Home GSR 130 AND 128

GSR 851a 役 from my last three posts looks like a semantophonetic compound of 彳 'to go' and a phonetic GSR 130a 殳, but 殳 is in fact a semantic component 'baton' (Karlgren 1957: 226), 'a kind of lance' (Schuessler 1987: 563).

The standard Mandarin reading of 殳 is shū with a high level tone normally pointing to a *voiceless initial. But other evidence points to a *voiced initial. GSR 128s 殊 'to cut off' > 'very', a homophone of 殳, transcribes Sanskrit ju in 文殊 for Mañju(śrī). And 殊 is also now shū in standard Mandarin. Why aren't 殳 and 殊 ˟shú with a high rising tone reflecting a *voiced initial?

Here's how I reconstruct the history of 殳 and 殊:

Scenario A: Primary *-d-

*CIdo > *CIduo > *duo > *dʑuo*dʑu > *ʑu > shū

Scenario B: Secondary *d-

*NITo > *NITuo > *NTuo > *nduo > *dʑuo*dʑu > *ʑu > shū

*N- is an unknown nasal. If Old Chinese was like Pyu, it had two possible nasal initials in presyllables: *n- and *m- (but probably not *ŋ-, unless ṅraḥ /ŋ.raH/ in PYU 20 is not an isolated oddity).

*T is an unknown dental stop: *t, *tʰ, or *d.

I can posit two parallel scenarios for 投 'to throw':

Scenario A: Primary *d-

*do > *dou > *du > *dəw > tóu

Scenario B: Secondary *d-

*NTo > *ndo > *do > *dou*du > *dəw > tóu

Schuessler (2007: 500) links 投 to Written Tibetan Hdor-ba 'to throw away' and gtor-ba 'to throw', but I would expect Written Tibetan -r to correspond to Old Chinese *-r, not zero.

殊 'to cut off' is cognate to 誅 'to punish, kill, reprove'. I assume both 殊 and 誅 had unaspirated or voiceless-initial roots, as there is no evidence for *tʰ or *d in 誅:

*RIto*RItuo > *Rtuo > *truo > *ʈuo > *ʈu > zhū [tʂu]

*R- might be *r- or *l-. *R- is so common in Old Chinese that I suspect it cannot simply be *r-. Written Tibetan has preinitial l- as well as r-, so Old Chinese may also have had preinitial *l-. EDACHI AGAIN: WHAT COUNTS AS OLD JAPANESE?

Continuing from my previous entry ...

岩波古語辞典 Iwanami kogo jiten (The Iwanami Dictionary of Old Words, 1990) gives an example of Old Japanese 役 edachi 'being forced to fight or work for the government' from 古事記 Kojiki (Record of Ancient Matters, 712):


Tsutsumi ike ni edachite, Kudara no ike wo tsukuriki.

'[They] were put to work on dikes and ponds, [and they] made the Pond of Paekche.'

Here is the context from 倉野憲司 Kurano Kenji's (1991: 145) reading of the Kojiki¹:


Mata Shiragibito maiwatarikitsu. Koko wo mochite Takeuchi no sukune no mikoto hikiite, tsutsumi ike ni edachite, Kudara no ike wo tsukuriki.

'Again Shilla people came over [to Japan]. Therefore Takeuchi no sukune no mikoto led them, had them put to work on dikes and ponds, [and they] made the Pond of Paekche.'

I fear that someone might see that entry and conclude that edachi is an Old Japanese word.

Why "fear"? Notice I wrote "reading" and not "edition". Kurano's (1991: 276) edition of the Kojiki - the text upon which his reading is based - doesn't have a single hiragana, since of course hiragana did not yet exist in 712:


It has punctuation marks that almost certainly weren't in the original text. I don't know what this passage looks like in the oldest surviving manuscript (the Shinpuku Temple manuscript from 1371-72), but you can see there is no punctuation in this image of a page from that manuscript. The punctuation is hardly the biggest problem, though.

