When looking at Andrew West's post about a Tangut hand mirror with the character


0645 2wuq1

which he translated as 祐 'to aid', it occurred to me that 2wuq1 sounds like 우 u, the Sino-Korean reading of 祐. (The Yale romanization of the reading is visually even closer - wu!)

That makes the Tangut word easier to learn. I try to take advantage of soundalikes whenever I can. But are the two forms related? I don't think so, because 2wuq1 is from Pre-Tangut *Sʌ-ʔwə/oH, whereas u is ultimately from Old Chinese *wəʔ(-s).

1.13.13:49: Commentary on the reconstructions


T0. The only remotely similar words I know of are Old Chinese Pa-type words (my ignorance of the rest of Sino-Tibetan is showing):

*Cɯ.P(r)a > *bɨa 'to help'

*Cɯ.P(r)a-ʔ > *bɨaʔ 'to help'

*Cɯ.P- may have fused into *b-: *N-p- > *m-p- > *m-b- > *b-. Another possibility is that *-P- was *-b-, and that *C- has left no trace: *Cɯ.ba > *Cɯ.bɨa > *bɨa.

輔 may be a *ʔ-suffixed variant of 扶.

The presyllabic and main vowels don't match.

There is no guarantee that Tangut -w- is from a lenited stop *P.

A medial *-r- cannot be ruled out; if it existed, it corresponds to nothing in Tangut.

T1. Pre-Tangut *S- conditions Tangut vowel tenseness that I indicate in my notation as -q.

T2. Pre-Tangut *-ʌ- conditions the grade of the Tangut syllable (-1). The phonetic value of -u1 was (partly) lower than [u]: e.g., [ou]. *-u (< *-ə or *-o) lowered to harmonize with the height of unaccented *-ʌ- which was later lost.

T3. I have projected Tangut [ʔw] (w- in my notation) back into pre-Tangut. But I suspect that at the pre-Tangut stage there was a sequence *-CVP- that was compressed into Tangut [ʔw]. Pre-Tangut *-ʌ- could have been in that sequence: e.g., *S(ʌ).Cʌ.PəH.

T4. The pre-Tangut vowel could be either or *o; both merged into -u1 (Jacques 2014: 206).

T5. Pre-Tangut *-H is a laryngeal that conditioned Tangut tone 2 which I write Arakawa-style at the beginning of my notation. *-H could correspond to Old Chinese *-ʔ-s. My assumption that Tangut tones originated Chinese-style from laryngeals could be wrong; they may preserve Proto-Sino-Tibetan tones or have some entirely different origin. But if Tangut and Chinese developed similar grade systems (possibly via contact), they might have developed tones in similar ways as well.

Old Chinese

C0. 祐 *wəʔ(-s) 'to assist' belongs to a large word family discussed at length in Schuessler (2007: 581-582). Schuessler reconstructs a Proto-Sino-Tibetan root *wəs. I don't know how he would account for the *-ʔ in the Old Chinese members of the family.

On the other hand, Matisoff (2003: 327, 591) relates 佑 (another spelling of 祐) to his Proto-Tibeto-Burman *grwak 'friend/assist'. The rhyme might work: Matisoff's Proto-Tibeto-Burman *a can be from Proto-Sino-Tibetan *ə. See Schuessler (2007: 31-32) on Tibeto-Burman -k corresponding to Old Chinese *-ʔ. I don't believe in 'Tibeto-Burman' except as a convenient term for 'non-Chinese Sino-Tibetan', so 'Tibeto-Burman' here is to be taken as the latter.

As for *gr-, see C1-C2 below.

C1. Baxter and Sagart reconstruct the 祐 word family with *[ɢ]ʷ-. The brackets indicate 'either *ɢʷ-, or something else that has the same Middle Chinese reflex as *ɢʷ-' (wording based on Baxter and Sagart 2014: 8): e.g., *N-qʷ- or *m-qʷ-. *ɢʷ- does look like Matisoff's Proto-Tibeto-Burman *g- (see C0 above). But I am suspicious of it - there is no Chinese-internal evidence that there was ever a stop in this word family. And Baxter and Sagart's system has no simple *w- which is what Schuessler and I reconstruct instead of *ɢʷ- in this word family.

