07.2.9.23:59: LIP SOUNDS LIGHT (PART 8)

Clauson's interpretation of the second chapter of Homophones (see the end of part 7) sounded promising at first. It explained the initial variation among transcriptions of members of 'homophone' groups: e.g. (using Nishida's [1964: 82-83] numbering of the groups):

4a: dwi, ywi

17: dwï, bwi

21: l-wi, b-wi

27: dwi, wwi, bwe

What mattered in, say, group 21, was not l- or b-, but the shared wi - the 'kernel', as David Boxenhorn would call it.

There are two problems with Clauson's hypothesis.

First, why are the only complex transcribed initials b(-)w-, d(-)w-, l-w-, and yw-? Why are there no instances of other initials like m(-)w-, n(-)w-, (d)z(-)w-, j(-)w-, g(-)w-, or ng(-)w- (assuming that only letters for voiced consonants could precede w in the transcriptions)?  Why are there no combinations of voiceless initials followed by -w-: p(h)(-)w-, t(h)(-)w-, (t)s(h)(-)w-, ch(h)-, k(h)-? (The transcription wh- could have represented [hw] or [xw].)  Middle Chinese, Tibetan, and more relevantly, Tangut's close relatives Mawo Qiang (Sun 1981) and rGyalrong (Jacques 2003) all permit many consonants to combine with -w- (examples added 07.2.10.1:44):

Middle Chinese w-clusters:

Dentals	Alveolars	Retroflexes 1	Retroflexes 2	Palatals	Velars	Glottals
tw-	tsw-	Tw-	Chw-	chw-	kw-	'w-
thw-	tshw-	Thw-	Chhw-	chhw-	khw-
dw-	dzw-	Dw-	Jw-	jw-	gw-
nw-				ñw-	ngw-
	sw-		Shw-	shw-	xw-
	zw-			zhw-	Gw-
lw-				yw-

(-w- could be preceded by any MC initials except for labials and the rare initials N- [retroflex n] and Zh- [retroflex zh].  It is not clear whether MC labials were followed by -w- or not. If there was a -w- after labials, it was nonphonemic.)

Written Tibetan w-clusters (Jacques 2003: 95-96 or this PDF; excluding marginal clusters):

Dentals	Alveolars	Palatals	Velars	Glottals
	rtsw- (but not tsw-!)
	tshw-		khw-
dw-			g(r)w-
		ñw-
		shw-		hw-
		zhw-
lw-, rw-

(Once again, there are no labial-w clusters.)

Mawo Qiang w-clusters (Sun 1981: 34-35):

Labials	Dentals/Alveolars	Postalveolars	Retroflexes	Velars	Uvulars
	tw-	(x)chw-	xChw- (but not Chw-)		qw-
			GJw- (but not Jw-)
mw-	nw-			ngw-
	sw-		Shw-
	(G, R)zw-		Zhw-
	lw-, rw-

Japhug rGyalrong w-clusters (Jacques 2003: 58, 62-63):

Dentals/Alveolars	Palatals	Velars	Uvulars
		kw-	shqw-
		khw-
			Xw-
zw-
lw-, rw-	yw-

Zbu rGyalrong also has qw-, qhw- (2003: 309).  Proto-rGyalrong had even more clusters:
tw-, thw-, (t)sw-, ndzw-, kw-, (ng)gw- but no labial-w clusters (2003: 331-332).

Tangut did not have to follow this trend, but the limited number of possible w-sequences in transcriptions for tangraphs within this chapter hints at another explanation.

Second, Tangut apparently did have other consonant-w clusters after all, but they were assigned to other chapters of Homophones. Here are the other complex-w- initial Tibetan transcriptions I found in Nishida (1964):

Chapter I. Labials: none

Chapter III. Dentals: Hdwi, btwi

Chapter IV. Retroflexes: none

Chapter V. Velars: rngwa, bkhwe, rngwi

Chapter VI. Alveolars: bswoH, Hdzwu, Hdzwui (sic), tshwi

Chapter VII. Palatals: none

Chapter VIII. Laryngeals: ywe, Hrgwe, d-wir, d-woH

Chapter IX. Liquids: none

(Gong reconstructs Cw- clusters for syllables found throughout Homophones*, but it would be circular to cite his reconstruction as evidence.)