Let's look at another reading of the Kojiki by 武田祐吉 Takeda Yūkichi (1977: 137)²:


Mata Shiraki hito maiwatarikitsu. Koko wo mochite Takeuchi no sukune no mikoto, hikiite, watari no tsutsumi no ike to shite, Kudara no ike wo tsukuriki.

'Again Shilla people came over [to Japan]. Therefore Takeuchi no sukune no mikoto led them, [and they] made the Pond of Paekche as a pond of the dike of the people who crossed over [i.e., from the Korean peninsula].'

(6.29.22:12: 'pond of the dike' makes no sense to me. The original text has 堤池 <DIKE POND>.)

It has no form of the verb edachi or even the character 役 (which Takeda reads as 渡 watari '[person who] crossed over').

Those two readings are not the only possibilities. Jidaibetsu kokugo daijiten (1967: 141) mentions two more readings of 爲役 or 役:

and proposes a third:

Which of these, if any, is right? There is no evidence within the original text to know. All the readings - the 讀み下し文 yomikudashibun - are translations into a stylized Japanese that is archaic but may not be identical to Old Japanese. That's why I romanize the readings in modern pronunciation.

The key word is phonograms. Unless an Old Japanese word is attested in phonograms, its phonetic value is unknown. All the e-readings of 役 are simply educated guesses. We don't know how yomikudashi worked in 712. 役 might even have been read as something like Go-on wiyaku!

6.29.22:20: The bottom line for me is that only Old Japanese words in phonograms can be cited in phonetic transcription. Old Japanese words in semantograms should be cited without phonetic transcription: e.g.,役 as <SERVICE>, not educated guesses like edachi, etachi, etatase, etate, etashi, etc.

古事記をそのまま読む Reading the Kojiki as It Is has more on the problem of interpreting 役 (or 渡 in the manuscript that it reproduces; Takeda's reading seems to be based on a manuscript with 渡): e.g.,

しかし、「役之堤池」は、全く不可解な構文である。まず、"之"が「の」を表すとした場合、「役の堤と池」は意味をなさない。 「役」を動詞とした場合も、目的語「堤池」の前に"之"を挟むことは絶対にない。 「役」がもし正しいとすれば、考えられる唯一の可能性は「"堤池之役"の誤写」である。

However, 役之堤池 is an completely incomprehensible construction. First, if 之 represents no [a genitive marker], 'dikes and ponds of service' makes no sense. [I had considered that possibility.] Even if 役 is taken as a verb, 之 would absolutely not be between it and its object 堤池 'dikes and ponds'. If 役 is taken as correct, the only conceivable possibility is an erroneous copying of 堤池之役 'service of dikes and ponds'.

6.29.21:50: Added English translations of the readings and derivations of the Jidaibetsu readings.

¹6.29.22:23: I have converted Kurano's hiragana back into the original kanji whenever possible to facilitate comparison with the original all-kanji text.

²6.29.22:24: I have converted Takeda's hiragana and postwar simplified kanji back into the original kanji whenever possible to facilitate comparison with the original all-kanji text. EDACHI

Continuing from yesterday's post about the unusual Japanese name 役小角 En no Ozunu:

The English Wiktionary lists edachi as a kun (native Japanese) reading of the Chinese character 役. (時代別国語大辞典 Jidaibetsu kokugo daijiten [The Great Dictionary of the National Language Categorized by Era] favors etachi.) I think edachi/etachi is not fully native. I regard it as a Chinese-Japanese hybrid *ye-(nV-)tat- 'to be forced to fight or work for the government'. The verb may be identical or similar in structure to modern 役に立つ yaku ni tatsu ~ 役立つ yakudatsu 'to be useful', lit. 'role DAT/LOC stand'. Yaku is an even earlier borrowing of Chinese 役 than e.

