I was happy to see Tangut, Jurchen, and Khitan discussed - and not just in brief mentions - in Peter Francis Kornicki's Languages, Scripts, and Chinese Texts in East Asia (2018). Search for 'Tangut', 'Jurchen', and 'Khitan' in the Google Books preview to see what I mean.

Kornicki's decision to avoid the sinocentric names Liao and Jin for the Khitan and Jurchen states reminds me of my own preference for 'Khitan Empire' and 'Jurchen Empire'. I have, however, been using Liao and Jin to refer to the ruling dynasties of those empires, not the empires themselves. But Kornicki has inspired me to start referring to the dynasties by their family names (cf the Yi dynasty of Korea).

Unfortunately, the reconstruction of the names of the Khitan and Jurchen ruling dynasties is still uncertain, so for now I use their Mandarinized versions. I'm more certain about the name 𗼨𗆟 Ngwimi, so I don't need to use the Mandarin version Weiming.

2. Speaking of names, last night I learned that 石田ひかり Ishida Hikari, an actress I haven't seen in over thirty years, has now married and taken the name 訓覇 Kurube which isn't in O'Neill's Japanese Names.

訓 is normally read kun. O'Neill doesn't include kuru as one of its name readings. I have wondered if  Chinese *-n characters representing Japanese CVrV sequences reflect (via Sino-Paekche) late survivals of an Old Chinese *-r that had not shifted to *-n. Unfortunately, there is no Chinese-internal evidence pointing to *-r instead of *-n in 訓.

覇 is normally read ha < *pa. Until now I would have regarded he < *pe as a hypothetical Go-on reading created for modern dictionaries by analogy with actual Go-on readings in the same rhyme class, but now I think there must have been a Go-on *pe since the name 訓覇 (prewar 訓霸) Kurube is attested in 和名類聚抄 Wamyō ruijūshō (Japanese names [for things], classified and annotated, 938), presumably before the creation of artificial readings for dictionaries and about four or five centuries after the importation of Go-on.

3. I still don't know why the Sino-Korean reading of 霸 is 패 phae < *phay. The expected reading is pha (for some reason p- cannot precede the rhyme -a in Sino-Korean, though pak, pan, pal, etc. are possible), and the idealized reading in 東國正韻 Tongguk chŏngun (Correct Rhymes of the Eastern Country, 1448) is pá, not páy.

Although it is tempting to link phay to the front vowel of Go-on *pe from topic 2 above, the latter reading reflects a Middle Chinese *æ̤ from Old Chinese *-raks without anything *-y-like. Korean *-y in 霸 seems to be a Korean-internal innovation. Crazy idea: *-y is the noun suffix *-y added to *p(h)a 'tyrant' and reinterpreted as part of the reading.

(I write *(h) in parentheses since I don't know whether the initial was aspirated before or after *-y addition.)

4. I checked 小學堂 Xiaoxuetang to see if any Chinese variety there had a *phay-like reading of 霸. None did. But I did find this bizarre triplet of readings in 三林塘 Sanlintang Shanghai:

p/tɕ uo/a/i 35/53

I think that's supposed to be read as

puo³⁵ / pa³⁵ / tɕi⁵³

I'm guessing that the first two are the same morpheme in two different strata (borrowed and native) and that the third is an unrelated native synonym.

5. I wish I could help make khitan.info and jurchen.info, hypothetical companions to Alan Downes' tangut.info.

6. Timothy Michael O'Neill's Ideography and Chinese Language Theory: A History (2016) looks interesting.

The introduction introduced me to 譱 <LAMB.SPEAK.SPEAK> (using the glosses of Joseph de Prémare, SJ, for the components), an old form of modern 善 for classical 'to regard someting as good, beautiful'.

7. I just realized that Japanese 匹 hiki < *piki (which does not match Middle Chinese *pʰit) might reflect (via Sino-Paekche) the late survival of an Old Chinese *-k that had not shifted to *-t after *i. Unfortunately, there is no Chinese-internal evidence pointing to *-k instead of *-t in 匹.