C2. There could not have been an *-r- in this word because *wrəʔ-s (or Baxter and Sagart's *[ɢ]ʷrəʔ-s) would have become Middle Chinese †wiʰ (cf. 鮪 *wrəʔ / *[ɢ]ʷəʔ > *wiˀ 'sturgeon'²), not *wuʰ (> Sino-Korean 우 u).

¹Samuel E. Martin designed the Yale romanization of Korean to be typeable on a standard US keyboard, so it has no diacritics or nonbasic Latin letters. w distinguishes labial wu [u] from nonlabial u [ɯ] (= ŭ in the modified McCune-Reischauer romanization of Korean on this site).

²The Sino-Korean reading of 鮪 should be †위 wi, but in fact it is 유 yu, presumably by analogy with 유 yu, the reading of the far more common character 有 'to exist'. There would be few opportunities to use 鮪 in Korean; the Korean word for 'sturgeon' is 鐵甲상어 chhŏlgapsangŏ 'iron armor shark'.

The suffix -ngŏ 'fish' is from Middle Chinese 魚 *ŋɨə, but in hangul it is written as <-.ng Ø.ŏ> across two syllables, so it is not associated with 魚 since character readings always only occupy single hangul blocks.

상어 sangŏ 'shark' is from Middle Chinese 鯊魚 *ʂæ ŋɨə, though it cannot be written as 鯊魚 in Korean since its parts do not correspond to syllable blocks/Sino-Korean readings:

Sinographs (aligned with Middle Chinese)
Middle Chinese
Korean (transliterated)

Sino-Korean (transliterated) s
Sinographs (aligned with Sino-Korean)

The Sino-Korean readings of 鯊 and 魚 are 사 sa and 어 ŏ, so 鯊魚 is read as saŏ. I suspect that sangŏ is an old borrowing from spoken Middle Chinese, whereas saŏ is a literary Korean creation combining the isolated readings sa and ŏ (< ngŏ). A TR-OUBLING TR-ANSCRIPTION

蔡同榮 Chai Trong-rong passed away fifteen years ago today. At first his name might look Vietnamese because of its tr, a letter combination not used in romanizations of the other major East Asian languages. However:

- Chai is not a Vietnamese surname. It is not even a possible Sino-Vietnamese syllable.

- Vietnamese names are usually made up of Sino-Vietnamese elements, and rồng 'dragon' is not one of them; it is a loan from Late Old Chinese 龍 *roŋ, but it is not Sino-Vietnamese in the strict sense: i.e., it is not the reading of 龍 which is long, a much later loan which may postdate rồng by a millennium.

- The Sino-Vietnamese reading of 蔡同榮 is Thái Đồng Vinh which is quite different from Chai Trong-rong.

- Nothing in Chai's background - beginning with his childhood in Japanese-ruled colonial Taiwan - points to a Vietnamese connection. (At first I thought he might be a Vietnamese immigrant to Taiwan. But he is ethnically Taiwanese.)

The tr doesn't match anything in the other forms of the name listed at Wikipedia:

Mandarin (IPA: [tsʰaj˥˩ tʰʊŋ˧˥ ɻʊŋ˧˥]):

Pinyin: Cài Tóngróng

Wade-Giles: Tsài Tóngróng (sic; the correct Wade-Giles is Tsʻai⁴ T'ung²-jung²)

Tainan Taiwanese Hokkien (IPA: [tsʰwa˥˩ tʰɔŋ˧ ʔeŋ˨˦]):

Pe̍h-ōe-jī: Chhòa Tông-êng

Then it occurred to me that Trong has the same letters as torng, the Gwoyeu Romatzyh (GR) romanization of 同. The -r- represents a high rising tone, not a consonant [r]. Could Trong be a metathesis of torng? And was the reordering of o and r accidental or intentional?

