... was in 西田龍雄 Nishida Tatsuo's 『言語学を学ぶ人のために』 For People Who Study Linguistics in 1992:

229-199-140 073-163 261-303-205

<t.ang.en ki.ên l.ing.de>

tang.GEN Qianling.DAT

'to the Qianling tombs of the Tang dynasty'

I thought that book also had an analyzed sample of Jurchen, but I think I confused it with an article of Nishida's that I saw later in the 90s.

That three-block Khitan sample is from line two of the Langjun inscription which was once thought to be in Jurchen. Turns out that the 'Langjun' in question was Jurchen - the younger brother of the emperor of the Jurchen Jin dynasty. So this post still fits this year's focus on Jurchen, though I do have many non-Jurchen topics in a queue.

The first word has bothered me for years because I would expect the genitive suffix to be -an rather than -en after tang. But today I realize that perhaps -en obeys consonant harmony rather than vowel harmony. Jin Chinese *tʰaŋ 'Tang dynasty' ended in a velar *ŋ, and I hypothesize that in Khitan, velar consonants went with higher series vowels like e [ə] rather than lower series vowels like a which went with uvulars. I am projecting the distribution of Mongolian and Manchu velars, uvulars, and vowel series back into Khitan.

I predict that if Khitan borrowed a noun ending in -əq - a sequence as impossible in native Khitan as Jin Chinese *-aŋ - its genitive would end in †-an with an †a to match uvular -q. BITHE ENDEHE¹

Today while handwriting the date in Jurchen, I noticed one mistake and then made another.

First, looking at my New Year's image which I made in haste, I spotted that I had written

minggan 'thousand' (cf. Chinese 千)


topohon 'fifteen' (cf. Chinese 十五 - if only the other Jurchen '-teen' characters were as transparent!).

I don't have a Jurchen input method set up yet. That's something I ought to do to celebrate the 900th anniversary of the Jurchen large script. So I copy and paste characters from lists. I have a text file of all Jurchen characters organized by number of strokes and 'radicals'. Without bothering to take note of the number of strokes, I spotted a character with a 五 'five'-like shape at the bottom and pasted it into the image file. Unfortunately, I pasted the wrong one. I've reuploaded the image with the right one.

Then I wrote 'first month' incorrectly as

emu biya 'one month'

a calque of Chinese 一月 'one month' = 'January'. I forgot that the correct term is

niyengniye(n) biya 'spring month'

or perhaps

se biya 'year month'

attested only in Chinese transcription in the Bureau of Interpreters vocabulary as (#297). Kane (1989: 194) thinks it might be an error for the Jurchen cognate of Manchu aniya biya 'first month', lit. 'year month'.

(1.5.1:21: Is it coincidental that the characters for niyengniye(n) and se are so similar? Was one derived from the other? Did a Chinese misreading of niyengniye(n) as se result in the Bureau of Interpreters term for 'first month'? Does any attested Manchu variety have an equivalent of se biya?)

Jurchen niyengniye(n) [ɲəŋɲə(n)] 'spring' was transcribed in Chinese as

捏年 *njenjen in the Bureau of Translators' vocabulary

捏捏 *njenje in the Bureau of Interpreters' vocabulary

The presence or absence of final -n may have varied by dialect; cf. how 'horse' had -n in the Bureau of Translators' vocabulary but -o in the Bureau of Interpreters' vocabulary.

As for the final nasal -ng in the first syllable, I reconstruct it because both Manchu and other several other Tungusic languages have it - and because Ming Mandarin did not have a syllable like *ɲəŋ. So I think 捏 *nje was an attempt to imitate the first two-thirds of Jurchen [ɲəŋ] at the expense of the third. Moreover, the Chinese may not have heard Jurchen -ng before n-, and in rapid speech, niyengniye(n) might have been simplified to [ɲəɲɲə(n)] or [ɲəɲə(n)].