19.11.8.23:59: TANGUT.INFO

Today I couldn't use my computer, so instead of writing part 2 of "Do Korean and Japanese Share a Copula?", I started reading Alan Downes' PhD dissertation "How Does Tangut Work?" on my Kindle. I downloaded it over a year ago but never gave it the attention it deserves until now. Unfortunately now that I can use my computer again, I can't give it the attention it deserves on my blog right now.

All I can do is link once again to his site tangut.info and recommend his Tangut Character Lookup tool on the front page. If you don't know any Li Fanwen numbers for Tangut characters, try inputting a random number between 1 and 6000 in the "Li Fanwen Number" field or inputting English into the "Meaning" or "Keyword" fields. When you reach the results page, select "Show Concordances" and be amazed. Right now Downes' concordance is limited to part of the Tangut law code, but imagine an even bigger future concordance encompassing a larger number of texts. And - once Khitan and Jurchen are in Unicode - similar tools for those languages.

I wish Downes applied his computational powers to the Khitan and Jurchen scripts. I don't know of any computational analysis of Khitan since the efforts of Starikov's team in Russia (1964-1986)*, and Jurchen might be terra incognita.

*11.9.0:12:

This was a significant step towards an actual decipherment of the [Khitan small] script, but, unfortunately, the work was discontinued before it had proceeded to the level of phonological reconstruction.

- Wu and Janhunen (2010: 22)

19.11.7.23:59: DO KOREAN AND JAPANESE SHARE A COPULA? (PART 1)

1. Bjarke Frellesvig (2001) proposed that Korean and Japanese had a shared copula, and Alexander Vovin (2008: 546-547, 2010: 73-76) proposed that this copula was a Korean borrowing in Japanese rather than an inheritance from a common proto-language.

There are two or three phonological problems with either version of the shared origin hypothesis (inheritance or borrowing).

(11.8.22:19: What follows deals with Vovin's  [2008: 547] formulation

Middle Korean írò- < Old Korean *ito > Western Old Japanese tə

converted above into my notation. Frellesvig's formulation is far more complex, and I cannot do it justice in a short post.)

First, the Middle Korean copula is írò- which could be from an earlier *írò- or *ítò- with *-t-lenition. Both versions of the shared origin hypothesis require *t, though there is no Korean-internal evidence for it; the only evidence is the t of Old Japanese tə 'be'. But I'm OK with using foreign data to resolve ambiguities in reconstruction, so I don't see this as an issue. I do, however, see the next two issues as harder to overlook.

Second, the Korean form has an í- that corresponds to nothing in Japanese. Why doesn't the Japanese form begin with i-? Could Middle Korean írò- be a  redundant compound of í- (also attested as a copula) with *tò, a root 'be' (?) possibly shared with tʌ̀ɣòy- ~ tʌ̀βʌ̀y < *tʌ̀pʌ̀y < *tò-pʌ̀y? 'become'. Cf. other redundant copulas like Classical Japanese tar- 'be' < tə 'be' + ar- 'be'.

Maybe Paekche, the Koreanic language most in contact with Japanese and hence the most likely source for Koreanic loans in Japanese, had an uncompounded root *tò- (or had reduced a disyllabic root *ítò- to *tò-).

Third, the o in the Korean form should correspond to Japanese o or u < *o, not ə.

So in short, the two copulas only have a single consonant t in common (and even that t is shaky in Korean - if not for the proposed Japanese etymology, there would be no reason to favor reconstructing *t instead of *r).

11.8.22:44: Here's a table showing how the Korean and Japanese forms line up:

There are only two Old Japanese verbs with infinitives in -ə, tə 'be' and its homophone tə 'say'. The Middle Korean infinitive is -a ~ -ə depending on the vowel of the preceding verb root. Could tə 'be' be a direct borrowing of an inflected Paekche *t-ə 'be' + infinitive suffix? If Frellesvig (2001) is right, and tə 'be' and tə 'say' are etymologically one and the same, then tə 'say' would also be a direct borrowing of a Paekche infinitive.

