14.8.23.23:48: WAS THE TANGUT IMPERIAL FAMILY THE MI OF WEI?

2339 1903 2ŋwɪ 1mi

from a hypothetical homophonous phrase

0395 4542 2ŋwɪ 1mi 'the cow feeds [someone]' / 'fed by the cow'

At the end of last month, I saw another derivation but couldn't remember what it was. I found it last night in Nishida (2010: 233):

It is very probable that the second syllable, miɦ (level 11), of ŋʷwɪ-miɦ (level 11) meaning "imperial family" was one of the corresponding cases of the [Tangut autonym] Mi. Its meaning might have been the Mi of Wei 魏.

The Tangut imperial family claimed descent from the Tuoba clan of the Northern Wei. Although this etymology is initially appealing, it has phonological problems.

First, the Tangut called the Wei

4962 2vɪ or 5574 2vɨi

rather than 2ŋwɪ. v- in those transcriptions reflects the loss of *ŋ- in the Tangut period northwestern Chinese pronunciation of 魏. Perhaps the imperial surname contains an earlier borrowing of 魏 preserving its nasal initial.

Second, the Tangut autonym

2344 2mi < *miH

has a 'rising' tone, whereas the second syllable of the surname has a 'level' tone. This tonal difference does not necessarily rule out a connection between the two names. The 'rising' tone of the autonym may be a reflex of a final glottal suffix *-H absent from the 1mi < *mi of the surname. Both 2mi and 1mi may be cognate to Tibetan mi 'person'.

14.8.22.23:44: A SILKEN SOURCE FOR THE RED RADICAL?

I'm surprised I was able to account for all uses of the 'red' radical

(Boxenhorn code: qie; Nishida radical 226)

in a straightforward manner in my last post. It means 'red' and/or is phonetic in all but one case (E):

A. n-phonetic	B. 'red'
E. 1tʂhɨĩ 'Chen' (a family on the land of the 2nie family?)	C. xŨ-phonetic in < B. 1xʊ̃ 'red'	D. -iã-phonetic in < B. 2ʔiã (1st syl of 'rouge')

I am normally at a loss to explain the function of a component in one or more tangraphs containing it. For instance, I have no idea what

the right side of

1671 1nie 'red'

is doing. It is in 65 other tangraphs. I think it is phonetic in

1674 2nie (second syllable of 2mi 2nie 'younger sister')

1809 2nie (second syllable of 1ɣɤə 2nie 'few')

which are near-homophones of 1nie 'red'. But what is it doing in, say,

3528 2tho' 'to harm, endanger'

whose analysis is unknown? Did red signify danger?

Going back to the other half of 1617 'red', I think

might be derived from the seal form of the top half (幺) of the Chinese 'silk' radical 糸 on the left side of Chinese 紅 'red'. The vertical line at the top of the Chinese 'silk' radical corresponds to the horizontal line of the Tangut 'red' radical, and the two circles correspond to the two

of the Tangut 'red' radical. If the admittedly vague similarity between the two radicals is just pareidolia on my part, did the Tangut simply draw a random line pattern and declare it to be 'red' and/or nie?

14.8.21.23:49: A RED RADICAL

Half of the nie-tangraphs (Tangut characters) from my previous entry contained the element

(Boxenhorn code: qie; Nishida radical 226)

that appears in twenty other tangraphs. Here are all 31 qie-tangraphs. Asterisks indicate words which are only in dictionaries to the best of my knowledge.

Class	Tangraphs	Li Fanwen 2008 #	Reading	Gloss
A1		0529 1671 1774 2231	1nie	2nd syl of 'be stifled to death'* red 2nd syl of 'servant' to try
A2		0547 0548 0593 1678 1723 1858 2239	2nie	surname syl 2nd syl of 'kind of insect'* 2nd syl of 'kind of grass' 2nd syl of 'chin' 2nd syl of 'colored silk'* 2nd syl of 'to hide'*, 'to turn around' bird name syl
A3	=+	0363	1nʊ	transcription
A4	=+	1235	1nĩ	red
B		0250 0260 0696 0820 1076 1164 1236 1402 1692 2237 4220 4880 4959 5102 5827	1bɪ̣ 1ŋa 1tʂwɤi 2dʐɨə̣ 1ʔie 1tʂhɨõ 2dʐɨə̣ 1xʊ̃2ʔiã 1dɪ̣ 2giẹ 2rəʳ 1tʂɨi 1giuʳ 2so ̣	sand red* red* red red jade necklace* red sand red red (Chn) 1st syl of 'rouge' (Chn) red soil* red wood* copper 2nd syl of 'rouge' (Chn) kidney millet
C	=+	1741	1xõ	transcription
D	=+	2049	2siã
E	=+	0298	1tʂhɨĩ	surname 陳 Chen