Before I look more into edachi/etachi, here's my take on the history of its first morpheme 役:

  1. The earliest reconstructible form for 役 is *CI-waj-k 'to do service' (Schuessler 2007: 568 reconstructs 役 *wai-k from 爲 *wai 'to do' = my *CI-waj.) *CI- may have been a causative prefix *SI-. (Cf. Baxter and Sagart's [2014: 56 ] causative *s-prefix.) The unknown high vowel *I is needed to account for the later vocalism (see the appendix). *-w- could be *-ɢʷ- (after Baxter and Sagart 2014), but I prefer to avoid exotic solutions if I can. See below for hard evidence for *-w- (or at least labiality).

  2. Fusion of *-aj- into *-e-: *Ciwajk > *Ciwek

  3. Vowel harmony-driven warping: *-e- breaks to *-ie- after a high vowel: *CIwek > *CIwiek

  4. Presyllable loss: *CIwiek > *wiek. Early Sino-Vietnamese việc was borrowed at this stage.

  5. *wi-fusion: *wiek > *ɥek

  6. *e > *a in palatal environments in some southern dialects (details unclear): ek > ak. Go-on yaku was borrowed from such a dialect before the 7th century. The earliest attestation of the Go-on reading that I know of is wiyau (sic; error for †wiyaku¹) in Ruiju myōgishō (c. 1100). It is remarkable that an un-Japanese [ɥ] or [jw] survived in spelling if not in pronunciation centuries after ak was borrowed. There is no trace of such an initial in modern Go-on yaku.

  7. -simpification: *ɥek > *jek
  8. Fronting of coda after a palatal vowel: *jek > *jejk or even *jec (if Hashimoto is right about Middle Chinese final palatals). Kan-on eki < 7th c. Kan-on *yeki was borrowed at this stage. But note that Sino-Korean 역 yŏk which probably slightly postdates Kan-on eki has nothing pointing toward a palatal component of the coda. The Sino-Korean reading isn't 옉 ˟yek < ˟yəyk < ˟(y)eyk. Maybe the coda had no palatal component in the source dialect of Sino-Korean. Or perhaps there was a phonotactic constraint against ˟-eyk in Old Korean. I don't know of any native Korean root with 옉 yek < yəyk < (y)eyk.

  9. *e-raising in some dialects: *jek > *jik. Sino-Vietnamese dịch < *jic was borrowed at this stage.

  10. Coda loss in some dialects: *jik > *jiʔ > standard Mandarin yi.

Back to edachi and the unusual name 役 En: Most Old Japanese speakers had no contact with Chinese speakers. The average speaker had few Chinese borrowings in their vocabulary. The elite, on the other hand, was more familiar with Chinese, and elite pronunciations of Chinese words may have been on an continuum from native speaker-like to heavily assimilated (i.e., Japanized). So the Kan-on reading (singular) of 役 was really a set of readings in En no Ozunu's time (the 7th century):

*ye is (questionably) attested in Old Japanese as a monosyllabic word 'corvee'. (But I write it with an asterisk because I reconstructed the *y-. The word is only known through the reading tradition²; there is no phonogram spelling pointing to *y-.)

The name En may have originated as *yek whose coda assimilated to the nasal of the following genitive marker (cf. the Korean rule /k n/ > [ŋ n]).

*yek nə wonduno > *yeŋ nə wonduno > En no Ozunu (with a pseudoarchaic -nu based on an erroneous reading of a phonogram for Old Japanese no).

¹19.6.29.21:19: wiyaku is the Ruiju myōgishō Go-on reading of 疫 'epidemic', a homophone of 役 in Chinese. The graph 疫 is a combination of 疒 'disease' (semantic) and 役 (abbreviated phonetic, itself a semantic compound of 彳 'to go' and 殳 'baton, beat' [Karlgren 1957: 226]). The Kan-on reading is eki, and the word was Japanized as *ye (now pronounced e), (questionably) attested in

according to Jidaibetsu kokugo daijiten (1967: 140) and Iwanami kogo jiten (1990: 201).

