11.10.1.23:59: IS TANGUT QIANGIC ... AND DOES QIANGIC EVEN EXIST? (PART 1)

In historical linguistics, subgroups are ideally supposed to be defined in terms of common innovations: e.g., what distinguishes Indo-Iranian is the merger of Proto-Indo-European nonhigh vowels *a(a) *e(e) *o(o) into a single nonhigh vowel *a(a). The long e [ee] and o [oo] of Sanskrit and the e(e) and o(o) Avestan are not retentions of Proto-Indo-European *e(e) and *o(o): e.g.,

PIE *toi > Proto-Indo-Iranian *tai > Skt te [tee], Av tee 'they'

PIE *deiwo- > PII *daiwo- > Skt deva- [deeva] 'god', Av daeeva- 'demon'

cf. Latin deus

PIE *gʷeHu- > *gʷeeu- (Beekes 1995: 35) or *gʷoou- (Watkins 2000: 35) >

PII *gaau- > Skt gau and Av gaau-, gao- 'cow'

cf. Latin bos

PIE *nepoot- > Skt, Av napaat-  'grandson'

cf. Latin nepos

PIE *wookʷ- > Skt, Av vaac- 'voice'

cf. Latin vox

Note how Latin generally preserves the original mid vowels *e and *o. (Deus has -u- instead of -o-. Old Latin had -os preserving PIE *o instead of -us.)

If Tangut belongs to a subgroup of Sino-Tibetan, that subgroup should ideally be defined in terms of shared innovations. Nowadays Tangut is often classified as Qiangic. What innovation sets Qiangic apart from other Sino-Tibetan languages? Katia Chirkova (2011: 5) wrote:

The only (phonological) innovation for the Qiangic subgroup proposed so far is brightening, that is, a strong tendency for the Proto-Sino-Tibetan rhyme *-a to be raised and fronted to -i or -e in Tangut and modern Qiangic languages, as proposed by James A. Matisoff (2004). Matisoff discusses this development essentially in relation to Tangut, but he also points out a number of parallels in modern Qiangic languages. He argues that this development is unusual in the Sino-Tibetan context, and it is therefore a valuable criterion for membership in the Qiangic group. At the same time, Matisoff (2004:350) notes that modern Qiangic languages do not display brightening to the same degree, and that the phenomenon is not regular, either within the same language or cross-linguistically. The following observations regarding this development can furthermore be made. Relatively few items [with brightening] shared by both Tangut and modern Qiangic languages have so far been proposed (33 words in total, Matisoff 2004). Of these, even fewer are shared by more than four Qiangic languages at a time. Conversely, those that are shared by most Qiangic languages, such as 'salt' (in 12 languages) and 'rabbit' (in 9 languages) appear to be good candidates for cultural loanwords, and are hence inconclusive as to the genetic relatedness between the languages in question. Finally, this phenomenon is equally attested in non-Qiangic languages of the area, such as Na [Chirkova's term for Naxi] and Yí.

Brightening is found even in Sinitic languages far from that area. When I was in school, I was puzzled by a classmate's surname (余 Yee). At the time I only knew the character's Sino-Japanese reading yo and its Mandarin reading yu. I was completely ignorant of Chinese linguistic diversity and Chinese historical phonology. My classmate explained that Yee was the Hakka reading of 余. Later I learned about the history of 余:

Old Chinese *la > Late Old Chinese *jɨa > Middle Chinese *jɨə > *jy > Hakka ji

Chaozhou (a Min language) also has brightening, though in fewer morphemes than Hakka:

旅 OC *raʔ > Chaozhou and Hakka li 'travel'

(but Chaozhou for 余 is ɯ, not the expected i!)

Although Chaozhou and Hakka apparently share this sound change*, I have never heard of anyone who would subgroup Chaozhou with Hakka rather than with other Min languages. And of course no one would subgroup Chaozhou and Hakka with the Qiangic languages solely on the basis of brightening.

Qiangic languages are supposed to share a complex of up to twenty features in addition to brightening (Chirkova 2011: 4). How many of those features does Tangut share?