That solution is not without its problems. First, no Paekche verb paradigms are known, so we don't know what the Paekche infinitive suffix was. Second, the Old Korean infinitive suffix seems to have been -a, not -ə; the latter developed after Korean developed vowel harmony.

2. You'd think that I'd know the songs of Kojiki well after having typed them all out for use as data in my PhD dissertation. No. It doesn't help that was over twenty years ago. So this morning I had to use this search engine to identify poem 31:

伊能知能

inəti-nə

life-POSS

麻多祁牟比登波

mata-k-em-u pitə-pa

complete-ATTR-TENT-ATTR person-TOP

多多美許母

tatamikəmə

幣具理能夜麻能

peŋguri-n-ə yama-nə

Heguri-be-ATTR mountain-POSS

久麻加志賀波袁

kuma-kasi-ŋga-pa-wo

bear-oak-POSS-leaf-ACC

宇受爾佐勢

unzu-ni sas-e

headdress-be-INF stick-IMP

曾能古

sənə ko

that child

Here's the full context.

Who converted the all-phonogram 8th century script into the modern mixed kanji-kana orthography that appears in the July 1958 issue of 日本古典文學大系月報　Nihon koten bungaku taikei geppō (Japanese Classical Literature Series Monthly)?

命の　全けむ人は　畳薦　平群の山の　熊白檮が葉を　髻華に插せ　その子

Chamberlain's translation (1932: 266):

'Let those whose life may be complete stick

[in their hair] as a head-dress the leaves

of the bear-oak from Mount Heguri, -

those children!'

11.8.23:39: I forgot the whole point of quoting that poem: commenting on the choices in the modernized orthography.

2a. kuma-kasi 'bear oak' as 熊白檮 <BEAR WHITE STUMP>

kuma is 'bear', but kasi 'oak' doesn't map in a straightforward way onto <WHITE STUMP>. I suppose oaks are characterized by white stumps.

Almost thirty years ago I asked H. Mack Horton if I had to be a botanist to be a Japanese literature scholar. I can't remember his answer. I'm obviously neither.

2b. unzu 'headdress' as 髻華 <MIZURA FLOWER>

The word unzu itself has no internal structure, so it doesn't split up into mizura (< Old Japanse mindura), an ancient Japanese male hairstyle, and 'flower' (Old Japanese pana).

The mizura (see it here) didn't exist in China, so its name was spelled semantically with various repurposed characters:

• 髻 <HAIR.COIL>

• cognate to 結 <TIE>?
  

• 鬟 <HAIR.COIL>

• cognate to 環 <RING>?
  

• 角子 <CORNER CHILD>

• 角 referring to how the tied hair of a mizura is on the sides (corners?) of a head?
  

• 子 is a noun suffix not to be taken literally
  

• 角髪 <CORNER HAIR>

11.9.20:09: Another name for the mizura is agemaki <  aŋgəy-maki 'raise-coil', written 總角 <ALL CORNER>.

3. The Japanese Wikipedia article on 漢文訓読 kanbun kundoku 'reading literary Chinese as Japanese' says that Khitan and Uyghur had their own versions of that. Is that a reference to Hong Mai's anecdote about how Khitan children read Chinese in Khitan word order? (See Kane 2009: 130.) What is the evidence for Uyghurized Chinese?

19.11.6.23:59: THE DAMAGED TEXT OF YELÜ DILIE'S EPITAPH: LINE 14

1. Looking at Andrew West's 2011 photo of the Khitan small script epitaph of 耶律迪烈 Yelü Dilie (1092), I see that the □ in Kane's (2009: 198) edition corresponds to damage in the third from last block in line 14, though Kane transliterates  □ as □, whereas he transliterates ⌧ as [damaged]. Unfortunately, all instances of Kane's ⌧ correspond to the tops of lines 33-38 which are hard to see in the photo. So I still can't figure out the difference, if any, between Kane's symbols ⌧ and □. For a moment I thought he might have changed his symbol for damage from □ to ⌧, but both symbols coexist in lines 33, 36, and 38 of his edition.

