If you were expecting me to reveal something personal about me here, you'll be disappointed.

I am disappointed because I can't reveal more about the origin of the Jurchen character

bitxə 'writing'

In my last entry, I described how the word probably ultimately comes from Middle Chinese 筆 *pit 'brush'.

But why is it written with what looks like Chinese 亻 'person'? I have no idea.

The right side of bitxə looks like the dotless variant of the Jurchen character


burə (phonogram; in first half of burəki 'dust')

burə 'prospective participle suffix' (< causative suffix -bu- + prospective participle suffix -rə)

burən 'prospective verbal noun' (with nominal suffix -n; at the end of a word: e.g., usgaburən 'grudge' [n.], cognate to Manchu usha- 'to bear a grudge'; tuwiburən 'postponing')

(definitions and analyses from Kiyose 1977: 72, 100, 119, 139)

which has no semantic resemblance to bitxə 'writing' and no phonetic resemblance beyond a shared b-. The graphs for burə(n) may be derived from Middle Chinese 夫 *pu 'man' ~ *bu (grammatical particle). Perhaps 夫 was borrowed into the (predecessor of the) Parhae (Bohai) writing system when it was still pronounced *pu/*bu in Chinese and carried over into the Khitan large script and later into the Jurchen script long after its Chinese reading had shifted to *fu.

Readings for 夫 and its derivatives 

Time period Chinese Parhae Khitan large script Jurchen
8th century *pu, *bu *bu (no Khitan script yet) (no Jurchen script yet)
10th century *fu ś (not bu!)
12th century (extinct) burə(n)

I do not understand why 夫 has an anomalous reading in Khitan.*

Could the graphic similarity between Jurchen

bitxə 'writing' and burə(n)

imply that the Parhae word for 'writing' was similar to the Parhae reading for 夫? Middle Korean put 'brush' (< Late Old Chinese 筆 *put 'id.') is phonetically close to Middle Chinese 夫 *pu, and there were Koreanic speakers in Parhae (see Vovin's "Korean Loanwords in Jurchen and Manchu" [2007] for the linguistic evidence).Perhaps the Parhae prototype of Jurchen

was used to write a Parhae *put 'brush' (with 亻 arbitrarily added to distinguish it from 夫)and was later recycled to write Jurchen bitxə (ultimately also related to the Parhae word.

There is a Khitan large script character

of unknown function. Could this have represented a Khitan word for 'writing' - the para-Mongolic source of Jurchen bitxə?

A possible graphic derivation of Jurchen bitxə


Parhae ?*pu (a phonogram?) > Parhae ?*put 'brush' > Khitan ?*biteke 'writing' > Jurchen bitxə 'writing' (reconstructed by Kiyose 1977 as bitehe)

NB: The Parhae characters are hypothetical. Only the last two readings may be directly related through borrowing; they are only indirectly related to Parhae ?*put 'brush'. Only the very last reading is attested.

I think this Koreanic-based graphic etymology is more probable than Yamaji's derivation of

bitxə 'writing'

from Chinese 便 *bien 'convenience' which has a partial phonetic resemblance (a shared bi) and no semantic resemblance.

The title of this post was originally simply going to be "More Personal", a reference to the components

亻 'person' + 更 'more'

of 便 *bien, but another derivation came to mind as I wrote it, so I added the second half of the title.



was the Khitan large script spelling for Liao Chinese 帥 *šoi 'commander'.

坐 was *tso 'sit' in Liao Chinese. I'll explain why it might have become a phonogram for Khitan oi later.

There is also a Khitan large script character

of unknown function that vaguely resembles Jurchen



This latter Khitan large script character appears in line 9 of 耶律褀墓誌 (1108):


<? ? ? ? iri>

iri means 'name', so maybe the preceding four characters represent a reduplicative name with the pattern 'A B A B'. FROM BRUSH TO BRUSH: THE SCRIBAL CYCLE

0. Introduction

I haven't written about Khitan or Jurchen in a long time. Alas, they only make cameo appearances in this post, but better something than nothing.

Today I was looking at the presumably Jurchen title of bitxəšï-史 'scribe-scribe' (the 史 is presumably the Jpn word fubito  'scribe' in the URL; see below) and wondering where its first half came from. Inspired by the Qing transcription of its Manchu cognate bithesi as 筆帖式 bitieshi 'brush note style', I thought it might be from Middle Chinese

筆記 *pɨt-kəʔ 'brush-record' plus 士 'person'

but I don't even know if the compound 筆記 existed back then, and I was only a *bit right at best. I know better now ... I think:

1. The Old Chinese *rut

Sagart (1999: 212-213) wrote about the word 筆 'brush' at length. Here's my take based on his with additions of my own:

An early word for 'brush' was written as a drawing of a brush 聿 without 竹 'bamboo' on top. It might have been *rut or *lut in Old Chinese. The Middle Chinese evidence for the initial is ambiguous (see below). In Wu, the word was disyllabic 不律 *pɯ-rut ~ *pɯ-lut corresponding to the compressed monosyllabic forms Qin 筆 *prut ~ *plut and Yan 弗 *put.

2. Early Sinoxenic

17th century Middle Vietnamese bút < *put 'brush' and 15th century Middle Korean put 'brush' were borrowed centuries earlier from Old or Middle Chinese dialects with *put-type words. (OC *pl- and *pr- would have become MV bl- and s-, not MV b-.)