Next: One Down, Twenty More to Go

*Perhaps Chaozhou never underwent the sound change *y > i. There are two Chaozhou vowels corresponding to Old Chinese nonemphatic *a. I suspect one is native and the other borrowed:

OC *a > *ɨa > *ɨə > *ɯ (native)

In some non-Chaozhou language: OC *a > *ɨa > *ɨə > *y >

borrowed into Chaozhou as i (Chaozhou has no y)

According to this hypothesis, Chaozhou 余 ɯ is native whereas 旅 li is borrowed from a non-Chaozhou *ly. li presumably displaced a native *lɯ.

11.9.30.23:59: MORE COGNATES FOR TANGUT 2DZWIO 'PERSON'

Last night, I wrote,

If [Tangut] 2dzwio and [Nuosu] co are cognates, that would not necessarily indicate that Tangut is closely related to Nuosu rather than to the Qiangic languages because a similar word is in non-Loloish languages

The only uncontroversially non-Loloish language I mentioned was Namuyi but even its Qiangic status is not absolutely certain. Thurgood (1993: 17) wrote that "the definitive subgrouping evidence remains to be presented" for Namuzi (= Namuyi) and a number of other languages. Moreover, Andrew West pointed out

as its [= Namuyi] speakers live in the Liangshan Yi autonomous prefecture in Sichuan, they must have a lot of contact with Nuosu speakers, so tsho may well be a borrowing from Nuosu.

However, Andrew and I have found further non-Loloish languages with similar words or roots for 'person' at STEDT. Such forms presumably motivated LaPolla to reconstruct *tsaŋ at the Proto-Tibeto-Burman level.

Even more tso-words for 'person' in the STEDT database

I have only chosen one language per subgroup.

This word is too widespread to help us subgroup Tangut.

Next: Is Tangut Qiangic, and Does Qiangic Exist at All?

10.1.3:18: ADDENDUM: Can Tangut -o correspond to non-Tangut -aŋ?

1. Sino-Tangut correspondences

Gong (1981: 774) found correspondences between Tangut -o and Middle Chinese *-aŋ:

4710 1lo : MC 郎 *laŋ 'young gentleman'

0443 1dʒɨo : MC 長 *ɖɨaŋ 'long'

However, MC *-aŋ became *-o in the dialect known to the Tangut, and it is not clear whether these words were borrowed before or after that shift. Did Tangut and NW Chinese undergo the same sound change *-aŋ > *-o?

2. rGyalrong-Tangut correspondences

Guillaume Jacques (2003; published in 2006) found that gDong-brgyad rGyalrong -o sometimes corresponded to Tangut -o:

2857 2ŋo 'sickness' : gDBrG kɤ ngo (ŋo?) 'be sick'

In his doctoral dissertation (2004: 143), Guillaume reconstructed Proto-rGyalrong *-aŋ as the source of gDBrG -o. (PrG *-o became gDBrG -u.)

Is it a coincidence that *-aŋ became *-o in two distantly related languages (MC and gDBrG) and possibly Tangut as well? Perhaps. In Japanese, Middle Chinese *-aŋ was borrowed as *-aũ (<au> in historical kana spelling) which later monophthongized to *-ɔɔ and is now [oo].

Should I reconstruct the pre-Tangut source of 2dzwio as

*P-Nɯ-tshaŋH

It just occurred to me that all three parts look like Sino-Tibetan words for 'person':

*P resembles Old Chinese 夫 *pa 'man' and Benedict's (1972) Proto-Tibeto-Burman *(p)wa 'person'

*Nɯ resembles Old Chinese 民 *min or *miŋ 'people' and Tibeto-Burman mi-words for 'person'

*tshaŋH resembles LaPolla's (1987) Proto-Tibeto-Burman *tsaŋ

But I doubt that 2dzwio originated as a triple synonym sequence. Do such sequences exist?

11.9.29.23:59: DO TANGUT AND NUOSU 'PEOPLE' SHARE A CO-MMON ORIGIN?

Last night, I wrote,

This begs the question of where the regular Tangut word for 'person'

2541 2dzwio

came from. I don't know, but I presume it was an innovation that filled the semantic space left by *mi which came to mean the Tangut in particular rather than people in general.