2. Sahaptin has long versions of all vowels except /ɨ/. Is there a historical reason of that?

3. Today I finally got around to looking at Valerie Henitiuk's Worlding Sei Shônagon: The Pillow Book in Translation (2012) when looking for Lone Takeuchi's A Study of Classical Japanese Tense and Aspect (1987). Wish I had the time to compare all 48 translation samples. Not that Google Books' preview would let me.

4. Michal Biran's "The Non-Han Dynasties" (2017) is a nice, short overview of the TJK (Tangut/Jurchen/Khitan) empires and their Mongol and Manchu successors.

5. Two lines of poetry caught my eye in Alexander Vovin's A Descriptive and Comparative Grammar of Western Old Japanese:

5a. 那賀那加佐麻久阿佐阿米能疑理爾多多牟叙

<na ŋga na ka sa ma ku a sa a məy nə ŋgɨ ri ni ta ta mu nzə>

na-ŋga nak-as-am-aku asa-aməy-nə kɨri-ni tat-am-u-nzə

you-POSS cry-HON-TENT-NML morning-rain-COMP fog-LOC rise-TENT-ATTR PT

'your weeping will rise into fog like the morning rain' (Kojiki song 4; analysis and tr. by Vovin 2008: 846)

Does the spelling 疑理 <ŋgɨ ri> indicate that /kɨri/ 'fog' was pronounced [ŋgɨri] with nasalization of /k/ spreading from the preceding comparative suffix /nə/ with a nasalized vowel [ə̃]?

5b. 吾妹子之阿乎偲良志

<I YOUNGER.SISTER CHILD si a wo LONG.FOR ra si>

wa-ŋg-imo-ko si a-wo sinop-urasi

I-POSS-beloved-DIM PT I-ACC long.for-SUP

'It seems that my beloved longs for me' (Man'yōshū XII: 3145; analysis and tr. by Vovin 2008: 681)

The modern Japanese reflex of sinop- is shinob- < *sinomb-. Is this another case of nasality spreading into a following consonant?

sinop- > *sinõp- > *sinõmb- > *sinomb-?

11.7.21:47: *sinomb- was later confused with the unrelated verb *sinomb- < sinəmbɨ- 'to conceal, to endure'. Both verbs are written with sinographs originally representing Chinese morphemes with different meanings:

• 偲ぶ <TALENT/URGENT bu> shinob-u 'to long for'

• 忍ぶ <ENDURE bu> shinob-u 'to endure/conceal'

偲 <PERSON.THINK> might have been interpreted as a semantic compound 'thinking of a person' appropriate for writing sinop- 'to long for'.

忍 <ENDURE> does match sinəmbɨ- in the sense of 'endure' but not 'conceal'. Is 'endure' an extended usage of 'conceal' ('conceal' > *'conceal discomfort' > 'endure')?

忍 in the sense of 'conceal' is well-known as the nin of 忍者 ninja 'concealer' and in 忍び shinobi 'art of the ninja' (lit. 'concealing').

6. My copy of 古代歌謡集 Kodai kayōshū (A Collection of Ancient Songs, 1958)  has a newsletter (日本古典文學大系月報　Nihon koten bungaku taikei geppō (Japanese Classical Literature Series Monthly) in postwar simplified Japanese orthography with at least one exception: 關聯 kanren 'connection' instead of postwar 関連. A slip?

11.7.22:12: I'm not counting how the title 日本古典文學大系 is consistently spelled that way instead of as postwar 日本古典文学大系. Or the subtly different prewar forms of 錄 and 卷 instead of postwar 録 and 巻 in the title section. (Postwar 巻 appears in the body on p. 1. I can't find 錄/録 in the body.)

7. Origin story of the day: Japanologist Hugh Cortazzi's in his own words. (Also found that when looking up Lone Takeuchi.)