Rong is the GR romanization of 榮, but Chai is not the GR romanization of 蔡 which is tsay with y signalling [j] preceding a high falling tone.

1.12.13:41: Later last night it occurred to me that someone unfamiliar with Chinese might have accidentally spread the r- of -rong to the preceding syllable: Tong-rong > Trong-rong. But why would Chai adopt someone else's error? A NEW INTERPRETATION OF 'OLD' IN THE SINO-JURCHEN VOCABULARY OF THE BUREAU OF INTERPRETERS

Two days ago I rediscovered this passage in Juha Janhunen's magnum opus Manchuria: An Ethnic HIstory (1996: 141; emphasis mine):

As a matter of fact, most of the differences conventionally assumed to exist between Jurchen and Manchu are illusory, being due to the poor compatibility of both the Jurchen and the Chinese script with the actual phonological shapes of Jurchen words.

It is reminiscent of what I wrote last week about "Manchu-like [Jurchen] forms distorted through the lens of the Chinese finite syllabary."

But I also wrote,

That is not to say that all apparent slight differences from Manchu in those transcriptions [of Jurchen] can be explained away as inescapable compromises. Some transcriptions do represent genuine dialectal variants [in Jurchen].

So one must avoid the temptation to see Manchu-like forms at all costs. One must always ask, does this Jurchen form differ from Manchu only because of the limitations of the Chinese script?

In the following case, I think the answer is 'yes'.

In the Sino-Jurchen vocabulary of the Bureau of Interpreters, the Jurchen for 舊橋 'old bridge' (#195) is listed as


Ming Mandarin *fo xufulun

The word for bridge - presumably hufulun or hufurun - doesn't look anything like Manchu doohan ~ doogan (< ?*daugan¹) 'bridge'. Kane read it as hufurun. I do not know why he decided the liquid was -r- rather than -l-. Ming Mandarin had no *r, so 倫 *lun could have represented a Jurchen lun or run.

In part 1 of my response to Kawasaki's article on Jurchen, I followed Kane and read the word for 'old' as fo. I thought fo and standard Manchu fe 'old' were different compressions of an earlier Jin Jurchen disyllabic

<pu (g)e> = pu(g)e (奧屯良弼餞飲碑 Aotun Liangbi picnic inscription 1)

But as I was editing the article for uploading, I realized that there was no Ming Mandarin *fə which would have been a perfect phonetic match for a Jurchen fe [fə]. Ming Mandarin *fo might have been the closest available phonetic match. That does not mean I think the Ming Jurchen dialect² in the Sino-Jurchen vocabulary of the Bureau of Interpreters was directly ancestral to standard Manchu. I simply think that the two varieties³ shared the same compression of pu(g)e.

¹1.11.1:23: Could *daugan be from a Khitanic language? It vaguely resembles Written Mongolian dughui 'narrow bridge', but the vowels rule out a direct borrowing from Mongolian. Hence it is not surprising that Manchu doohan ~ doogan  is not in Rozycki's Mongol Elements in Manchu (1983).

²1.11.0:26: Perhaps I should have just written "Ming Jurchen", as closer examination of the vocabulary may reveal a mix of dialects or a single dialect with a smattering of loans from other dialects.

³1.11.0:36: That is, the variety that is the source of this phrase fe hufu(r/l)un 'old bridge' and standard Manchu. I am hesitant to regard all of the Jurchen forms in the Bureau of Interpreters vocabulary as being from the same dialect. My suspicion is that the vocabulary is a collection of bits and pieces gathered from informants with diverse bakgrounds. But the problem awaits detailed examination. TODAY IN JURCHEN HISTORY

By coincidence, two major anniversaries today are exactly 23 years apart:

- the fall of the Northern Song capital of Bianjing (now Kaifeng) to the Jurchen:

Emperor Qinzong and his father, Emperor Huizong, were captured by the Jin army. The Northern Song dynasty came to an end.