Today Andrew West suggested that 2dzwio may be a cognate of Nuosu (Liangshan Yi) ꊿ co [tsho] 'person'. If the pre-Tangut root of 2dzwio were similar to the Nuosu form, it could have developed a voiced initial, medial -w-, and medial -i- through the addition of affixes:

*P-Nɯ-tshoH > *P-Nɯ-tshioH >  *P-NtshioH > *P-NdzioH > *P-dzioH > *dzwioH > 2dzwio

*P- conditioned medial -w-

*Nɯ- conditioned both voiced dz- and medial -i-

could this presyllable have been a reduction of *mi 'person'? Is 2dzwio a redundant compound 'person' + 'person'?

*-H conditioned tone 2, the 'rising tone'.

If 2dzwio and co are cognates, that would not necessarily indicate that Tangut is closely related to Nuosu rather than to the Qiangic languages because a similar word is in non-Loloish languages:

tso-words for 'person' in the STEDT database

(See the next post for more words in other subgroups.)

Jinuo is either a sister of Loloish according to STEDT or a Central Loloish language (Thurgood 2003: 8). In any case, Jinuo and (the rest of) the Loloish languages are in the Lolo-Burmese subgroup.

I wonder how I reconstructed 'person' in my 1993 Proto-Lolo-Burmese reconstruction which I don't have on hand.

2dzwio and *tsaŋ could be independent retentions from some higher-level subgroup of Sino-Tibetan*. Does Tangut share any innovations with Nuosu or other Yi languages?

*E.g., Proto-Qiangic-Lolo-Burmese. But I know of no reason to subgroup Qiangic with Lolo-Burmese.

11.9.28.23:59: DO THE QIANG AND TANGUT SHARE A COMMON AUTONYM?

(The post that was supposed to be here today was lost when I accidentally shut off my computer. I am reconstructing it and will post it later.)

I have long thought that the Tangut autonym

2541 2mi

was cognate to Written Tibetan mi 'person'* and that the Tangut called themselves 'the People'.

This begs the question of where the regular Tangut word for 'person'

2541 2dzwio

came from. I don't know, but I presume it was an innovation that filled the semantic space left by *mi which came to mean the Tangut in particular rather than people in general.

Today, Guillaume Jacques suggested that 1mi may be cognate to the Qiang autonym** which varies by dialect (Sun 1981: 2): rma, ʐme, χmɑ, mɑ. The autonym in the standard romanization "based on the Qugu variety of the Yadu subdialect of the Northern dialect" is RRmea [ʐme] (Huang and LaPolla 1996: 3, 6).

The correspondence of Qiang a-vowels to Tangut i reminds me of the correspondence between Tangut

1918 1mi 'not' (homophonous with 2mi 'Tangut' except for its first [= level] tone)

and Old Chinese 無 *ma. However, most of the Qiang forms have preinitials which I think go back to *r-. There is no trace of *r- in 2mi. A pre-Tangut preinitial *r- should have resulted in a Tangut retroflex vowel:

*rmiH > 2miʳ

There is, however, a Tangut word with similar semantics which has a retroflex vowel and is a good phonetic match for the Qiang words for 'Qiang':

3818 2mieʳ < *rɯ-meH 'person'

cf. gDong-brgyad rGyalrong tɯ rme 'person'

Another similar word is

3818 1miəʳ < *rɯ-mə 'person'

cf. gDong-brgyad rGyalrong tɯ rme 'person' (Jacques 2006: 12)

whose vowel reminds me of Taoping Qiang mə.

Could

3818 1miəʳ< *Cɯ-mə 'he, other'

cf. gDong-brgyad rGyalrong tɯ rme 'person' (Jacques 2006: 12)

also be cognate in spite of its semantics? Is there any other language in which 'person' became 'he' or 'other' (via 'that person', 'other person'?).

If the first syllable of the disyllabic autonym

3752 3296 2miə 2niaa < *məH Cɯ-nakH (ɲakH sans presyllable/)?

Written Tibetan Mi-nyag preserves the earlier *-k

is cognate to the preceding m-words, what is the etymology of the second syllable?***

Qiang has an r-less word for 'person' (Huang and LaPolla 1996: 368, Sun 1981: 201):

Ronghong mi

Taoping mə

Mawo nə < ?*ni < ??*mi

Perhaps most or all of these words share a common root with or without affixes and/or ablaut:

Note that Tangut and Qiang have opposite patterns.

