Today is a

<RED¹.giyan OX².an> ful(a)giyan wihan inenggi 'red ox day'

in the Jurchen calendar.

It is hard at a glance to tell whether 'red' was disyllabic [fʊlɢʲaʜ]³ (i.e., identical to later standard Manchu fulgiyan⁴) or trisyllabic [fʊlaɢʲaʜ]. The Bureau of Translators vocabulary (early 1400s?) has the trisyllabic transcription

弗剌江 *fu la kjaŋ (#617)

whereas the Bureau of Interpreters vocabulary (c. 1500?) has the disyllabic transcription

伏良 *fu ljaŋ (#1100)

The obvious solution would be to posit [a]-loss: earlier trisyllabic [fʊlaɢʲaʜ] became later disyllabic [fʊlɢʲaʜ]. But it is not clear that the varieties of Jurchen within the two vocabularies are two snapshots of the same dialect at two different points in time. It is not even clear that each vocabulary is homogeneous: i.e., reflecting only a single dialect rather than a mix learned from various informants who may not even have been contemporaries. Lastly, it is possible that Ming Mandarin *la was merely a device to write a simple Jurchen [l]. There was no Ming Mandarin syllable *ful (and hence no character for such a syllable), so [fʊl] might have been transcribed as 弗剌 *fu la. On the other hand, other Tungusic languages do have an a after l in 'red', and the undoubtedly related Proto-Mongolic word for 'red' does have an *a between *l and *g: *hulagan, suggesting that the *a at least dates back to when Tungusic borrowed the word from Mongolic (or vice versa?). I should look into this more.

As for 'ox', the Bureau of Translators vocabulary has the transcription

委罕 *wej xan (#143)

whereas the Bureau of Interpreters vocabulary has the transcription

亦哈 *i xa (#411)

Jin (1984: 128) takes the transcription *wej xan at face value, reconstructing Jurchen weixan (= weihan in my notation) which violates vowel harmony (e and a belong to opposing vowel classes and should normally not be in the same root). Kiyose (1977: 105), on the other hand, disregards the *w- without explanation and reconstructs Jurchen ihan which matches Manchu ihan [ɪχaʜ] 'ox'. Kane's (1989: 216) reconstruction of iha is straightforward.⁵

Once again, the obvious solution is to posit loss over time: earlier wi- became later i-. The *-e- of the transcription simply reflects the fact that Ming Mandarin had no syllable *wi; *wei was the closest match for Jurchen [wɪ]. Manchu has no wi, so all early Jurchen wi became later Jurchen/Manchu i. Japanese had the same wi > i change, which is why the kana / for <wi> are now obsolete.

The trouble is that there is no support elsewhere in Tungusic for an initial w- in 'ox'; all the non-Jurchen forms in Cincius (1975: 299) start with i-type vowels. (Oddly I cannot find the 'ox' cognate set at starling.)

Is it possible that Jurchen once preserved a Proto-Tungusic *w- that all other languages lost before *i⁶, even t though Jurchen/Manchu is considered innovative? There is no a priori reason to reject that possibility; a language that is innovative in many ways can still be conservative in at least one way. Ideally I would like to find other cases of Jurchen wi- corresponding to i- elsewhere in Tungusic.

¹2.10.13:45: I have only seen this character followed by <giyan> in the Bureau of Translators vocabulary. Nonetheless I don't think it was a Ming dynasty addition to the Jurchen character set. It is not attested in words other than 'red'. So I suspect that it was originally a standalone logogram <RED> and that <giyan> was added later to represent its final syllable.

²2.10.13:55: This character appears by itself in the Jurchen Character Book manuscript thought to be an early catalog of characters. That suggests it was originally a standalone logogram <OX> and that the <an> in theBureau of Translators vocabulary is a later addition.

³2.10.23:29: Or perhaps [fʊlʁʲaʜ] with [ʁ]. The Bureau of Interpreters transcription without *k might indicate that the uvular stop had lenited to the point where it was hard to perceive.

⁴2.10.17:17: The Manchu spelling fulgiyan appears trisyllabic, but -iy- is just a means to write palatalization.

⁵2.10.14:19: The absence of an -n present in Manchu is a common trait of the Bureau of Interpreters inscriptions. See Kane (1989: 112) for other cases of a Jurchen zero : Manchu -n correspondence.

⁶2.10.23:27: Cincius' enormous Tungusic dictionary (1975) only has six pages of entries for в- <v> and only two entries for ви- <vi>, both for Evenki words without cognates elsewhere. So it does not appear there is any obvious modern (i.e., non-Jurchen) evidence for reconstructing Proto-Tungusic *wi-. I suspect *w- was once far more frequent and lost in most environments (e.g., in Manchu w is only possible before a and e). The only *w-word I could find in starling's Proto-Tungusic is *wa- 'kill' which is solidly attested throughout the family.

As tempting as it may be to reject wi- in Jurchen (and, by extension, earlier Tungusic), the Chinese transcription 委 *wej is difficult to explain away since (1) the Chinese could have easily chosen an *i-character to write a Jurchen i- and (2) I cannot think of any *i-character that might be miswritten as 委 *wej. WAS G'AG'AI KOREAN?