If Chapter II really contained all Tangut syllables with initial Cw-,  why were these Cw-syllables scattered among other chapters?

Next: My answer to that question.

*07.2.10.2:02: Guillaume Jacques' alphabetical index to Gong's (1997) reconstructions contains the following w-clusters:

The only Tangut initials that did not precede -w- in Gong's reconstruction were all labials except p- (ph-, b-, m-, w-) and r-.  Perhaps pre-Tangut rw- > Tangut w- (cf. the Elmer Fudd pronunciation of rabbbit as wabbit).

07.2.8.23:54: CHINESE EYES AND EARS (what does this refer to?*)

Yes, I can still write about topics other than Tangut. It's just that I only have so much time to post, and I have to be selective.

A few parts of this article about Chinese calligraphy got to me (emphasis mine).

Today there are still professional calligrapher artists and it is a very popular amateur pastime. Maybe its popularity has something to do with the palaver that Chinese speakers need to go through to use a computer. They have to spell out what they want to say phonetically, using the Roman alphabet keyboard and a programme that will translate that into Chinese characters. Even then Chinese is so sensitive to tone that a choice of four translations may be offered.

It looks like the reporter heard that 'Chinese' (actually, Mandarin) has four tones and concluded that there is one character per syllable/tone combination: e.g., if I type "shi", I get four choices, each with a different tone (that I indicate with a number):

shi1 失 'lose'

shi2 十 'ten'

shi3 使 'send / use / cause'

shi4 市 'city'

But in fact I have many more choices.  Even if I specify, for example, shi with tone 1 only, here's what I get using Windows XP's Chinese (Taiwan) input method:

失施師詩濕溼獅屍蝨噓虱尸蓍葹鳲邿湤絁鰤溮鶳箷褷襹螄瑡籭蒒釶 鍦褜魳鯴 (33 non-simplified characters)

师湿狮诗酾鲺鳁 (7 simplified characters)

plus the phonetic symbol ㄕ for sh- (doubling as the symbol for the syllable shi; cf. 尸 shi 'corpse' above)

Now let me drag Tangut into this.  (It's inevitable!)  If there were a phonetic Tangut input method, I could type in we2 (we with tone 2 = rising tone) and get 18 choices (from the first homophone group of the labiodental chapter of Homophones) including

TT3901 TEACH

TT0645 CITY

TT0624 FIFTH HEAVENLY-STEM (see the full list here)

TT5661 PERSONAL-SIGN (looks like WAIST jyiw 1.45 [!?] plus the previous homophonous tangraph as phonetic)

TT0631 KUMISS (phonetic 干丨? + WATER)

TT0627 SLAVE (phonetic 干丨? + the phonetic element ya [!?])

(Are 干 and 干丨 always phonetic symbols?**)

TT3213 (kind of grass; GRASS + what looks like the left of SLAVE above)

TT5286 CLIFF (BORDER + HIGH)

TT4029 DONKEY (read on)

TT4527 SOIL, etc.

Not specifying the tone (1 or 2) would result in 31 different choices (18 we2 tangraphs + 13 we1 tangraphs). 

In everyday Chinese life a vocabulary of 3,000 characters will be enough to get by, but calligraphers call on up to 50,000 in a selection of scripts such as seal, official, running, regular and mad.

I wouldn't call 3,000 characters "a vocabulary" because that implies that characters equal words, whereas the reality is that many, if not most characters only represent parts of words.  In everyday Chinese life, people use more than 3,000 words, but they can write nearly all of those words with only 3,000 characters.

I'm not sure whether the 50,000 figure includes different versions of a character in different calligraphic styles (as if one counted print and cursive a as two 'different' letters) or is one of those attempts to enumerate all the Chinese characters that ever were, regardless of calligraphic style.

He showed me 40 versions of the character for donkey, with its circular component referring to the millstone powered by a donkey in Ancient China.

I have no idea what this is talking about.  Here are the complex and simplified versions of 'donkey':

驢 驴

Neither version contains any "circular component referring to [a] millstone".