*Old Chinese 民 *min or *miŋ 'people' may also be cognate.

**9.29.00:19: 羌 Qiang is a modern Mandarin descendant of an Old Chinese exonym variously reconstructed as

*kɯ-hjaŋ or skɯ-jaŋ (this site)

*k-hlaŋ (Sagart 1999: 238)

*kh(i)aŋ (Schuessler 2007: 426)

*C.qhaŋ (Baxter and Sagart 2011)

The graph 羌 contains 羊 'sheep' as phonetic and the word it represents may be cognate to 'sheep'. Schuessler compared it to Written Tibetan skyong-ba 'to guard, keep, tend (animals)' and glossed 羌 as 'herders'.

9.29.2:49: Could 2niaa be cognate to

0176 1nɨaa 'black'

cf. Written Tibetan nag-po 'black' (not ny-!)

in spite of the different medials (and different initials in WT?). Were the 2miə 2niaa the 'black people'? Cf. the term

2750 0176 1ɣʊ 1nɨaa 'black head'

for one of the two groups of Tangut.

Could

2519 2niaa 'wisdom'

be an adjective 'wise' after 'people'?

The other two niaa might be cognate to 1nɨaa 'black':

2015 2nɨaa 'excrement' (< *Cɯ-naH 'black thing'?)

<

4229 4116 1lhia 1nɨaa (kind of tree?; < 5167 0176 'black deer' with 'wood' on top?)

1lhia can also be 0480 'sage':

though if the tree were named 'black sage', I'd expect 4229 to be 0480 with 'wood' on top.

1lhia 'sage' looks like Written Tibetan lha 'god', though the -i- would not be expected if the former was borrowed from the latter.

11.9.27.10:10: HAPPY GREAT CYCLE, JIM SHOOTER!

Today is the 還 暦 kanreki (lit. 'return calendar') of Jim Shooter (emphasis mine):

One's "sixtieth" birthday, or alternately one's "sixty first" calendar year (Traditionally in Japan, when a person was born they were said to be "one," and at every New Year's day thereafter turn a "year" older. This leads to an age count that is usually one or possibly two years greater than one's age count in Western countries). When a person lives to see their sixty-first calendar year [= 60th birthday in the West] they have lived through the entire sixty-year cycle of the traditional eto (ten stems and twelve branches) calendar and returned to the same "year" and horoscope sign in which they were born.

Jim has completed the sexagenary cycle of the traditional East Asian calendar.

I wish to congratulate him in Tangut, the extinct language I have been studying since 1996:

Le chiu be re, Kha Jin!

lit. 'great cycle happy, Shoot(er) Jim!'

I don't know what the Tangut would have called kanreki, so I decided to call it le chiu, the 'great cycle'.

Tangut adjectives generally follow nouns.   be re 'happy' follows 'great cycle'. (But le 'great' can precede nouns.)

Tangut surnames precede given names.

Kha 'to shoot' is a translation of Shooter which in turn is a translation from Polish.

Jin has no meaning. It is a character created to represent non-Tangut syllables like Jim. Tangut doesn't allow final -m at the end of syllables. Jin (pronounced with a nasal [ĩ] rather than a true -n) is the closest possible Tangut approximation of Jim's given name.

Jim, may your next great cycle be even greater than the first!

11.9.26.1:26: RETURNING TO RHINOS

Andrew West brought up a number of points about my post on rhinos:

1. I linked to the Shuowen entry for 犀 'rhino' without commenting on it. Andrew drew my attention to its analysis of the graph as

㞑 (= 尾; other variants with up to 19 strokes) 'tail' (phonetic) + 牛 'ox' (semantic)

(1:33: Straightening the 乚 of 㞑 'tail' to 丨results in the top half of 犀 'rhino' which itself is a variant of 犀 'rhino' in 龍龕手鑑.)