Tonight it occurred to me that the Manchu script was ironically credited to two men with non-Manchu names, Erdeni and G'ag'ai.

Erdeni is the Mongolian borrowing of Sanskrit ratna- 'jewel' with an initial vowel added to avoid an initial r- forbidden by Mongolian phonotactics.

Crossley (2000: 185) wrote,

Like those of many leaders of the Nurgaci period, Erdeni's origins are difficult to characterize. He had a Mongol name and certainly could write Mongolian [the written language used by the late Ming Jurchen, just as the Jin Jurchen before them had used Khitan]; he may have been a native of a Mongolian-speaking region. But the early Manchu records suggest that he was also expert in Chinese, and that in his contemporary frame he functioned as a Nikan.

Nikan is Manchu for 'Chinese', and in Crossley's view, the term does not simply mean 'of Chinese descent'; it refers to "those who behaved as Chinese" (2000: 55). One could be ethnically Mongol - or Jurchen or Korean - and function as Nikan.

Crossley (2000: 188) speculates that G'ag'ai might have been of Nikan "background" (ethnicity) "but in fact it [his heritage?] was irrelevant" since his "responsibility for literate acts under the [Jurchen] state" made him live as a Nikan.

So was G'ag'ai [kakaj] a Nikan - er, Chinese - name? It has the velar-a sequence absent from native Jurchen/Manchu (and Mongolian) words¹. However, I don't know of any plausible Ming (or even modern standard) Mandarin name element pronounced [ka]². Here's a wild guess - might the name be Korean: i.e., something like 가개 Kagae ([kakaj] or [kagaj]³ in the 16th century)? Googling for 김가개 Kim Kagae, I found this 2009 article by 최범영 Chhoe Pŏm-yŏng which not only mentions an attestation of the name in 1404 but independently speculates that Korean Kagae is the source of G'ag'ai's name.

2.9.0:59: The name Kim Kagae appears as 金加介 in the entry for day 29, month 8 of the 12th year of King Sejong's reign (1430) in 世宗實錄 Sejong shillok 'Veritable Records of [King] Sejong'. I can't find any mention of a Kim Kagae in the entry for 1404 in 太宗實錄 Thaejong shillok 'Veritable Records of [King] Thaejong', so I don't know if the Kagae of 1404 is also spelled 金加介. It may be a native Korean name with varying Chinese character spellings.

¹2.9.0:05: The situation with g'a [ka] and ga [qa] in Jurchen/Manchu is similar to that for k'a [kʰa] and ka [qʰa].

²2.9.1:21: There are, in fact, no [ka] syllables in 'Phags-pa transcription a few centuries earlier, and [ka] only has a marginal status in modern standard Mandarin⁴. Windows 10's Pinyin IME's first suggestion for Pinyin ga [ka] is the transcription character 噶 for foreign ga: e.g., 喀什噶爾 Kāshígá'ěr 'Kashgar' and ... 噶蓋 Gágài, the Chinese transcription of 'G'ag'ai'. The other ga-suggestions are 嘎尕尬旮呷軋釓尜伽咖戛夾胳嘠錷玍魀, none of which I've ever seen in a name.

³2.9.1:01. I don't know if intervocalic voicing already existed in 16th century Korean. [aj] did not monophthongize to [ɛ] until the "end of the eighteenth century" (Lee & Ramsey 2011: 264).

⁴2.9.1:12: Old Chinese was full of *ka (= Baxter and Sagart's *kˁa) which became Middle Chinese *ko which in turn became modern standard Mandarin [ku].

Middle Chinese gained a new *ka from Old Chinese *kaj. (The final *-j shielded *-a from raising before being lost.) This too was lost in modern standard Mandarin: the new *ka became *ko and then [kɤ].

You can see part of a vowel shift chain: *aj > *a > *o > *u.

In tabular form:

Old Chinese
Middle Chinese
Early Mandarin
Modern standard Mandarin
[ɤ] after velars

I have excluded reflexes of early Mandarin *o after other initials. FUK'ANGGAN I GEBU (THE NAME OF FUK'ANGGAN)

The 乾隆 Qianlong emperor died 220 years ago today. He appointed 福康安 Fuk'anggan to lead the troops in the Sino-Nepalese War.

Fuk'anggan has an interesting name for two reasons:

1. It has the typical Chinese three-syllable pattern, it contains the Chinese syllable k'ang with a velar-a combination [ka] absent from native Manchu words¹, and it even has a meaningful, positive Chinese character spelling: 'good-fortune health peace'. Yet it is romanized as a trisyllabic single name because it is not a Chinese name - 福Fu is not his surname, though it may have been influenced by his clan name Fuca (spelled with a different fu, 富 'rich', in Chinese: 富察). And his personal name was not 康安 K'anggan; it was Fuk'anggan. At most I could say that Fuk'anggan is a Sino-Manchu hybrid; it wouldn't have been a Jurchen name many centuries ago.