8. I might have seen an aphid on my phone screen last night. Wikipedia says 'aphid' in Korean is 진딧물 <c.i.n t.i.s m.u.r> chindinmul which looks like chindi- + genitive -s + -mul. (Genitive -s assimilates to the following nasal: /sm/ > nm.) Martin et al. (1967: 1545) define 진디 chindi by itself as 'aphid' and chindinmul as 'a nest of aphides' (not 'aphids'; 11.7.0:22: aphides is the plural of Latin aphis; the stem is aphid-).

물 mul looks like a shortened (or unsuffixed?) form of 무리 muri 'group'. I suspect Japanese 群れ mure 'group of animals' < *mura-i is a borrowing of a Koreanic *mur plus a Japanese filler vowel *a and noun suffix *-i.

(11.7.0:13: Sakihara's 2006 Okinawan dictionary has no entry for a cognate of *mura-. If there are no Ryukyuan cognates of *mura-, then *mura- may be a borrowing from Koreanic [specifically Paekche?] into mainland Japanese after it split from Ryukyuan.)

Martin et al. derive chindi 'aphid' from <c.i.n t.ŭ k.i> 진드기 ~ <c.i.n t.ŭ.k Ø.i> 진득이 chindŭgi 'tick, mite, louse' with irregular -g-loss. They in turn derive chindŭgi from chindŭk-i with a noun suffix -i but do not define chindŭk. I suspect that chindŭk is the same chindŭk that is in

• 진득거리다 chindŭk-kŏrida 'to keep sticking'

• -kŏrida is 'to do ... repeatedly'
  

• 진득진득하다 chindŭk-chindŭk-hada 'to be sticky'

• hada is 'to be, do'.
  

Was chindŭgi originally 'clinger'?

1. I'm still slowly copying out the Khitan small script epitaph of 耶律迪烈 Yelü Dilie (1092) as published in Kane (2009: 191-211) because I don't have access to any photos. (11.6.1:00: I do now!)

I forgot to ask last week - what is the difference, if any, between ⌧ and □ in the printed text? ⌧ is transliterated as [damaged] whereas □ is transliterated as □. But aren't both forms of damage?

What caught my eye was what appeared tho be a rare eight-character block <⌧⌧⌧⌧⌧⌧⌧eu> at the start of line 33. I don't remember seeing any blocks with more than seven characters. (Here are the standard block layouts. Other possibiities are vertical stacks, 'diamonds', and 'pyramids'.)

2. Last night I saw Jennifer Taylor (née Bini) on Two and a Half Men. Bini is a name reduced to a b- (e.g., Iacobo) plus a diminutive suffix -ini. Are there any other surnames of that type?

3. I would have never guessed that Touchet was pronounced [ˈtuːʃi]. The -t seems nonetymological (the original name was Sahaptin tu-se) and presumably was added by analogy with rhyming French words ending in -t. I'm guessing French [tuʃe] was then Anglicized as [ˈtuːʃi] with final [i] in place of [e] (cf. how karate was borrowed as [kʰəˈɹɑti]). American English has [ej] and [i] but not [e].

4. I also would have never guessed that Muth in America was pronounced [mjuːθ]. I would have guessed [muːt] as in German or an Anglicized [mʌθ].

5. Proof Hollywood German has only one gender: Der Waffle Haus. GAHHHH!!

Or should I say Vancouver German? Dead like Me was filmed in Vancouver.

19.11.4.23:14: PRE-TANGUT *O BEFORE *CORONALS

1. Yesterday, I mentioned how pre-Tangut *o shifted to *y before *-r but not in open syllables in my interpretation of Jacques (2004: 206). Compare:

• *-or > -yr¹

• *-o > -u (not ˣ-y!)

A similar *y-shift occurs before *-t (which is subsequently lost):

• *-ot > *-yt > -y
  

Other codas do not have y-reflexes in Tangut:

• *-ok > (> *-ow? >) -o

• *-oj > -o

• *-op > (*-yw? >) -ew (did this ever undergo *y-shift? Should -ew be rewritten as -yw?)
  

• *-om > -on (nasalized o, not o + a coda n)

If *-oŋ and *-on existed, I don't know what their reflexes are.