¹1.5.0:08: Bithe endehe is Jurchen for 'made a mistake in writing'. THE FIRST SYLLABLE OF THE JURCHEN WORD FOR 'NOSE'

There is disagreement not only about how the Jurchen word for 'nose' was written in the Jurchen large script but also about how it was pronounced. Jin Qizong (1984: 287) provides three reconstructions:

Jin Guangping: ʃoŋgi

Yamaji Hiroaki: šonggi

Kiyose Gisaburō: songi

Kane (1989: 314) reconstructed it as sunggi.

There is no doubt that the word is related to the root songgi- of Manchu songgiha ~ songgin 'tip of the nose'. But on the surface the Chinese transcriptions seem to represent slightly different forms:

Bureau of Translators: 雙吉 (for *šuwanggi?)

Bureau of Interpreters: 宋吉 (for *sunggi?)

I have long thought that the extant sources for Jurchen are not directly ancestral to Manchu. Do those forms contain retentions or innovations absent from Manchu? Apart from perhaps preserving an original final -i sans suffix, I think they are artifacts of Chinese transcription.

The problem was that Ming Mandarin did not have a syllable *soŋ that would have been a good match for the Jurchen syllable song. So the transcribers of the two bureaus found different solutions to this dilemma:

The Bureau of Translators solution: Ming Mandarin 雙 *ʂwaŋ approximated Jurchen o with wa (a labial glide-nonhigh vowel sequence). The syllable swaŋ did not exist in Ming Mandarin, so *ʂwaŋ was the next best approximation.

(There was a phase when early Mandarin had < *wɑ, but 雙 had *wa with a different vowel and was not affected by the shift.)

The Bureau of Interpreters solution: Ming Mandarin 宋 *suŋ approximated Jurchen o with u (a labial nonlow vowel).

It would be worthwhile to go through all the Bureau of Translators and Bureau of Interpreters transcriptons of Jurchen with an eye (or should I say an ear?) for the limitations of Ming Mandarin phonology and see if there are other cases of Manchu-like forms distorted through the lens of the Chinese finite syllabary.

That is not to say that all apparent slight differences from Manchu in those transcriptions can be explained away as inescapable compromises. Some transcriptions do represent genuine dialectal variants: e.g., in the Bureau of Interpreters vocabulary, the Jurchen word for 'sky' is transcribed as 阿瓜 pointing to agwa rather the expected abka [apka]. The Chinese could have transcribed Jurchen [pq] as *-pu-k-, but they didn't because they heard [ɢw] (or [ɣw]? - 1.5.0:05). The Bureau of Interpreters dialect had shifted *-pk- to -gw-. WHO KNOWS THE RIGHT WAY TO WRITE 'NOSE'?

Not me. At least not in the Jurchen large script.

The Unicode proposals I've seen for Jurchen (N3628 and N3788) have only two forms of songgi 'nose' from Jin Qizong's 1984 dictionary (pp. 267 and 257):

The dotted form is attributed to the Berlin text of the Hua-Yi yiyu; the dotless form is an unattributed variant.

I have not been able to find a third form from p. 267 of his dictionary:

He attributes it to Grube (1896), but I think it's a ghost, because I see three different forms on pages 26, 53, and 85 of Grube:

The first form is identical to Jin's Berlin form, but the others don't match the one he says is from Grube.

Kiyose (1977) has one more form from the Hua-yi yiyu:

Apart from the possible ghost, which of the five remaining forms is real? The only way to know is to examine the original texts or high-quality reproductions of them. This exercise demonstrates the limits of relying on publications with handwriting or fonts in lieu of the actual forms. 900TH ANNIVERSARY OF THE JURCHEN LARGE SCRIPT

Happy New Year!

This year is the 900th anniversary of the Jurchen large script.

I have been getting back into Jurchen over the past few weeks, and I hope to post more about it this year.