The 31 fall into five categories:

A. qie as n(ie)-phonetic: 13 tangraphs

A1. qie as 1nie-phonetic: 4 tangraphs

A2. qie as 2nie-phonetic: 7 tangraphs

A3. qie as n-phonetic: 1 fanqie tangraph (2nie + 1tʊ = 1nʊ)

A4. qie as n-phonetic / semantic for 'red' (see category B below): 1 fanqie tangraph (1nie 'red' + 1ʔĩ = 1nĩ 'red')

B: qie as semantic abbreviation of 'red': 15 tangraphs

C. qie as xŨ-phonetic: 1 tangraph

Cf. 1402 1xʊ̃ 'red' (Chinese loanword) in category B.

The Tangraphic Sea derived 1741 1xõ from 1671 1nie 'red' (whose Chinese translation was 紅 *xʊ̃) and 3682, first syllable of 2mə 1ʔɤõ 'merit' (which could be translated into Chinese as 勳 *xiũ)

D. qie as -iã-phonetic: 1 fanqie tangraph (1si + 2ʔiã = 2siã)

E. qie as abbreviation of a surname Ne or a surname containing the syllable ne: 1 tangraph

Was a Chen family in the Tangut Empire on the land of a Tangut family whose name contained Ne? The Tangraphic Sea derived the right half of 0298 from 2107 1tsɪʳ 'earth'.

14.8.20.23:59: PROXIMATE PRONUNCIATION

Yesterday I wrote about the transcription evidence and potential cognates of Tangut

1nie 'relative'

which originally may have meant 'near' (relatives being the people nearest to oneself).

Fanqie spellings expressing the pronunciation of tangraphs ( Tangut characters) in terms of the initials and rhymes of other tangraphs are only available for a little over half of the 6,000+ known tangraphs. Unfortunately, no fanqie are known for either 1nie or its second ('rising') tone counterpart 2nie.

Usually first ('level') tone tangraphs have fanqie in the surviving first volume of the Tangraphic Sea, but that volume is missing some pages including those which probably contained tangraphs for 1nie and other syllables with the 36th rhyme of the first tone.

The fanqie for most second tone tangraphs is probably in the lost second volume of the Tangraphic Sea. (Some second tone fanqie are in the surviving third volume Mixed Categories.)

Homophones lists 22 characters in a homophone group mixing 1nie and 2nie. All but one (0548) can also be found in Precious Rhymes of the Tangraphic Sea which has no fanqie for any of them.

Homophones	Tangraph	Li Fanwen number	Reading	Gloss	Tangraphic Sea	Precious Rhymes
13B31		1723	2nie	second syllable of 2ŋwəʳ 2nie 'colored silk' (only in dictionaries?)	in missing second volume?	✓
13B32		1671	1nie	red	in missing pages of first volume?
13B33		0547	2nie	the surname Ne (occurs as a first or second syllable in disyllabic surnames but unclear if it can occur by itself); transcription character	in missing second volume?
13B34		1858	2nie	second syllable of 1lɨa 2nie 'to hide' (only in dictionaries?; the first half can mean 'to hide' by itself) and 1gie 2nie 'to turn oneself around, look around; the other way around'
13B35		0593	2nie	second syllable of 2khwa 2nie 'a kind of grass'
13B36		0548	2nie	second syllable of 1lhə 2nie 'a kind of insect' (only in dictionaries?)	not in either book
13B37		1678	2nie	second syllable of 2miə 2nie 'chin'	in missing second volume?	✓
13B38		1774	1nie	second syllable of 2nieʳ 1nie 'servant'	in missing pages of first volume?
13B41		0529	1nie	second syllable of 1nie' 1nie 'to be stifled to death' (only in dictionaries?)
13B42		1732	1nie	first syllable of the surname 1nie 1xɤu
13B43		0806	2nie	second syllable of 2mɪ 2nie 'wind' (only in dictionaries?; the first half by itself is the name of the 'wind' trigram ☴)	in missing second volume?
13B44		1674	2nie	second syllable of 2mi 2nie 'younger sister' ('ritual language'? only in dictionaries?)
13B45		1809	2nie	second syllable of 1ɣɤə 2nie 'few'
13B46		1926	2nie	in the past
13B47		0213	1nie	relative	in missing pages of first volume?
13B48		2231	1nie	to try, second half of 1lɨe 1nie 'emissary' (the first syllable is 'to serve' by itself), first half of 1nie 2ʔwiəʳ 'writing on silk [cf. 1723 above with a different tone], written correspondence' (the second syllable is 'writing' by itself)
13B51		2239	2nie	second half of 2biu 2nie 'nightingale', first half of 2nie 2no 'cuckoo, oriole'	in missing second volume?
13B52		3671	1nie	first syllable of 1nie 2riaʳ 'father' (only in dictionaries?; the second half needs to be combined with either 1nie- for 'father' or -2si for 'mother')	in missing pages of first volume?
13B53		5147	2nie	first syllable of 2nie 1ɣa 'dog' (only in dictionaries?)	in missing second volume?
13B54		3846	2nie	optative prefix (< 'downward'), you (is this pronoun only in dictionaries?)
13B55		3817	2nie	to present a gift (only in dictionaries?)
13B56		0638	2nie	to compel, drive