- the assassination of Emperor Xizong in 1050


Emperor Xizong felt depressed by the loss of his sons that he developed an addiction to alcohol and started neglecting state affairs. He also became more violent and ruthless, and started killing people indiscriminately. One of his victims was Ambaghai, a Mongol chieftain and great-granduncle of Genghis Khan.

Emperor Xizong was overthrown and murdered by his chancellor, Digunai, and other court officials in a
coup d'état on 9 January 1150.

Xizong is of linguistic interest as the man attributed with the creation of the mysterious Jurchen small script whose fate may have been intertwined with his:

During the 1970s a number of gold and silver paiza with the same inscription, apparently in the small Khitan script, were unearthed in northern China. Aisin-Gioro has analysed the inscription on these paiza, and although the structure of the characters is identical to the Khitan small script she concludes that the script is not actually the Khitan small script but is in fact the otherwise unattested Jurchen small script. She argues that this small script was only used briefly during the last five years of the reign of its creator, Emperor Xizong, and when he was murdered in a coup d'état the small script fell out of use as it was less convenient to use than the earlier large script.

If Aisin Gioro Ulhicun is right, the only two surviving samples of the script are those that she has identified. I have written about the first twice before; I should finally get around to writing about the second six years later. THE JURCHEN SCRIPT: INNOVATION OR DERIVATION? (PART 1)

(Edited 1.10.0:41 before posting.)

川崎保 Kawasaki Tamotsu's 「渤 海」文字資料からみた女真文字の起源に関する一考察 ('An Observation Concerning the Origin of the Jurchen Script as Seen from Parhae Script Materials', 2014), a response to Alexander Vovin's "Did Wanyan Xiyin Invent the Jurchen Script?" (2012), contrasts two views of the Jurchen large script: 発明 hatsumei 'invention' (the ex Khitanis¹ hypothesis) and 発展 hatten 'development' (the ex Parhis² hypothesis).

To try to parallel how both terms begin with the root 発 hatsu- (hat- before t-) 'go out', I have loosely rendered 発展 hatten as 'derivation' in the title so that it has the same ending as invention.

Kawasaki first summarizes Vovin's  English-language paper in Japanese before presenting his own views.

Vovin read a Parhae stamped tile in Jurchen as

pe gorhon ni

'old thirteen GEN' = 'of Old Thirteen'

Kawasaki interpreted the first two characters as a single Jurchen character looking like Chinese 舍 'to set aside; lodging'.

I am not sure what to make of this stamp for several reasons. Here are the first two.

1. The first character on the stamp has 人 on the top rather than ス. There is no evidence that those two elements were interchangeable in the Jurchen large script, as no ス-graphs have 人-variants in Jin Qizong's dictionary (1984: 23-24). Nonetheless that does not refute Vovin's reading because 人 and ス could have been interchangeable in the earlier Parhae script. Alternately, the stamp may show an older Parhae form with 人 that was replaced by ス in Jurchen.

2. Vovin interprets

as Jurchen pe 'old'. At first this seems plausible given that (1) the Manchu word for 'old' is fe and (2)  Jin Jurchen p- corresponds to Ming Jurchen and Manchu f-. Parhae Jurchen predated Jin Jurchen and probably would also have had p-.

However, the word for 'old' is attested in the Jurchen large script as disyllabic

<pu (g)e> = pu(g)e (奧屯良弼餞飲碑 Aotun Liangbi picnic inscription 1)

and not as a monosyllabic pe. The contraction of pu(g)e to Manchu fe may have been a post-Jin innovation in some but not all varieties of Ming Jurchen. The Bureau of Translators vocabulary has disyllabic fuwe (transcribed as 弗厄 *fu ə and spelled in the Jurchen large script as in the Aotun Liangbi picnic inscription; #667) whereas the Bureau of Interpreters vocabulary has a monosyllabic transcription 佛 *fo, presumably for a form like fo in a Ming Jurchen dialect that compressed fuwe differently than the ancestor of standard Manchu. (o is labial like -uw- and mid like e [ə].)