Both versions have the semantic component 'horse' on the left (complex: 馬, simplified: 马).

The complex version contains the phonetic component 盧 'food vessel' (mostly used to write a homophonous surname).  In late Old Chinese, the surname 盧 la and 驢 lïa 'donkey' were nearly homophonous.  (驢 is not attested in early Old Chinese.)  They are still similar in Mandarin today: the surname 盧 is Lu and 驢 'donkey' is lü.

The simplified version reduces the phonetic component 盧 to 戶 which looks like the character 戶 for the unrelated word Md hu 'door' and vaguely resembles 石 Md shi 'stone'.  Maybe the reporter heard that 戶 'door' resembled 石 'stone' and thought that 驴 'donkey' actually contained  石 'stone'.

None of the eight versions of 'donkey' at the Taiwanese goverment dictionary of sinographic variation contain 石 'stone'.

The Tangut word for 'donkey' was

TT3585 + TT3561

lyị̈²we²

Its tangraphs contained nothing that implied 'donkey':

Tangraph 1:

left: unknown element (Nishida's radical 276) resembling PERSON:

Nishida's (1966: 492) list of tangraphs with this element includes FIGHT, HARM, the surname khyiy 1.36 (see below),  HOWEVER, OLD, HOLD/HIDE, AGAIN, DWELL,  a place name thəw (Grinstead: BUT/ALTHOUGH, Gong: dyiy 2.33), ABOLISH, MUSTARD, CONQUER (Grinstead: TURKS [!?]), CONQUER

right: TT1730 tsyï ALSO 1.30 (click here to see it) which almost rhymed with TT3585 lyị̈² 2.61

Tangraph 2 (from the first homophone group in the labiodental chapter of Homophones):

left: non-independent element EARTH

right: non-homophonous (!) surname khyiy 1.36 (containing the same left element as tangraph 1 above; click here to see it)

The first half of the word lyị̈²we² 'donkey' resembles 驢 Early Middle Chinese lïə, Late Middle Chinese lüö, and probable Tangut period NW Chn lü 'donkey', so it could be a Chinese borrowing.  The trouble is ... the tangraph (composed of plus  could also mean 'hare', so it did double duty for a Chinese loan and a homophonous native word for a different animal.

The second half of the word is obscure.  Perhaps the word lyï¹we² 'donkey'is a redundant compound combining Chinese and native words for 'donkey'.

*Korean 唐나귀 tangnagwi 'donkey' sounds vaguely like English donkey.  The 唐 Tang refers to China's Tang dynasty but in this context probably just means 'Chinese'. 나귀  nagwi is a native word for 'donkey'.  So 唐나귀 tangnagwi is literally 'Chinese donkey'.

"Eyes" refers to 나 na which sounds like Korean 나 na 'I'.  (The Korean word for 'eye' is 눈 nun which sounds like 'noon' - and looks like Korean 눈 nuun 'snow' which has a long vowel not indicated in the spelling.)

"Ears" refers to 귀 -gwi which sounds like Korean 귀 kwi 'ear'.  (Korean k becomes g after a vowel: na + kwi = nagwi.  This change is not reflected in spelling: ㄱ can stand for k or g depending on its environment.

**07.2.9.2:30: Sofronov (1968 II.292-293) has a list of 45 tangraphs with 干(丨) on the left.  Here is how they are distributed among the chapters of Homophones:

Their rhymes are also all over the map:

1.16, 1.20, 1.29, 1.32, 1.36, 1.47, 1.49, 1.56, 1.57, 1.69, 1.74, 1.77, 1.80, 1.92

2.3, 2.4, 2.7, 2.23, 2.25, 2.28, 2.33, 2.40, 2.55, 2.73, 2.78

And they have no unifying semantic link: e.g., what do the 干-graphs above (CITY, FIFTH-HEAVENLY-STEM, KUMISS, SLAVE) have in common with

TT0640 po 1.49 HOE

TT0642 nyiy 2.33 WIND

TT0643 pyịy 2.55 SCOOP

TT0647 pya 1.20 BUTTERFLY

TT0648 re^rw 2.78 REPENT

TT0651 nywiy 1.36 GULLET

TT0653 dyï 2.28 WITH (cognate to COMBINE below?)