However, Old Chinese 尾 *məjʔ 'tail' has an *m- that cannot be reconciled with any reconstruction of 犀 'rhino'. Even if I rewrote my proposed *Nʌ-si as *mʌ-si*, there is no precedent for writing *C₁V₁C₂V₂ with a phonetic for *C₁V₁V₂. The *-əj of 尾 is similar to -ʌ .. -i, but the medial *-s- of 犀 is a barrier that cannot be surmounted.

Could 'tail' be semantic? Andrew noted that rhinos are better known for their horns than their tails. However, he also pointed out that

Of course yaks are renowned for their tails, so maybe the character [犀] originally meant yak, and was later transferred to mean rhino ... in which case, maybe the early pre-Han sources that mention 犀 and 兕 actually mean yak and buffalo [more on this below] and rhinoceros was a misinterpretation applied during the Warring States period or the Han Dynasty.

If 犀 was 'yak', perhaps it was the native counterpart to 犛 'yak'***.

2. Andrew thought that 兕  looked like a water buffalo to him. If it originally represented a rhino, I'd expect a one-horned head 凸 instead of a two-horned head 凹 atop 儿 legs. (凹凸 normally mean 'concave' and 'convex'.)

3. Andrew linked to Berthold Laufer**'s Chinese Clay Figures. On p. 77 is a picture of a rhinoceros which a 17th century Jesuit called a 鼻角獸 'nose horn beast', a calque of rhinoceros 'nose-horn'. Could this term 鼻角獸 have reached Korea and inspired the Korean calque 코뿔소 khoppulso 'nose-horn-ox'?

*This *mʌ- might be part of a disyllabic root, as it could not be related to Sagart's (1999: 85) *m-prefix for the "names of small animals" in Old Chinese. In any case, there is zero reason to reconstruct a nasal initial of any sort for the presyllable of 犀 apart from a desire to regard it and 兕 as near-doublets deriving from a common *Nʌ-si with an additional *-ʔ suffix in 兕.

**Laufer was also a pioneer in Tangut studies. His "The Si-hia [i.e., Xixia = Tangut] Language: A Study in Indo-Chinese Philology" was published in 1916. Small world.

***犛 'yak' has two very different Middle Chinese readings, *lɨ and *mæw. Pulleyblank (1991: 51) derived them from a common Old Chinese root:

MC *lɨ < OC *R-ŋʷə̀w

MC *mæw < OC *R-ŋʷáw

I propose a common root *mV-r-w with vowel alternation:

MC *lɨ < OC *rə < *mɯ-rəw (lost presyllable; *r- > *l-)

MC *mæw < OC *m-raw < *mɯ-raw

But maybe the two are merely synonyms sharing the same character, as proposed by Karlgren (1957: 259).

Schuessler (2007: 348) proposed that MC *lɨ < OC *rə is "prob. cognate to, if not a loan from" Written Tibetan Hbri-mo 'domesticated female yak'.

If the root of 犛 was *mɯ-r-w, its phonetic 𠩺 could have represented words with a similar structure. 𠩺 has two Middle Chinese readings, *lɨ (like 犛) and an anomalous *xɨ. The latter may be from *hmrə < *smrə < *s-mɯ-rə, a prefixed variant of *lɨ < *rə < *mɯ-rə.

11.9.25.21:39: BUYANG: THE MISSING LINK?

0. Overview

On Friday night, I was surprised to see this in the opening paragraph of the Wikipedia entry on 布央语 Buyang:

It [Buyang] is important to the reconstruction of Austro-Tai [the hypothetical ancestor of Kra-Dai and Austronesian] as it retains the disyllabic roots characteristic of Austronesian languages [unlike other Kra-Dai languages which have monosyllabic roots]. Examples are /matɛ́/ "to die", /matá/ "eye", /qaðù/ "head", and /maðû/ "eight".

In short, Buyang might be the most Austronesian-looking Kra-Dai language.

I don't recall such disyllabic words in Ostapirat's 2008 book on the Kra languages including Buyang. Ostapirat reconstructed those words as monosyllables in Proto-Kra since they were monosyllabic in his Buyang data:

Ostapirat regarded Paha as a language distinct from Buyang rather than as a dialect of Buyang.