It is unlikely that a monosyllabic Parhae Jurchen pe expanded into a Jin Jurchen pu(g)e and then recontracted into fo or fe. Parhae Jurchen pe may be an anachronism unless it is from a dialect not ancestral to the more conservative varieties of Jurchen with disyllabic pu(g)e/fuwe. Could standard Manchu fe be a descendant of the Pohai Jurchen pe-dialect (or another Pohai Jurchen dialect with the same type of *uge > -e compression)?

'Old' in Jurchen: a simplified possible family tree

(† = expected but not attested)

Proto-Jurchen *puge
Parhae Jurchen dialect 1: pe Parhae Jurchen dialect 2: †pu(g)e
Jin Jurchen dialect 1: †pe Jin Jurchen dialect 2: pu(g)e (Aotun 1)
Ming Jurchen dialect 1: †fe Ming Jurchen dialect 2: fo (Bureau of Interpreters) Ming Jurchen dialect 3: fuwe (Bureau of Translators)
Standard Manchu fe (Did these dialects survive into the Manchu era?)

Maybe, though Manchu does have a single word with -uge: buge ~ buhe (rather than †be < *buge) 'gristle'. Are those loans from noncompressing dialects? How heterogenous is standard Manchu?

And I would like confirmation of the sound value of


Ideally I'd like to see a polysyllabic word written with that character in the Parhae material that corresponds to a Jin Jurchen pe or Ming Jurchen/Manchu fe. Without such interlocking of both internal and external evidence, I have no way of knowing how

would have been read in Parhae. Graphic similarity does not entail phonetic similarity: e.g., Jurchen 日 'day' and 月 'month' look exactly like Jin or Ming Mandarin 日 and 月 but are pronounced completely differently: inenggi and biya instead of *ʐi and *ɥe. I am convinced that there is graphic continuity between the Parhae and Jurchen scripts. I am more agnostic about projecting Jin Jurchen values back onto an earlier script that may have been used to write other languages, related or otherwise. (In theory even Koreanic and para-Japonic speakers in Parhae could have used the Parhae script.)

I am not even sure there is graphic continuity between this particular Parhae script character and its Jurchen (near-)lookalike. I have already mentioned the problem of the different shapes of the top elements. Might the Parhae character be the source of

<?> '?' (N4631 #1355; font from CCAMC.org)

in the Khitan (not Jurchen!) large script?

To the best of my knowledge, the character

is not attested in Jin Jurchen; it is only known - at least to me - from the Ming Jurchen 永寧寺碑 Yongning Temple Stele (lines 3 and 4) where

<pe ing>

transcribes Ming Chinese 平 *pʰiŋ.

There is no evidence that

was read with final -e or any other vowel since it is only followed by -ing in the corpus. Could that character have been devised in Ming Jurchen to write  Ming Chinese *pʰ, a consonant absent from native Jurchen words (in which p [pʰ] had shifted to f)? I would prefer to read that character simply as p.

Jin Qizong (1984: 24) suggests that

is derived from Jurchen

<FORTY> dehi 'forty'

or from the aforementioned Chinese character 平 *pʰiŋ.

I would add that there is an even closer match for the shape of Jurchen <p> in the Khitan large script:

<FORTY> (北大王墓誌 Epitaph for the Grand Prince of the North 5; font from CCAMC.org)

One other potentially related Khitan large script character

<?> '?' (N4631 #2026; font from CCAMC.org)

is a lookalike for simplified Chinese 圣 'sage'. Unfortunately nothing is known about the phonetic or semantic value of 圣 in Khitan.

Although <FORTY> in the Khitan and Jurchen large scripts and what I read as p certainly look similar, I can't understand why one would decide to write a phonogram <p> as a modification of a logogram <FORTY>. Jurchen dehi has no p in it, and the Khitan word for 'forty' probably sounded something like Written Mongolian döcin 'forty' which also lacks p. Why not add a dot to, say,

be (accusative marker)

to create a phonogram for <p(e)>?

In fact a dotted phonogram derived from <be> already exists - the aforementioned


which is only attested in noninitial position after sonorants (vowels and l; Jin Qizong 1984: 100).