TT0655 dyï 2.28 PROTECT

TT0656 thywï 2.28 COMBINE (cognate to WITH above?)

TT0659 syu 2.3 LIKE, etc.

(I have cited every tangraph between 0640-0663 which Grinstead has glossed other than TT0645 we 2.7 CITY which I've already mentioned.)

Thus 干(丨) has no consistent semantic or phonetic value (from a Tangut A perspective).

***07.2.9.2:44: Schuessler (2006: 367) reconstructs 驢 'donkey' in Old Chinese as ra. Although the graph is not attested in earlier OC,

Unger (Hao-ku 13, 1989) points out that the donkey must have been known in China before its first mention during the Han dynasty because 'mule' [Md] luó 騾 ... occurs already in [the pre-Han text] Lüshi chunqiu [Wikipedia].

騾 would be early OC roy 'mule' in my reconstruction, which would be hard to relate to a hypothetical early OC 驢 ra 'donkey'.  They have nothing in common beyond initial r-. I don't know of any other OC word families with -oy and -a words.

07.2.7.4:29: LIP SOUNDS LIGHT (PART 7)

I've been thinking of making a few minor changes to Gong's Tangut reconstruction (just as Gong himself started tinkering with Sofronov's reconstruction before releasing his own state-of-the-art reconstruction).  Here's one of those changes:

Like Li Fanwen before him, Gong reconstructed only a single bilabial (not labiodental!) initial w- for the syllables in the second chapter of Homophones. I found this w- difficult to reconcile with the Chinese and Tibetan transcription evidence:

- Why would Tangut w- be transcribed with an f-initial sinograph (^口縛)?  The diacritic suggests that the sinograph was not supposed to be read with initial f- (though Nishida reconstructed Tangut f- partly on the basis of this transcription).

- Why would Tangut w- be transcribed with wh-, ww-, bw- as well as w-?  wh- suggests a sound that is w-like but is a fricative like h-.  ww- and bw- suggest a sound that has more friction than the glide w-.

(The use of chapter II tangraphs to transcribe Sanskrit v-syllables does not tell us anything, since a language without a v- could approximate Sanskrit v- with w-.  Written Tibetan, Thai, and Burmese use w- for Indic v-.)

(Moreover, the title of the chapter is not necessarily meaningful.  If the Tangut lacked f- and v-, they could have placed their native w- into the 'labiodental' category in order to preserve the Chinese initial categories as closely as possible.)

I concluded that Gong's w- may have actually been a fricative v-.  Written Tibetan has no letter v, so wh-, ww-, and bw- may have been attempts to write a Tangut v-.

But what about b-w-, d-w-, dw-, l-w-, and yw-? Why would those Tibetan letter sequences be used to transcribe Tangut v- (or w-)?

In 1964, Sir Gerard Clauson wrote (p. 62; emphasis mine),

The only native Tangut sound which can reasonably be brought within the scope of Chapter II of the Homophones, with its seventy-eight different word forms [= syllables which may or may not have been independent words], is the bilabial w.  In [Classical] Tibetan less than half a dozen words begin with this sound [there are more now due to loanwords], but this initial was commoner in Nam (see Thomas, op. cit.*) and  may have been fairly common in Tangut.  Some of the seventy-eight word forms, but perhaps not very many, no doubt represent Chinese loan words, names, etc. with initial f-.  The only obviously possible explanation of the remainder is that in Tangut medial w was regarded as the principal sound in an initial consonant cluster of which it was the second member and that words containing it were included in this chapter.  This would of course be different from the practice in Tibetan where medial -w- is represented by a subscript and the consonant to which it is attached is regarded as the initial.

Next: The consequences of Clauson's solution.