Sources of ABVD data (click on links for full bibliographical data):

Ecun Buyang: Kosaka, Li, & Zhou 1998, 仡央语言词汇集

Langjia Buyang: Li 1999, 布央语研究

Paha: Li and Luo 2010, The Buyang language of South China (listing 7 tones instead of Ostapirat's 5!)

PAN: Blust 1999

Notes on individual words

1. 'to die'

What is the source of Wikipedia's Buyang matɛ́ which doesn't match the other sources?

Ostapirat (2000: 239, 241) reconstructed 'die' and 'kill' nearly identically. Letter-number codes refer to tones.

None of the above resembles matɛ́.

Does Paha really have a single word for 'to die' and 'to kill'?

2. 'eye'

Ostapirat (2000: 220) listed no Kra forms with m-. Nonetheless, on p. 208 he reconstructed 'eye' with *m- to account for Paha d- as opposed to t- in other languages:

Although I agree with Ostapirat that there was a four-way distinction, I don't understand why he reconstructed a retroflex stop as the source of nonretroflex fricatives ð- and z-.

I also don't understand why he reconstructed *m- as a source of voicing in Paha as opposed to a generic nasal *N-. Was his choice of *m- influenced by the *m- of the Proto-Austronesian word for 'eye'? Even if Proto-Kra and Proto-Austronesian do share a word for 'eye', why can't the Proto-Kra word be *Nʈa [ɳʈa] with a nasal that assimilated to the following retroflex stop? Ostapirat's preinitial nasal *m- does not contrast with *n- or other preinitial nasals.

3. 'head'

Ostapirat's Proto-Kra *krai bears no resemblance to Proto-Austronesian *qulu. Is *krai a native Kra word? Are the qa ðu-type words from a loanword into Proto-Kra? Or are qa ðu-type words inherited from a common ancestor of Proto-Kra-Dai and Proto-Austronesian? Was *krai a South-Western Kra innovation? (Cognates of *krai are absent in Central-East Kra languages: Paha, Buyang, and Pubiao.)

Thai เกล้า klaw 'head' (whose tone probably goes back to *-ʔ) is similar to Old Chinese 首 *hluʔ. Does OC *hluʔ go back to *qluʔ? Is 'head' an area word that spread among unrelated languages, or is it evidence of shared ancestry? I would not link Sino-Tibetan, Kra-Dai, and Austronesian together solely on the basis of one word.

I wonder if 首 *hluʔ is a loan in Chinese that displaced 后 *ɢoʔ 'head' which became 'ruler' in OC. I don't know of any non-Chinese cognates of 首 *hluʔ in Sino-Tibetan, but 后 *ɢoʔ 'head' has cognates: Written Tibetan mgo and Tangut 1ɣʊ.

Did medial *-l- become -ð- in Buyang, or does Buyang preserve a *-ð- lost in Proto-Austronesian? (There is no *ð in the Proto-Austronesian consonant inventory.) According to Ostapirat (2000: 189-191), Buyang ð- is from *r- and *-r-, not *l. So do the non-Ostapirat Buyang forms come from *qaru with *-r-? Was some descendant of PAN *qulu  like *qəru or *qaru borrowed as *qaru?

Is ABVD Paha mau unrelated to the other forms, or is it a reduction of a prefixed *m-qaru?

4. 'eight'

The various Kra forms point to *maru. (See above for Buyang ð- indicating *r-.) Is the resemblance to Proto-Austronesian *walu coincidental? Was some descendant of PAN *walu  borrowed as *waru and then prefixed with *m- (why?)?

Thai แปด pɛɛt is a borrowing from Middle Chinese 八 *pɛt 'eight'.

5. Why I care

The "Significance" section of the Wikipedia article on Kra languages begins,

Several Kra languages have consonant clusters and disyllabic words, whereas other Kradai languages only have single consonants. One such language, Buyang, has been used to support a proposed connection with the Austronesian family.

The assumption - which I share - is that more complex forms with clusters and disyllables are original and that simpler forms are secondary. The reverse is highly unlikely:

*CV > CCV

*CV > CVCV

The development of the longer proto-forms implied by Kra into the shorter forms in other Kra(-Dai) languages may be parallel to the development of longer proto-forms into shorter Chinese and Tangut forms. Thus studying Kra may lead to insights in Chinese and/or Tangut historical phonology.