But I don't know for sure how 疫 was originally intended to be read in those texts. In fact, 倉野憲司 Kurano Kenji's (1991: 255) edition of Kojiki has 伇 (an archaic variant of 役, not 疫) which appears as 役 eyami (! < *ye-yami, a hybrid of Chinese 'epidemic' and native Japanese 'illness') in his 讀み下し文 yomikudashibun on p. 101. 武田祐吉 Takeda Yūkichi's (1977: 96) yomikudashibun of Kojiki has 伇 which he reads as e (< *ye).

²*ye appears in Man'yōshū 3847 in the semantogram combination 課役 <IMPOSE SERVICE> which has been read as edachi (< *yendati), etsuki (< *yentukɨ), and mitsuki (< *mitukɨ) 'tax' (Ōno et al. 1990: 205). (6.28.21:57: All three possibilities fit the meter.)

Appendix: Evidence for labiality in 役 Karlgren (1957: 226) reconstructed Old Chinese 役 as *di̯ĕk without any labial segment. Thirty years later, Schuessler (1987: 743) reconstructed the word as ?*ljik without any labial segment. Standard Mandarin yi and Cantonese jik have no labial segment. However, both internal and external evidence point to a labial segment.

1. Internal evidence

1.1. Mandarin:

(Do any Jin varieties have a labial vowel in this morpheme? Xiaoxuetang only lists Taiyuan 太原 ieʔ.)

1.2. Most Wu varieties at Xiaoxuetang have labial vowels: y, u, or o; 莊村 Zhuangcun has ʯʔ [ʐ̩ʷʔ]

1.3. All Xiang varieties at Xiaoxuetang have y.

1.4. Some Gan varieties at Xiaoxuetang have y or u; 平江 Pingjiang has ʯɤt [ʐ̩ʷɤt].

1.5. A few Hakka varieties at Xiaoxuetang have labial vowels:

1.6. Some Yue varieties at Xiaoxuetang have v-, y, or u.

1.7. Some Pinghua varieties at Xiaoxuetang have v-, ʋ-, y, u, or o.

1.8. Min varieties (list not exhaustive):

1.8.1. Southern: 揭陽 Jieyang uek

1.8.2. Pu-Xian: yʔ in both 莆田 Putian and 仙游 Xianyou

1.8.3. Eastern: 福安 Fuan peik with p-!

1.8.4. Northern: 石陂 Shibei ɦy (with level tone!)

Are the initial consonants of Fuan and Shibei evidence for a proto-obstruent like Baxter and Sagart *ɢʷ-?

1.8.5. Central: 明溪 Mingxi y (with departing tone!)

1.8.6. Other: 隆都 Longdu uɐk (with upper register!) and 將樂 Jiangle y

1.9. Some of Xiaoxuetang's unclassified varieties also have labial segments:

2. External evidence

2.1. Early Sino-Vietnamese việc and Muong (variety unidentified) [wiək] (Pulleyblank 1994: 83, cited by Schuessler 2007: 563)

2.2. Ruiju myōgishō (c. 1100) Go-on wiyau (sic; error for †wiyaku)

2.3. Borrowings in Tai: Saek viak D2L 'work', Siamese wiek³ (Maspero 1912: 73, cited by Schuessler 2007: 563; thai-language.com regards เวียก wiak as Isan - i.e., not standard Siamese - and the example implies it means 'work') Added the forms from Schuessler (2007: 563). EN NO OZUNU

役小角 En no Ozunu, founder of 修驗道 Shugendō, was banished by the Japanese court 1320 years ago today. The spelling of his name is doubly interesting.