*Thomas, FW.  1948. Nam, an Ancient Language of the Sino-Tibetan Borderland.  Publications of the Philological Society 14. London: Oxford University Press.  I would love to see this book because Clauson (1964: 73) suggested that Nam might be "pre-1038 Tangut" (i.e., Tangut before tangraphy):

... since the interpretation of these documents [in Nam and other lost, 'unknown' languages] is at present no more than an exercise in more or less inspired guessing, a great deal more progress will have to be made with the reconstruction of the meanings [of the words in those documents?] as well as the sounds of Tangut words before any such identification becomes plausible.  This, however, can be said, that the phonetic structures of Nam and of the language represented by the long text in British Museum MS.Or. 8212 (188) are very much like what the phonetic structure of [earlier] Tangut is likely to prove to have been, with a considerable apparatus of initial and final consonantal clusters and, at any rate prima facie, a fairly limited stock of vowels [an artifact of an Indic writing system?]

07.2.6.2:57: LIP SOUNDS LIGHT (PART 6)

The most notable thing about the transcription data for labiodental homophone group 1 is the presence of initial d- in the Tibetan material: dwi, dwiH. The initial cluster dw- is rare in Written Tibetan and as far as I know only occurs in three syllables: dwa, dwags, dwang. Therefore I don't believe that

- there was a Tibetan syllable spelled dwi but pronounced [wi]

- and that Tangut we (or wI or vI) was transcribed as Tibetan dwi [wi]

Is there any Tibetan dialect today in which the written cluster dw- is pronounced as [w]: i.e., a dialect in which d- in dwa, etc. is as silent as the k- in Eng knee?

In Goldstein's (2001: 547-548) dictionary of modern standard (i.e., Lhasa) Tibetan, dw- is pronounced as [t], not [w].

dw- is not the only Tibetan cluster that appears in the transcriptions of Tangut syllables in the labiodental chapter of Homophones. Here is a complete list of all initial transcription data in Nishida (1964: 82-83), organized by Nishida's reconstructed initials:

The Chinese transcription initials are from pre-Tangut northwestern Middle Chinese: e.g., PTNWMC nggw-, mv-, 'w-, yw- may have simply been w- or v- in Tangut-period NW Chinese. I believe that all of the Chinese transcription initials are now w- or v- in modern NW Chinese dialects (with the exception of fH- which has simplified to f-).

bw- does not occur in Written Tibetan to the best of my knowledge.

d-w-, b-w-, l-w- represent the un-Tibetan letter sequences da-wa-, ba-wa-, la-wa- without any subscripts (or syllabic division-indicating tsheg dots) rather than as dw-, bw-, lw- (da, ba, la plus subscript w). (The syllable sequences da.wa., ba.wa., and la.wa. would be divided by tsheg dots.)

Did Tibetans really hear w- (and perhaps also f-, v-, Mv-) and write it (them?) as d(-)w-, b(-)w-, l-w- with non-labial initial letters?

Next: Clauson's solution.

07.2.5.3:06: LIP SOUNDS LIGHT (PART 5): _INTER_LUDE

I don't have time to do a full post tonight, so I'll just post a couple of notes:

First, I left out one type of main tangraph-clarifier pattern which Nishida (1964: 21) noted. I'll call this the

TER

lude in

type since the main tangraph represents the second syllable of a trisyllabic word: e.g., Homophones

entry 24B78

jywiy²khwə¹naa²

translated as 半春菜 'half spring vegetable' in the Pearl. The tangraphs are read in the sequence

2 khwə¹

3 1 naa² jywiy²

i.e., from the bottom right to the top center and then down to the bottom left.

Second, I asked whether the syllables in group 1 of Homophones' labiodental chapter could be reconstructed as vI (Nishida 1964: 85).

Sofronov, Li Fanwen, and Gong Hwang-cherng's answer would be 'no'. Here's how they reconstructed the pronunciation of

TT3901 TEACH 2.7

the first tangraph in group 1. Everyone agrees on the rhyme (rising tone, rhyme 7) but only Sofronov and Gong's reconstructions are identical:

Note that Li Fanwen has abandoned his 1986 reconstruction in favor of Gong's. I cite it to show the diversity of existing reconstructions.

Li Fanwen reconstructed all the other syllables in group 1 as (near-)homophones of TEACH. (Recall that some of them had the level tone instead of the rising tone. He reconstructed those syllables as wI 1.8 with the same segments w and I.) Similarly, Nishida reconstructed all group 1 syllables as either vI 2.7 or vI 1.8. I assume that Sofronov and Gong also reconstructed identical readings (except for tone) in this group. (I don't have time to check 58 [< 29 x 2] readings tonight. Li and Nishida's reconstructions can be easily accessed at a glance, whereas Sofronov's and Gong's are scattered.)