If *-op never became *-yw, I can say that *o became *y before coronals *-t and *-r (and *-n, the nasal counterpart of *-t?). But why would *o lose its labiality and become achromatic (nonlabial and nonpalatal) y before coronals?

Perhaps the key lies in the fact that *i also underwent *y-shift in even more environments than *o: before *-Ø and *-p as well as before *coronals.

Given how *o fronted before coronals in Lhasa Tibetan

• *od > [ø]

• *on > [ø̃ː]

• but *or > [oː] (not [øː]!)

I think this might have happened in Tangut:

*o-fronting in Tangut is like *o-fronting in Lhasa Tibetan, though the conditioning codas differ.

It is interesting that *o did not front (i.e., become palatal) before the palatal (and hence dorsal and noncoronal) coda *-j. *o was preserved before final glides: *-j and *-w < *-k.

The *ø that resulted from*o-fronting has no relation to Grade IV o which I think might have been [ø]. Compare:

(I leave out Grades II and III for simplicity.)

In the above scenario, grades developed after *ø > *e but before *e > *i > y:

• words without *high vowels developed Grade I

• words with *high vowels developed Grade IV

By the time *e raised to high *i, height-driven grading was over, so *e1 would not become *i4; it became *i1, retaining its grade.

*ø > *e may have something to do with the *o > *e shift before *-p. The lack of parallelism between *-op and *-om might indicate that the latter no longer had a labial coda by the time *o dissimilated to *e before *-p (or did *-p already lenite to -w?).

Did *-om become a nasal vowel when *-op > *-ep/*-ew? Or did *-om become *-on after *o-fronting (so this new *-on never became *-øn)?

Later, *-eT (*T = *t or *r) merged with *-iT. That resulted in a large number of *-i(T) syllables which were all subject to the *i > y shift. I don't know whether that shift predated or postdated coda loss. (The -r in the Tangut column is not a coda.)

¹-r in unstarred Tangut forms indicates vowel retroflexion unlike pre-Tangut *-r which is a true liquid coda.

In yesterday's tables, none of the Tangut forms ended in -r because all of the forms had a preceding *S- at the pre-Tangut level. That *S- conditioned vowel tension (written as -q) which could not coexist with retroflexion: *SCVr > CVq (not ˣCVrq).

2. This is a strong statement (via Joanne Jacobs; emphasis mine):

Once a week, children wear a vest that includes a pocket for a listening device officials refer to as a “word pedometer.” The device, made by the nonprofit LENA, syncs with an online program that counts how many words the children hear each day, but it does not recognize which words are exchanged. The system works with any language, and it can differentiate between words broadcast by a TV or computer and those spoken by a person.

If the online program "does not recognize which words are exchanged", how does it 'know' what is and isn't a word? My fear is that it doesn't 'know'; it might be a syllable detector which indeed would work "with any language", and the word counts might be the number of syllables divided by some figure for the average number of syllables in an English word. Which might not be the average number of syllables of a word in some other language. And let's not even get into the issue of what counts as a word.

3. What is the invertive case of Kabardian? Is it the same thing as the adverbial case?

4. I just heard Kelly Clarkson pronounce Bebe Rexha as [ˈbiːbi ˈɹɛksə]. Ouch. Will Albanian xh = [dʒ] ever become common knowledge? Will it become fashionable to pronounce xh as [ʒ], a consonant that doesn't even exist in Albanian? Some English speakers seem to think [ʒ] is the foreign sound (e.g., Beijing as [bejˈʒɪŋ]), presumably due to its low frequency in English and its presence in French.

19.11.3.23:59: ABC COMPRESSION

1. Today I found the Korean-language YouTube channel of Oliver Ssaem. 쌤 ssaem is a compressed slang form of Korean sŏnsaengnim 'teacher'. I'll call that ^AB^C compression: i.e., compression at both ends that turns a trisyllabic word into a monosyllable. Here's a case of Tangut  ^AB^C compression from Jacques (2014: 126; somewhat modified here; he reconstructs disyllabic *S-kar-u):

𗕐 1252 1kyq4 'to frighten' < *SI-kar-u

The A (*SI-) has left traces as -q4: i.e., tension (written as -q) and Grade IV (high and perhaps palatal?): y4 = [jɨ]?).