角 'horn' is normally read as tsuno. In a compound, I would expect ts- to voice to z-: -zuno. But I wouldn't expect a final -u. Iwanami kogo jiten (1990) says tsunu is an Edo period error for tsuno based on the misinterpretation of man'yōgana for -no as nu. So is Ozunu an Edo period misreading of 小角? (The genitive marker no between En and Ozunu is unwritten.) Or is Ozuno (another reading of 小角) a regularization of an original Ozunu reflecting a dialect in which *-o raised to -u? (Cf. forms like Hitachi Old Japanese yu [Kupchik 2011: 374] corresponding to Western Old Japanese and even modern standard Japanese yo 'night'.)  

役 (Wiktionary) is normally read as yaku or eki. Both of those readings are Chinese loans. 役 has never had a nasal-final reading in Chinese. So why is 役 read En in this name? If the name is native, it shouldn't end in -n since all Japanese words in the 8th century ended in vowels. I wonder how 役 was read when he was alive. ASADY

I was surprised to learn last night that بشار الأسد‎ Bashar al-Assad is pronounced [baʃˈʃaːr elˈʔasad] in Levantine Arabic with a gemihnate [ʃʃ] and a single [s]. Why isn't it written as Basshar al-Asad in English?

The Polish Wikipedia reflects the geminate [ʃʃ] and a single [s]: Baszszar al-Asad.

The Slovak Wikipedia even reflects the long [aː]: Baššár al-Asad. (But the Czech Wikipedia lacks the geminate: Bašár al-Asad.)

The Hungarian and Albanian Wikipedias have e for at least one short [a]: Bassár el-Aszad (but not ˟Eszed!) and Beshar el-Asad (but not ˟Esed!).

The Thai Wikipedia has บัชชาร อัลอะซัด  <ɓăjjāra ʔălaʔaḥzăɗa> whihch I assume is read as [bàtsaːn ʔan ʔasát]. Alas, I don't know of any entries for Assad in the Lao, Khmer, or Burmese Wikipedias.

The Tamil Wikipedia has another drastic localization: பசார் அல்-அசத் <pacār al-acat>  [pasaːr al asat]. (Tamil has no initial [b] or final [d].)

I don't know how typical those renderings are. I wish I had time to investigate how Arabic names are localized in various languages.

The title is from my attempt to Tangutize Assad's name as


1637 5994 4541 2682 4541 1693 0804

2ba1 1shar3 1a? 2lu3 1a? 1sa4 2dy4

using conventions originally developed for Sanskrit.

I am unaware of any reasoning for choosing either tone 1 or tone 2 for transcribing Sanskrit, so I have not taken tones into consideration when choosing transcription characters from the set used for Sanskrit as compiled by Arakawa (1997).

1. 1637: Sanskrit -a is generally Tangutized as -a4, though Sanskrit ba can be Tangutized with -a1. I should look into exceptional cases of a1-transcription.

2. 5994: In theory I could have transcribed [ʃʃaː] as shy sha, but I have chosen an English-like solution with just one fricative.

Tangut has no syllables ending in [r]. The -r of my Tangut notation indicates vowel retroflexion, not an [r]-coda.

Tangut has no word spacing. Perhaps modern Tangut would have had a dot here to separate foreign names.

3. 4541: Devised to write Sanskrit a. Sanskrit a after consonants is normally transcribed as -a4, so I suspect 4541 is 1a4. But 1a1 is also possible since Sanskrit ba can be transcribed as 1637 2ba1. I do not think 1a2 or 1a3 are likely, as neither is the known reading of any other tangraph. 1a4 is attested as the reading of 𗅹 2375 'east, tail'. 1a1 is also not attested as the reading of any other tangraph, but Sanskrit a was transcribed in Chinese as *1a1, and Tangut  -a1 can correspond to Sanskrit -a. Hence 1a1 is not impossible as a reading of 4541.

I do not know why Li (2008: 721) does not list a tone for 4541 which appears in the level tone (i.e., tone 1) section of Mixed Categories of the Tangraphic Sea.