Since Tangut dictionaries - both ancient and modern - are character-based, one might think 22 characters stood for 22 monosyllabic words pronounced nie, but in fact there are only three monosyllabic 1nie words and only three or four monosyllabic 2nie words:

1nie: 1. 'red', 2. 'relative', 3. 'to try'

2nie: 1. 'the surname Ne' (? - unsure if it can occur by itself), 2. 'in the past', 3. 'to present a gift', 4. 'to compel, drive'

8.21.2:42: It is tempting to try to derive the polysyllabic words from the monosyllabic nie-words, particularly since some of them are combinations of nie with monosyllabic words: e.g.,

1995 0806 2mɪ 2nie 'wind'

whose first syllable is also the Tangut name of the 'wind' trigram ☴. Could that word literally be 'red wind'? The trouble with that case and others is that 'red' is 1nie, not 2nie. One could try to salvage the etymology by proposing that 2nie in compounds is from 'red' plus an *-H suffix conditioning the second tone. But it is dangerous to build speculations atop speculations. Moreover, in this particular case, perhaps 2mɪ is an abbreviation of a monomorphemic, disyllabic 2mɪ 2nie 'wind'.

14.8.19.23:39: PROXIMATE PEOPLE

Today I saw Tibetan nye 'near' which brought to mind a possible Tangut cognate:

0213 1nie 'relative' (i.e., one's near relations; I covered related characters here)

This word was transcribed in Tangut period northwestern Chinese as 你 *ni. No Tibetan transcription is known, but its near-homophone

3830 2nie 'king'

with a different tone was transcribed in Tibetan as nye(H) and ne(H). (Tibetan ཉ ny- [ɲ] and ན n- are different letters.)

I normally derive Tangut rhyme 37 -ie from pre-Tangut *Cɯ-e:

*Cɯ-ne > *Cɯ-nie > nie

The -i- is a trace of the lost presyllabic high vowel *ɯ.

However, Tibetan nye makes me wonder if Tangut -i- in 'relative' is primary rather than secondary:

*ɲe > nie = [ɲe]? [nje]?

Similar Qiangic and rGyalrongic words for 'near' (see sections 3.2 and 3.3 of this list) have palatal ȵ- (= ɲ) or dental n-. (See items #1757-1758 here for very different rGyalrong words.)

Possible Old Chinese cognates have n-:

邇 *Cɯ-ne(j)ʔ (< *√n-e-j + -ʔ?) 'near'

暱 *Tnik (< *T- + √n-j + -k?) 'near, be familiar with'

Could those words contain e-grade and zero-grade forms of a root *n-j? Could the root-initial consonant have been *ɲ-? Were *Cɯ- and *T- the same prefix with and without a presyllabic vowel? Were *-ʔ and *-k variants of the same suffix?

14.8.18.23:36: SREDNJI KITAJSKI JĘZYK

I felt uncomfortable about mentioning Middle Chinese reconstructions in my last post because they may give the false impression that Chinese in the past was more homogeneous than it actually was.

It occurred to me last night that Middle Chinese is about as real as Interslavic, my favorite constructed language. If future linguists knew nothing about Russian, Polish, Serbo-Croatian, etc. - i.e., specific actual languages - Interslavic would have to do for comparisons with other European languages. The title of this post is Interslavic for 'Middle Chinese language'.

I suspect that diversity within Middle Chinese was like that between Slavic languages today. So my *kon for 昆in this table is to real Middle Chinese forms what Interslavic koń 'horse' is to these modern Slavic words: similar but not necessarily identical. Interslavic koń happens to match the actual Polish word for horse, but its vowel is very different from that of Ukrainian кінь [kinʲ] 'horse', and it is completely different from Russian лошадь [loʂətʲ] 'horse', a loan from Turkic. If a language borrowed a word kin 'horse' from Ukrainian, it would be strange to say that kin is from a 'Slavic' koń. Yet how many would blink if I wrote that Sino-Vietnamese mã is a loan from 'Middle Chinese' 馬 *mɤaˀ? (The actual source of mã was more like *ma with a 'rising' tone in a southern late Tang variety of Middle Chinese.)

'Middle Chinese' may sound specific, but it's actually a generic term like 'Middle Indic' which could refer to Pali, Gandhari, Ardhamagadhi, etc.