Might the graphic similarity between

<be> and <(g)e>

indicate that *-g- might have lenited to [ɣ] ~ [β] - the latter perhaps just after u? - justifying the choice of <be> as a basis for <(g)e>? Cf. how Proto-Koreanic *-p- lenited to [ɣ] and [β] in different dialects:

Proto-Koreanic *tupur 'two' > southeastern Old Korean 二肸 <TWO.> tuɣur but western Middle Korean 두ᄫᅳᆯ tuβur

(This is a revision of Alexander Vovin's proposal that Proto-Koreanic *-b- became [ɣ] and [β] in different dialects. Now neither he nor I follow S. Robert Ramsey's proposal to reconstruct *b in earlier Korean.)

Medial lenition is common to both Jurchen/Manchu and Korea, and a detailed comparison of the process in the two might be interesting.

I'll get to more of the problems with 'Old Thirteen' in part 2.

¹I originally wrote ex nihilo, but that could be misinterpreted as a straw man, as no one has ever claimed that Wanyan Xiyin had invented the Jurchen large script without any influence from other scripts. 'From the Khitans' is better since the orthodox view is that Wanyan Xiyin took the existing Khitan large script and arbitrarily changed it.

On the other hand, I side with Janhunen who thinks the Jurchen script is a modification of the Parhae script which was a sister of the standard Chinese script. Contrasting the two views:

Ex Khitanis

Chinese script
Khitan large script
Jurchen large script

There is no Parhae script for the Jurchen script to derive from in the orthodox view which holds that the Parhae wrote exclusively in Chinese characters.

Ex Parhis (Janhunen)

Proto-Chinese script
Standard Chinese script
Nonstandard northeastern Chinese script
Parhae script
Khitan large script
Jurchen large script

Today it occurred to me that the lost Tabghach script might fit into the above schema as the ancestor of the Khitan large script:

Ex Parhis (this site)

Proto-Chinese script
Standard Chinese script
Nonstandard northern Chinese scripts
Tabghach script? Parhae script
Khitan large script
Jurchen large script

In the above scenario, one could speak of a  para-Mongolic (or 'Xianbeic' = Shimunek's 'Serbic'?) line of scripts (Tabghach and Khitan) and a  possibly 'Tungusic' line of scripts (Parhae and Jurchen). But this is extremely speculative, as we have no idea what the Tabghach script looked like; it might not have been what Janhunen called 'sinoform' (i.e., Chinese-like). It may have had no relationship to any of the other scripts in the diagram above.

Janhunen (1996: 153) wrote,

In view of the later ethnic situation in the border zone between Korea and Continental Manchuria, and in the absence of any contradicting evidence, the most natural assumption about the states of Koguryo and Bohai [= Parhae] is that they were dominated by people ethnically ancestral to the Jurchen. It is well known that the Bohai ruling elite as largely formed by descendants of Koguryo nobility [...] there are no indications that any significant number of people linguistically connected with the modern ethnic Koreans would, during this period, have been present outside of the United Shilla territory.

To some extent, the above conjecture about the possible Jurchen identity of the Bohai population is complicated by the fact that the Bohai people continued to be counted as a separate ethnopolitical entity even after the fall of the Bohai kingdom. Not only the [Khitan] Liao [dynasty] but also the [Jurchen] Jin [dynasty] system of ethnic administration registered the Bohai people as distinct from the Khitan and Jurchen populations.

Another complicating factor is the absence of Jurchenic elements in the Koguryo onomastic material which is split between Koreanic and Para-Japonic items.

My attempt to reconcile the above points:

- Koguryo was a multiethnic state with a Chinese-influenced Koreanic-speaking elite ruling over Tungusic (including Jurchenic), Koreanic, and Para-Japonic-speaking subjects.

- As there is no evidence for Para-Japonic in Parhae, the Para-Japonic language(s) of Koguryo may have become extinct by the time Parhae was established in 698. Or Para-Japonic speakers were in the part of Koguryo that Shilla conquered: i.e., the part that did not become part of Parhae.