2. I would not expect 安 to correspond to Manchu gan; it was read an in Beijing.

Does gan for 安 reflect the influence of a Mandarin dialect in which 安 was read ŋan? (2.8.0:04: There are many such dialects today.) [ŋ] was not a possible syllable-initial consonant in Manchu, so [ŋŋ] was not possible word-internally in Manchu, and [fukʰaŋɢan] would be the closest Manchu approximation of a Mandarin *fu kʰaŋ ŋan.

¹2.8.1:08: ka in the Möllendorff romanization of Manchu that I use represents [qʰa] with a uvular [qʰ]. [qʰa] is more common in Manchu than the loan sequence [kʰa], so it makes sense to use ka for the more frequent syllable and k'a for the less frequent syllable.  The apostrophe after velar letters corresponds to velarity, not aspiration as in the Wade-Giles romanization of Mandarin.

Möllendorff did, however, use the apostrophe for aspiration to romanize other Manchu letters for Chinese transcription: ts' [ts] and c' [tʂʰ] (the latter only before y in his romanization). I favor Norman's decision to drop the aspiration in those cases since there is no native [tsʰ] that contrasts with ts'. Nor is there a native cy that contrasts with c'y. HAVE AN ICE DAY

Today is the first day of the new year. The first of the month - ice inenggi 'new day' in Jurchen (see Andrew West's online Jurchen calendar):

Jin Qizong derived the character for ice 'new' (pronounced with two syllables: [itɕə]) from the left side 亲 of Chinese 新 'new'. I couldn't quite buy that because of the asymmetry of the Jurchen character and the symmetry of 亲. But I just found the asymmetrical Chinese variant 𢀝 from the Jin (!) dynasty dictionary 四聲篇海 Si sheng pian hai 'Sea [of Writings] Arranged by the Four Tones'. (I got the title translation from Imre Galambos.)

As an adherent of Janhunen's ex Parhis² hypothesis, I don't think the Jurchen script was Chinese mutiliated on the spot by Wanyan Xiyin in 1119. Rather, I think 完顏希尹 Wanyan Xiyin adapted an existing Parhae script that was a local (i.e., Manchurian) variant of the Chinese script. And the character for 'new' in the Parhae script might have been that variant 𢀝 or something close to it - possibly even identical to the Jurchen character. CIKOSKI'S NOTES FOR A LEXICON OF CLASSICAL CHINESE

Today I discovered John Cikoski's Notes for a Lexicon of Classical Chinese, Volume I (2011) while looking for Bernhard Karlgren's (1954) quotation about the excesses of phonemics. The book would have strongly appealed to me if I were still a Karlgrenian.

In the early 90s I borrowed every book of Karlgren's I could find. My favorite remains his 1954 Compendium of Phonetics in Ancient Chinese and Archaic Chinese which walked me through the reasoning behind his reconstructions. I no longer agree with him on many matters, but at least I know why he did what he did. A scientist must insure that his results are replicable and not seemingly pulled out of a hat.

When I first saw Pulleyblank's Middle Chinese (1984) in 1992, my gut reaction was disbelief. Chinese couldn't have looked like that! Too bizarre! It would be another year before a second look at Pulleyblank persuaded me.

If I had never become a Pulleyblank fan, I would enjoy Cikoski's book more. Cikoski picks up where Karlgren left off and builds upon the master's reconstruction while still avoiding what he perceives as the pitfalls of modern approaches. Details later.

2.5.21:17: But in the meantime I found the other volumes of his Lexicon with a copyright notice, covers, and a non-Unicode Grammata Serica font with a key here. THE FATHER OF JURCHEN LANGUAGE STUDIES

Today I realized that's who Wilhelm Grube was when I read his Wikipedia entry. I've known about him since the mid-90s. I have no idea why it took me so long to see the obvious.

I also saw Andrew West's scan of Grube's seal (葛祿博藏書印 'Seal of the Library of Ge Lubo', read from top to bottom, right to left):




cáng 'to store'

I didn't recognize the seal form of 藏 'to store'; it's so much simpler than the regular print form 藏. The closest Unicode match is in CJK Unified Ideographs Extension B: 𤖋. I'm surprised 𤖋 is not in this list of 28 variants of 藏.

As simple as 𤖋 is, it's not as simple as the proposed second-round simplified character 䒙 - one of the lucky ones in Unicode (CJK Unified Ideographs Extension A, to be exact). Some second-round characters still aren't in Unicode (and are marked in red on Andrew West's page). It's incredible ... we can type Tangut in Unicode but not "newspapers, books, and publications of all kinds" written in second-round simplified characters in 1978.

藏/𤖋/䒙 has two standard Mandarin readings, cáng and zàng. Neither quite matches the reading of the phonetic of 䒙, 上 shàng. However, 上 is a very transparent phonetic for 䒙 in Wu varieties like Suzhou in which both 藏/𤖋/䒙 and 上 can be [zɒŋ]¹ (ignoring tonal differences; compare the readings here and here).

¹2.4.1:10: 上 also has a colloquial Suzhou reading [zaŋ] with a different vowel.

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