Which reconstruction, if any, is correct? Here's the foreign transcription evidence that Nishida collected for group 1:

Chinese transcriptions: ^o嵬, 外

Tibetan transcriptions: dwi, dwiH, whï

Sanskrit transcription value: none known

Unfortunately, Nishida did not specify which transcriptions were used for which tangraph (probably due to a lack of space).

Notes:

- the only evidence we have for Tangut-period northwestern Chinese is in Tangut, so we risk circularity by citing reconstructions and must turn to non-Tangut based sources: i.e., pre-Tangut and post-Tangut Chinese dialects of the region

- in northwestern Late Middle Chinese prior to the Tangut period, 嵬 (no circle) and 外 were nggwi and nggwai (the sources of Sino-Japanese nggwi, nggwai > modern gi and gai) but in modern northwestern dialects, 外 is wE or vE (Coblin 1994: 186) and 嵬 is probably something like wey or vey, judging from its near-homophone 危 (Coblin 1994: 216)

- pre-Tangut Tibetan transcriptions of 外 (Hgwe [nggwe?], gwe) imply that -ai had monophthongized to -E (Coblin reconstructed -Ei) even before the rise of the Tangut; the alternate transcription HgoHi [nggoy?] implies a different shift of wa > o: nggway > nggoy

- it's not clear to me when the -i of 嵬 broke to -ey; maybe it occurred during/by the Tangut period, explaining why ^o嵬 (-ey?) and 外 (-E[y]?) were used to transcribe the same Tangut rhyme (2.7; rhyme 1.8 tangraphs were only transcribed with ^o嵬)

- the function of the superscript circle in ^o嵬 is uncertain; presumably it means 'pronounce like 嵬, but not quite like 嵬'

- wh- is not a normal Tibetan initial cluster

- the sound value of the letter transcribed as ï (gi-gu inversé; gi-gu phyir-log) is uncertain.

So how do you think group 1 was pronounced?

Next: What about the other labiodental groups?

*Li Fanwen reconstructs rhyme 2.7 elsewhere in his study of Homophones as -IUI (1986: 188). I don't know why he didn't reconstruct TEACH and its (near-)homophones as wIUI. Perhaps he thought the rhyme -IUI was simplified to -I after w.

[07.2.7.3:57: I misread Li.  I think -IUI is really -I -UI - the rhyme without and with a preceding medial -U-.]

Strangely, there is no rhyme with simple -i in Li's rhyme table (1986: 188-89), though he did reconstruct long -ii for rhymes 1.30 and 2.28 (equivalent to Nishida's -ïH, Sofronov's -I, and Gong's -yï).

07.2.4.5:11: LIP SOUNDS LIGHT (PART 4)

When I color-coded the first page of the labiodental section of Homophones in part 3, I regretted not color-coding the clarifier characters written beneath the main characters. The color-coded graphic erroneously indicates that the syllables represented by the main characters and their clarifiers had the same tones. This is not always the case. For example, the syllable represented by the first labiodental tangraph in group 1 (09B21) had a rising tone but one of its clarifiers had a level tone:

Main tangraph: TT3901 教 TEACH we 2.7

Clarifier 1 (bottom right): TT3924 教 TEACH tshow 1.54 (1 = level tone, not 2 = rising tone; should have been marked with grey)

Clarifier 2 (bottom left): TT3589 指數 INDEX? lyïy 2.37

(In Li Fanwen [1986: 672], the second clarifier is miswritten as

TT3584 SMALL tsəy 1.40.)

All three tangraphs put together represent a trisyllabic word we²tshow¹lyïy² 'instruct'.