4. 2682: Isolated Sanskrit consonants are usually Tangutized as -y syllables. However for some reason l is Tangutized as either 2682 ending in -u or 3284 𗥰 2la3. I have opted for 2682 2lu3 since it sounds like the Japanese solution for writing -l: -ru.

5. 4541: See above.

6. 1693: An example of an -a4 character used to Tangutize a Sanskrit syllable. Contrast with 1637  2ba1 for Sanskrit ba. Gong reconstructed -a4 as [ja], but there is no [j] in most Sanskrit syllables transcribed as -a4. (An obvious exception is 𘁂 5314 2a4 for Sanskrit ya.)

7. 0804: An example of an y-syllable used to Tangutize a Sanskrit consonant. Contrast with 2682 above which ends in -u rather than the usual -y. CIR: THE THREE-AXIS MODEL OF ORTHOGRAPHIC REFORM

Having recently finished reading Robbins Burling's Spellbound: Untangling English Spelling, I've been thinking about how to characterize different proposals for reforming English orthography. The CIR model has three axes:

  1. C: Continuity

  2. I: Internationality

  3. R: Regularity

I could describe a proposal in terms of these three features using this notation: [±lowercase letter of feature].

Continuity refers to whether a proposal incorporates an existing practice, either by leaving it alone or by expanding its domain.

To write all English [dʒ] as <j> is [+c] since some English [dʒ] are already written as <j>.

To write all English [ʃ] as <x> is [-c] since English [ʃ] is not written a <x>. (I am not counting foreign names like <Xi>.)

Continuity is of interest to both existing users of English (native and nonnative) and learners who would want to access literature in the prereform orthography.

Internationality refers to whether a proposal is compatible with non-English orthographic practices.

To write all English [i] as <i> instead of, say, <ee> is [+i] since <i> represents [i] in most Latin-alphabet orthographies.

To write all English [ʃ] as <s> is [-i] since no Latin-alphabet irorthographies have <s> for [ʃ] with the major exception of Hungarian. (Also, <s> is [ʂ] for southern Vietnamese speakers - not [ʃ], but close.)

Internationality is of interest to learners who would benefit from an orthography using conventions they are likely to already know.

Regularity refers to whether a proposal has one symbol (or symbol sequence) per sound.

To write all English [k] as <k> is [+r].

To write English [k] as <c> before nonfront vowels and <k> before front vowels is [-r]. (But still more regular than the current spellings of [k]!)

Regularity is of interest to learners who do not want to be burdened with irregularities.

Obviously those features are not really binary; there are degrees of CIR. I don't want to assign arbitrary numerical values, so maybe I could double plus and minus signs: e.g., to switch English to the Shavian alphabet would be [--i] ('doubleminus international'? - cf. Orwell's 'doubleplusgood') since no other language is written in that script.

Shavian in terms of all three features is [-c -i +r]:

The existing orthography is [+c -i -r]. It is irregular with many language-specific eccentricities.

A 'perfect' orthography that is [+c +i +r] seems impossible. To maintain continuity to some nontrivial degree, a new orthography would have to abandon internationality: e.g., reject international <i> for English-specific <ee> as the spelling of [i].

A [+c -i +r] orthography would require learning a lot of English-specific conventions, but those conventions would be consistent: e.g., <mee> and <eet> instead of <me> and <eat> (cf. <eel> and <feet> which would remain unchanged).

A [-c +i +r] approach would require all four of those [i]-example words to have new spellings: <mii>, <iit>, <iil>, <fiit>.

Maybe I should call regularity between two features, 'regularity' and 'monophoneticity/monophonemicity' (?). It is possible to have regularity without absolute one-to-one correspondences: e.g., [i] could be <ii> in closed syllables but <i> in open syllables: e.g.,< iit>, <iil>, <fiit> but <mi> (since there is no [mɪ] that would be written <mi> if <i> = [ɪ]). Another example of this type of 'split' (or environment-conscious) regularity is my proposal above for <c> and <k> which I regarded as [-r].