Unfortunately there is no analogous established terminology for specific varieties of Middle Chinese. It is easier to type a simple name like Pali than a phrase like 'Tangut period northwestern Chinese' (TPNWC), the dialect in the Timely Pearl in the Palm that is also the source of Chinese loans in Tangut.

Tonight I momentarily considered renaming TPNWC 'Zaric' after Tangut

1ɮar 'Chinese'

but that term would make no sense to those who didn't know the Tangut word. Although my older term is more tedious, it is also more transparent.

One could think of Middle Chinese reconstructions as being as open to intrepretation as Interslavic pronunciation: e.g., the ę of język 'language' in the post title could be [ʲa] ~ [ʲɛ] ~ [ʲɛ̃] ~ [ʲɔ̃] ~ [ɛ]; [ʲæ] is a suggested average.

That description of Interslavic states that "[a]ccentuation is free." Hence there is no way that one could figure out Serbo-Croatian tones from Interslavic: e.g., the falling tone of konj 'horse'. (Most of Slavic lacks tones, so Interslavic also lacks them.)

The situation is a bit different with Middle Chinese tones. The Old Chinese sources of Middle Chinese tones are known (e.g., *-ʔ in 'horse'), but their phonetic realizations are not. It is likely that *-ʔ left a trace as glottalization which disappeared at different times in different places (and is still present in today's Xiaoyi), and pitches once associated with glottalization became phonemic.

Although 'rising', the traditional name of the tone category for 'horse', suggests the tone was rising, that may not have been true in all Middle Chinese varieties, and it is certainly not true today: e.g., in Taiwanese, 馬 bé 'horse' has a high falling tone (indicated with an acute accent in romanization!). See Sagart (1998) for more on Chinese tonal history.

I have similarly used an acute accent to indicate the 'rising' tone in Middle Chinese varieties after glottalization was lost, but that accent may imply a rising or even high tone though I am actually agnostic about its contour, so I am reluctant to use it now. Maybe it's time to dust off my tone codes.

All of the above also applies to Old Chinese except for the part about tones since Old Chinese didn't have any. Old Chinese was not uniform before the mid-first millennium AD. In fact, 揚雄 Yang Xiong (53-18 AD) wrote the first Chinese dialect dictionary, 方言 Fangyan 'Areal Speech', toward the end of the old chines period. I think that oddities in Chinese loans in Vietnamese and Tai may in part reflect Old Chinese diversity that has been lost. Proto-Indo-European must also have been diverse.

Speaking of Proto-Indo-European, I don't understand how Proto-Indo-European *ḱem- 'hornless' became Proto-Slavic *konjь; why didn't PIE *ḱ- become PS *s- (cf. Sanskrit śama- 'domestic'), and why didn't PIE *-m- become PS *-m-?

14.8.17.23:51: WHEN B IS SPELLED G

If Vietnamese mắm [mam] < *ɓamʔ 'salting' could be written with the velar-initial phonetic 禁 cấm [kəm] (see my last two entries), could labial-initial syllables be written with velar-initial phonetics in sawndip, the traditional Zhuang script, as well? I looked through Sawndip sawdenj [Traditional Zhuang script dictionary] which I admit is a problematic source* and found the following characters with velar-initial phonetics for Zhuang [p]-initial syllables:

(8.18.0:54: Added Zhuang reading of phonetic component column. The title of this post should makes more sense now. I was referring to how Zhuang b-syllables were written with g-phonetic components.)

At least two phonetic components may actually be semantic: e.g.,

弓 *kuŋ 'bow' (archery) could refer to bending down in 足+弓 bomx 'to crouch'

外 *ŋwajʰ 'outside' could refer to going outside in 足+外 byaij 'to walk'

or it could have been chosen for a labial or labiodental initial ([w]? [v]? [m]?) close to by- [pj]

乎 *ɣo may be a reference to its homophone, the first syllable of 葫蘆 'gourd'.

The other phonetic components are baffling. If they are really phonetics, were they chosen only for their rhymes? Or did they have labial-initial readings in local varieties of Chinese?

*Holm (2011: 2) pointed out that

The Sawndip sawdenj is a useful compendium, but it provides no information about where the dialect forms come from, so it is impossible to see any patterns in geographic variation from this source.

Moreover, all the readings in Sawndip sawdenj are in standard Zhuang, even though the characters could be from all over the Zhuang-speaking world. Hence I presume many actual readings have been converted into hypothetical standard Zhuang equivalents. Such readings are strictly speaking not readings at all, since no literate native speaker would have ever used those hypothetical readings. Nonetheless I hope those hypothetical readings are close enough to the originals for my purposes here: e.g., b-[p] readings are most likely from nonstandard [p]-readings.