If English were written in tangraphy, the character for the syllable in- of instruction would be found in the glottal section (since it has initial glottal stop) with clarifiers for -struc- and -tion (read from right to left):

IN

TION STRUC

Most main tangraphs have only one clarifier. This clarifier can either be on the bottom left or the bottom right. Its position indicates whether the clarifier is to be read before or after the main tangraph:

CLARIFIER ON BOTTOM RIGHT: the second labiodental tangraph in group 1 (09B22):

_____

Main tangraph: TT0645 城 CITY we 2.7

Clarifier: TT4793 牆 WALL dzywï 1.69

These two tangraphs form the word dzywï¹we² 城壁 'city wall', 城堡 'castle'. The clarifier is read before the main tangraph above it.

(The translations are Li Fanwen's [1986: 240]. I was initially inclined to translate it as 'walled city', since Tangut noun1-noun2 sequences are modifier-modified sequences: 'wall' describes the 'city', just as 'dharma' in

tsyii^r1baa^r1 法鼓 'dharma drum' [Pearl 212]

specifies the type of drum [Nishida 1966: 266, 568]. However, perhaps this is a noun-adjective sequence: Tangut 'wall city' = Eng 'urban wall'.)

CLARIFIER ON BOTTOM LEFT: the first labiodental tangraph in group 2 (09B22):

Main tangraph: TT2393 塞 BLOCK we 1.8

Clarifier: TT2699 滿 FULL sə 1.27

These two tangraphs form the word (phrase?) 塞滿 we¹ sə¹ 'be fully stuffed'. The clarifier is read after the main tangraph above it.

Not all clarifiers form words or phrases with the head tangraph. Some are explanatory notes: e.g.,

Main tangraph: TT4962 we 2.7 (a surname)

Clarifier 1 (bottom right): TT0029 myï^r 1.86 (< earlier r-myï? cf. rGyalrong tU rme 'person' [Jacques 2003: 13], Written Tibetan mi 'person')

Clarifier 2 (bottom left): TT3890 SURNAME/CLAN mə 2.25

Li Fanwen (1986: 241) translated the clarifiers myï^r1mə² (lit. 'person-surname') as a word meaning 族姓 'clan surname'. There is no trisyllabic word we²myï^r1mə² - the sequence is to be interpreted as

WE

NAME SUR

i.e., as 'We - a surname'.

Nishida (1966: 21) lists other two-tangraph clarifier sequences which do not form words with the main tangraph: 'personal name', 'place name', 'Sanskrit', 助語 'help word' (i.e., a grammatical morpheme), 'mantra', and the puzzling (to me)

NOT GO

There are also two-clarifier sequences that serve as phonetic-graphic fanqie explanations: e.g.,

initial of clarifier 1 + rhyme of clarifier 2 = reading of main tangraph

part of clarifier 1 + part of clarifier 2 = main tangraph

thya 2.17 = th(yi 1.11) + ya 2.17

Main tangraph: TT0504 thya 2.17 (phonetic symbol for Skt tya [sic!; for thya?; Grinstead 1972: 196]; 反 MOUTH + phonetic symbol ya; similar to many phonetic symbol sinographs which consist of 口 mouth + phonetic: e.g., 咖啡 Md kafei 'coffee' ['mouth' is arguably even semantic in that case])

Clarifier 1 (bottom right): TT0510 DRINK thyi 1.11 (反 MOUTH is semantic; according to Tangraphic Sea, the right side is from the slightly different-looking bottom left of

TT1941 尋找 SEARCH bẹ 2.58! [!? why not, say, LIQUID?])

Clarifier 2 (bottom left): TT2156 ya 2.17 (phonetic symbol; cf. Chn 羊 [pron. yang in Md])

To review the different types of clarifiers, here's what the [fi] group in an English version of Homophones might look like:

FEA

ble si

PHOE

nix

FY

__taf

FEE

name sur

(OK, it's actually an Irish surname, but it has been Anglicized.)

The syllable fi in a foreign name (Mufi comes to mind) might be written as a fanqie character to visually distinguish it from [fi] in foreign words.

I selected [fi] for my example because Nishida (1966: 85) reconstructed the syllables of Homophones labiodental group 1 as vI, and I can't think of many English words with [vI]: e.g., video [vIdiow]. Moreover, [vI] ends in [I] which cannot appear at the end of an English monosyllabic word, and I wanted a monosyllabic example (which ended up being the surname Fee above).

Next: Was group 1 really vI?