Lastly, the CIR terminology or something like it could be used to describe any script. Invented scripts would be [-c]. Adaptations of existing scripts could be regarded as [-c] if they bear little or no relation to a previous script for a language. The modern Turkish alphabet is

The Tangut script is

[-c -i -r] scripts like Tangut are the hardest to learn because they are sui generis.

¹6.26.17:26: I inserted "almost" because of ambiguous cases like gâvur [ɟaʋur] 'infidel' which in theory could also be read as ˟[ɟaːʋur] with a long vowel as well as a palatal consonant (cf. kâfir [caːfir] 'infidel') or ˟[gaːʋur] with a long vowel. Google Translate's TTS (?) 'knows' that gâvur and kâfir both have palatal consonants but that only kâfir has a long vowel.

I'm surprised that gâvur doesn't have a long vowel. It is borrowed from Persian گاور gāvur (before the u > o shift in modern standard Persian) which does have a long vowel. And kâfir (from Arabic via Persian) demonstrates that palatals can precede long vowels in Turkish.

I confess I was tempted to derive gāvur from kāfir, but there would be no reason for Persians to change k, f, and i to g, v, and r. The earlier form of gāvur is گبر gābr which is phonetically even further from kāfir - the second consonant b is a stop, not a fricative, and there is no second vowel. gābr is from Aramaic, not Arabic. THE BATTLE OF MANG YANG PASS

The Battle of Mang Yang Pass occurred sixty-five years ago today:

It was one of the bloodiest defeats of the French Union together with the Battle of Dien Bien Phu in 1954 and the Battle of Cao Bằng in 1950.


The ambush and destruction of GM 100 [Groupement Mobile No. 100] was considered the last significant battle of the First Indochina War. Three weeks later, on Jul. 20, 1954, a battlefield ceasefire was announced when the Geneva agreements were signed, and on Aug. 1, the armistice went into effect, sealing the end of the French Indochina and the partition of Vietnam along the 17th parallel. The last French troops left South Vietnam in April 1956, upon request from President Ngô Đình Diệm.

What kind of name is Mang Yang? Vietnamese syllables normally do not begin with Y-. Mang Yang is in Gia Lai Province which has many obviously non-Vietnamese names. The one I recognize is Pleiku which is un-Vietnamese in four ways:

  1. It begins with p-. ph- is permissible but not p- (because earlier Vietnamese *p- became b-).

  2. It begins with a consonant cluster containing -l-. All native Vietnamese *Cl-clusters became tr-. Wiktionary has a phonetic Vietnamese spelling pờ lây cu splitting the first syllable Plei in two. (Why is a huyền tone assigned to pờ?)

  3. The first syllable ends in -ei, a rhyme unknown in Vietnamese.

  4. The second syllable has a k- instead of c- for [k] before a back vowel. Vietnamese k- is normally written only before front vowels.

Wikipedia and Wiktionary derive Pleiku from Jarai Plơi Kơdưr, lit. 'village north/above'.

The Vietnamese Wikipedia says Mang Yang is Bahnar for cổng trời, lit. 'gate sky': i.e., 'sky gate'. But the dictionary of the Plei Bong-Mang Yang Bahnar dialect by the Bankers and Mơ (1979) has no words like mang 'gate' or yang 'sky'. I cannot find a word for 'gate' in the dictionary's English-Bahnar index, and the only word for 'sky' I can find using that index is plĕnh on p. 99. (There is supposed to be another word for 'sky' on p. 110, but I don't see one.) There is a yang 'spirits, nonhuman beings that affect humans' on p. 145. Perhaps that is the Yang of Mang Yang.

Tangut Yinchuan font copyright © Prof. 景永时 Jing Yongshi
Tangut character image fonts by Mojikyo.org
Tangut radical and Khitan fonts by Andrew West
Jurchen font by Jason Glavy
All other content copyright © 2002-2019 Amritavision