The C (*-u) has fused with *-ar into *-or which ultimately became -y. Normally *-r conditions vowel retroflexion, but Tangut vowels cannot be both tense and retroflex at the same time. It seems that if the conditioning factors for both tenseness and retroflexion coexist - as they do in - tenseness dominates. *-r must have been lost after *-o > *-u. Otherwise *-o after *-r-loss would have raised to *-u and become -wy, not -y.

11.4.17:28 (based on Jacques 2014: 206):

What I think actually happened to rhymes of *S-r syllables (omitting how *I conditioned Grade IV for simplicity)

*-o shifted to y (which was some sort of achromatic vowel: neither palatal nor labial) before *-r. *-u underwent a similar shift to *-wy with or without a following *-r. Following y-shift, pre-Tangut had no *-u, so *-o raised to fill that gap.

By the final stage (Tangut), *S- was lost after conditioning vowel tension (written as -q) at some earlier point, and *-r was lost. I have deliberately ignored the *S- > -q shift until the Tangut stage, since (1) I cannot date that change and (2) my interest here lies in the relative chronology of *o-raising and *r-loss.

*r-loss normally left behind retroflex vowels (which I still write with final -r for convenience), but in the above cases, *-r left no trace when preceded by *S-V because Tangut does not permit tense retroflex vowels: i.e., ˣVrq.

What would have happened if *o-raising occurred after *S-tension and *r-loss and before *y-shift

In the above incorrect scenario, the *-r that blocked *-o from *y-shift disappeared early, so *S-or and *S-o merged into *-oq which merged with *-uq and underwent *y-shift.

What would have happened if *o-raising occurred after *S-tension and *r-loss and after *y-shift

That's closer to what I think actually happened, but still not quite right.

I'd like to work out a full relative chronology of Tangut sound changes.

2. John McWhorter uses Hmong as an example of a language with lots of tones. Dananshan Miao (= Hmong) has eight tones. I think of Kam as the record-holder with nine tones, but I just learned that

Preliminary work on the Wobe language of Liberia and Côte d'Ivoire and the Chatino languages of southern Mexico suggests that some dialects may distinguish as many as fourteen tones, but many linguists believe that many of these will turn out to be sequences of tones or prosodic effects.

Wikipedia describes the fourteen tones of Wobé (the height classes are mine):

Wikipedia describes a ten-tone variety of Chatino (converted to a 4 = high/1 = low scale; again, the height classes are mine):

I wonder what the frequency of each tone is.

(11.4.16:52: I forgot to mention this hypothesis:

tone languages are less likely to develop in dry environments because dry air deprives the vocal cords of the suppleness required to produce subtle differences in tone.)

3. When did ants first arrive in Hawaii? Ants were first scientifically recorded in Hawaii in 1879. Yet there are several native words for 'ant' which appear to be related to each other:

naonao, nonanona, ʻānonanona

There was a newspaper, Ka Nonanona 'The Ant' (1841-5), named after this saying:

E hele ʻoe i ka ʻānonanona, e nānā i kona ʻaoʻao e hoʻonaʻauao iho.

'Go to the ant, study her ways and learn.'

Did Europeans (unintentionally, of course) bring ants to Hawaii? Or did Polynesians do so long before Captain Cook?

None of the Hawaiian words for 'ant' have an etymology even though words for 'ant' are reconstructible at the Proto-Oceanic and higher levels. I suppose they are unique to Hawaiian.

I'm reminded of the native word for horse, lio. Horses definitely weren't here before contact with Europeans? So how did they get a native name? I forgot this etymology at wehewehe.org:

a shortening of ʻīlio, formerly a generic name for quadrupeds

I prefer that proposal to one identifying the word as an extended usage of lio 'tight, taut, as a rope, or of hair or horse's ears pulled back tightly'.