- The majority population of Parhae was Tungusic-speaking; their languages were not prestigious and hence almost totally absent from written records except in the Parhae script at a local and unofficial level.

- 'Parhae' as an ethnonym could have been a cover term for various peoples speaking Jurchenic and/or para-Jurchenic languages: i.e., Tungusic languages more or less related to Jurchen.

- The Jurchen script, then, was an official version of the previously informal Parhae script originally used to record one or more relatives of Jurchen but possibly not Jurchen itself. Wanyan Xiyin may have introduced new usages or characters to adapt the script to Jurchen. This introduction may not have been by him alone; it could have paralleled what happened when the Mongolian script was adapted for Manchu five centuries later. Roth Li (2000: 13) wrote,

The process of modifying [the Mongolian] script [for Manchu] occurred over at least a decade and was not, as some Chinese, sources made it appear, carried out singlehandedly by Dahai in 1632.

Dahai may have been the new Wanyan Xiyin: i.e., the man later attributed with the result of a slow process involving multiple people. As far as I know, there is no contemporary documentation indicating that Wanyan Xiyin 'created' the Jurchen large script in 1119 according to his biography in History of the Jin Dynasty 73; that 'fact' may be a later oversimplification. The later dates from other sources (1121 in the Wanyan Xiyin inscription and 1123 in the Record of the Great Jin State) could be reconciled by viewing the 'creation' of the Jurchen script as a process spanning years before and after 1120. Cf. Wikipedia:

The date of the creation of the script (1119 or 1120) varies in different sources. Franke (1994) says that "[t]he Jurchens developed ... [the large script] ... in 1119". Kane (1989) (p. 3) quotes the Jin Shi [History of the Jin Dynasty], which states that "[i]n the eighth month of the third year of the [天輔] Tianfu period (1120), the composition of the new script was finished". The two dates can be reconciled as one may imagine that the work started in 1119 and was completed in August–September (the eighth month of the Chinese calendar) of 1120.

In fact Tianfu 3 is 1119, so Kane's (1989: 3) 1120 may be an error; his 2009 book has 1119 on p. 3. (The date of the script is on page 3 of both books!)

²My guess at the Latin ablative of Parhae. I declined Parhae like Thebae 'Thebes' and Sinae 'China' which are pluralia tantum (though of course no such concept exists in Korean or Chinese [the Korean name is the Sino-Korean reading of the Chinese place name 渤海). WHAT IF THE LANGJUN INSCRIPTION WERE IN JURCHEN?

It just occurred to me that the Jurchen large script was fifteen years old when the 郎君 Langjun inscription was written in Khitan in 1134. The Arkhara inscripton of 1127 demonstrates that the Jurchen large script was already in use in the years between its 'creation'¹. So why wasn't the Langjun inscription about a Jurchen aristocrat - none other than the emperor's brother - written in his language?

It would be interesting to try to construct a Jurchen version of the inscription. At least I can guess that they would have written 'Tang dynasty' as

<ta ang> (attested in 1185 in 大 金得勝陀頌碑 26)

which partly parallels the structure of the Khitan small script spelling


from the Langjun inscription, though of course the Khitan small script fuses two characters (<ta> and <ang>) into a single block unlike the Jurchen large script characters which remain full-sized.

All that makes me wonder about the influence of Khitan writing practices on the Jurchen script - and if they can be differentiated from Parhae writing practices. Were the Parhae the first to write CVC syllables as <CV VC> sequences?

¹Not the best word, as I agree with Juha Janhunen (1994) who first proposed that the Jurchen script is actually derived from a preexisting Parhae script rather than being invented on the spot. It was Alexander Vovin who introduced me to Janhunen's idea over twenty years ago. I just found 川崎保 Kawasaki Tamotsu's 「渤 海」文字資料からみた女真文字の起源に関する一考察 (2014), a response to Vovin's "Did Wanyan Xiyin Invent the Jurchen Script?" (2012).

Next: My thoughts on Kawasaki's article.

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 © 2002-2018 Amritavision