Key to the Jurchen large script characters above:

1. In the center: ju(r?)šen amba(n) bithe 'Jurchen big script'

1a. ju(r?)- 'Jur-' (red)

Could this have originally been a logogram for 'Jurchen'? This character is not attested without the following character which might have been added as a phonetic clarifier: <JURCHEN.šen>. Moreover, it is never used to write any word other than 'Jurchen'. If it originally stood for a simple CV syllable ju, I would expect to see it in more words. Hence I suspect its original reading was longer: jur or even juršen.

Unfortunately this character has no attestations outside the Hua-Yi yiyu.

I would like to say that *rš or *rc regularly became š, but the high frequency of and rc in Manchu casts doubt on that proposal. Nonetheless it is hard to believe that Mongols inserted an -r- into jürcin for no reason unless there is some analogy I'm missing (cf. -r-insertion in Tartar because of Tartarus).

I wondered if Mongolian -r- might correspond to the *-k of the Middle Old Chinese name of an early Manchurian people,  肅愼 *siwk dinh: *-k > *-g > *-ɣ > *-r? Janhunen (2004: 70) independently came up with the same idea years earlier, citing Dagur rhotacism. But other mismatches remain to be explained:

OC *s : j-

OC *d : -c- (but the affrication of Jurchen t/d to Manchu c/j would not happen for centuries, so Jurchen -c- cannot be from a stop!)

1.1.18:54: Ligeti reconstructed the Ming Jurchen reading as from Jin Jurchen jür. See Kiyose (1977: 90).

1b. -šen '-chen' (yellow)


1c. amba(n) 'big' (green)

I suspect that this was originally a logogram for amban 'big' and that the spelling <BIG.an> with a phonetic clarifier <an> was a later innovation.

The Jurchen term for 'large script' is not attested, so I don't know if it contained amba or amban or something else entirely.

1d. bithe 'script' (blue)


2. Around the center: uyewun tanggū 'nine hundred'

2a. uyewun 'nine' (orange)

Logogram obviously related to Chinese 九 'nine'.

2b. tanggū 'hundred' (pink)

Logogram presumably related to Chinese 百 'hundred'.

Last night it occurred to me that Jurchen/Manchu ū could be romanized as ů to signify how [ʊ] is between [u] and [o]. But I will go with the traditional letter here.

3. The corners: juwe minggan oniyohon aniya 'two thousand nineteen year'

3a. juwe 'two' (yellow-green; top right)

Logogram identical to Chinese 二. Later written with a L-shaped second stroke that distinguished it from the Chinese form.

3b. minggan 'thousand' (blue-green; bottom right)

Logogram presumably related to Chinese 千 'thousand'.

3c. oniyohon 'nineteen' (dark purple; top left)

Logogram possibly related to Chinese  九 'nine'

-hon may be a Khitanic word for 'ten' related to Mongolian arban (< *xarban?) 'ten'. The Khitan word for 'ten' is unknown. The word for 'nineteen' in Khitan was 'ten-nine', not 'nineteen', but a related language or dialect could have had the opposite order.

If -hon is Khitanic, oniyo- should be too, but the Khitan word for 'nine' was completely different: is.

3d. aniya 'year' (light purple; bottom right)

4. Background: indahūn 'dog' (the animal for 2019 which also happens to be mine)

Logogram. The word was later spelled with a phonetic clarifier as <DOG.hūn>.

It is tempting to link this word to Japanese inu 'dog', but the mismatch between -dahūn < *-dakun and -u would have to be explained. I would expect a Japanese cognate of indahūn to be †idaku with †-d- < *-nd-. And I'm overlooking the problem of the initial consonant - Jurchen has lost an initial *ŋ- still preserved in some other Tungusic languages.

1.1.16:04: The character could be related to Chinese 犬 'dog'. If Janhunen's hypothesis that the Jurchen script is derived from a sister of the Chinese script is correct, then <DOG> might be a more elaborate descendant of the drawing of a dog ancestral to犬.

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