

i-ta 'went'

hu-ta 'sacrificed'

kR-ta 'done'

- the guNa grade (-a-, -e- [< -ai-], -o- [< -au-], -ar-, -al-) of a root is found in the 'strong' forms of verbs: e.g., singular presents of the 'second conjugation' (verb classes II, III, V, VII, VIII, IX):

e-mi 'go'

ju-ho-Shi 'you (sg.) sacrifice' (ju- is a distorted reduplication of the root hu)

kar-o-ti 'he/she/it does'

- the vRddhi grade (-aa-, -aay- [< -aai-], -aav- [< -aau-], -aar-, [-aal- is not attested]) is found in third person singular perfect forms of verbs with vowel-final stems: e.g.,

iy-aay-a '(he, she, it) went'

ju-haav-a (< -aau-) '(he, she, it) sacrificed'

ca-kaar-a '(he, she, it) did'

(These are generalizations and are not comprehensive. Each grade has other uses.)

So far, no one has fully worked out the functions of vowel alternation in OC.* Pulleyblank believes that the vowels ə/a** represent a 'extroverted'/'introverted' dichotomy, but it is hard to see what such a dichotomy has to do with a pair like (in a Pulleyblank-style reconstruction):

OC 皐 kə́w 'high' : OC 高 káw 'high'

Next: From nasals to vowels: further implications of the OC zero grade hypothesis.

*06.11.6.1:12: And I doubt anyone will. I was going to say that OC vowel alternations look about as opaque as those in English: e.g.,

(One could try to build a theory around the forms of sing, but notice how quickly it would fall apart.)

but at least we know they occur in English conjugation (and irregular plurals), even if we can't state simple rules for using each vowel. I still can't make such a statement about OC:

"So when is your so-called zero grade used?"

"Uh [əəə] ..."

(06.11.6.1:33: See the note below about my zero grade equalling Pulleyblank's -ə-grade!)

**06.11.6.00:48: Pulleyblank also (1991: 44) described OC ə/a alternation as a zero vs. infixed -a- contrast

... since ə is not an underlying phoneme but inserted epenthetically by rules of syllabification.

Therefore his -ə-grade is his version of my 'zero grade' (and about half the words in OC are infixed!?).

06.11.6.00:48: ADDENDUM: In 1998, 林德威 David Prager Branner asked, "Did Early Chinese Really Have Morphology"? In 2003, he revised this paper as "On Early Chinese Morphology and its Intellectual History":

The medieval Chinese tradition tells us that a given Chinese character may change its meaning when its reading is altered slightly. Modern scholars have sought principles for these changes, and from those principles have reconstructed a skeletal system of early Chinese morphology - with such elements as derivation by tone change, causative infixes [who proposed those? -AMR], transitivising prefixes, etc. Yet it is an arresting fact that some of pre-modern China's linguistically most astute scholars inveighed against the multiple readings on which this research is based.

Maybe everything you just read was a total waste! But even if we completely threw away everything but modern Chinese languages, some sort of morphology is implied by the everyday double readings of characters like 好:

Md hǎo 'good' (low rising tone) : Md hào 'to like' (high falling tone)

< OC xu' : xu'-s