Old Japanese pumi 'writing' is unrelated unless it is a simplification of a Paekche *put-mi. I have wondered if pumi is derived from pum- 'to step on'; writing is like tracks (footsteps) on paper. The fubito 'scribe' in the URL of bitxəšï-史 is from Old Japanese pumbitə, a contraction of pumi-pitə 'writing-person'. OJ punde 'brush' may be from *pumi-te 'writing-hand'.

3. Early Middle Chinese

In Middle Chinese as recorded by the dictionary tradition,

聿 was *jwit, implying OC *lut

筆 was *pɨt, implying OC *prut (OC *plut would have become MC *put)

If one assumes that phonetics and their derivatives sounded almost alike, then one would like to reconstruct the phonetic 聿 and its derivative 筆 with either *l or *r, though the MC readings point to different liquids. Perhaps that assumption is wrong or was valid only for the dialect of the OC speaker who created the graph 筆. We also cannot be certain that the MC dictionary tradition readings descend from the same dialect; 聿 *jwit < *lut could be from an *l-dialect and 筆 *pɨt < *prut could be from an *r-dialect.

4. Turkic

At some point, MC shifted to *i. Early Turkic bitiː- 'to write' (Clauson 1972: 299) may have been based on an unattested *bit 'brush' borrowed from MC *pit after that shift. I would expect Turkic *bït with back ï from MC *pɨt.

Chinese *p- was borrowed as b- because Turkic lacked initial p- in native words (Erdal 2004: 64-65; pre-Turkic *p- became zero except in Khaladj where it became h-).

bitiː- had the derived noun bitig 'a general word for anything written' with a deverbal suffix -g (Clauson 1972: 303).

Clauson (1972: 304) listed three different early Turkic words for 'scribe':




Only the first two are related to bitiː- with the suffix -çi indicating a human agent. (So much for my old idea that the third syllable of Manchu bithesi was from Chn 士 'person'!) The third is from Syriac.

5. A Khitan detour

I am not aware of a biti-type word for 'writing' in Khitan. Kane (2009: 127) transliterated the Khitan small script spellings for 'writing'


as <us.gi>. I think it might have been usig (assuming that <CV> symbols did double duty for VC as well as CV sequences); cf. Written Mongolian üsüg 'writing'.

6. Jurchen and Mongolian

I am not aware of an usig-type word for 'writing' in Jurchen. Jurchen

bitxə < ?*bitkə 'writing'

is identical to Manchu bithe 'book' (which could also be romanized as bitxə). This word and other Tungusic bit-type words for 'writing' (see Cincius 1975 I: 86 for a list) were presumably borrowed from some Para-Mongolic cognate of Written Mongolian bicig < *bitig 'writing' (obviously in turn from Turkic bitig 'id.').


bitxəšï < ?*bitkəsi or *bitkəši 'scribe'

must then be from some Para-Mongolic cognate of WM bicigeci < *bitigeci 'scribe' (obviously in turn from Turkic *bitigçiː 'id.').

7. A golden detour

The correspondence of Jurchen š* < *s to WM c in 'scribe' is reminscent of the Manchu s : Jurchen c correspondence in

Manchu aisin : Jurchen ~ alcun 'gold'

For much more on these words for 'gold', see my posts from November and December 2011:

"From Alcun to Aisin?"

"From Alcun to Aisin Again?"

"Dye and Dry"

"Golden Bones and Tigers"

"Ligeti on 'Gold'"

"Ligeti on 'Jurchen'" (Part 1 / Part 2 / see the end of Part 3 for Janhunen on Pre-Para-Mongolic *c > Para-Mongolic  > Manchu š; one might expect this PM language to have been Khitan, but I know of no evidence for *c > in Khitan)

"Heavy Mountains"

"Going An about Gold"

I forgot I had written that much about 'gold'. It's an important word for the Jurchen since the Khitan called them the 'Gold' and their dynasty was called 'Gold' in Jurchen and Chinese.

Enough about 'gold'. Back to 'writing' ...

8. Why not just derive the Jurchen words from Pre-Mongolian?

J bitxə < ?*bitkə implies Para-Mongolic *bitke which does not match Pre-Mongolian *bitig. Moreover, bitxəšï < ?*bitkəsi or ?*bitkəši 'scribe' implies Para-Mongolic *bitkesi with *s or *bitkeši with instead of the c of Pre-Mongolian *bitigeci.

9. Manchu

Although Jurchen bitxə 'writing' and Manchu bithe = bitxə 'book' are identical, the Jurchen and Manchu words for 'scribe' are slightly different:

J bitxəšï : M bithesi = bitxəsi

I don't think the final vowels were different. So far I've been using bitxəšï-史's spelling of J bitxəšï for convenience. Jin Qizong's Jurchen reconstruction has šï but no ši. I suspect that all his šï were really ši. Jin (1984: 147) derived the Jurchen graph

for šï from Chinese 失 which was pronounced *ši (not *šï) during the Jin Dynasty. Jin's choice of šï may be influenced by its modern Mandarin reading which could be romanized as šï (= Pinyin shi). The shift of *ši to *šï occurred by the late Ming.

As for the consonant before that final vowel, standard Manchu is not a direct descendant of Jurchen. The latter had palatalized *si to ši, whereas Manchu preserved si (Kane 1989: 113). Hence M bithesi is more conservative than J bitxəši, even though the former is attested after the latter.

Another possibility is that Manchu s is a hypercorrection by Manchu speakers who noticed that standard Manchu si corresponded to Jurchen ši and who assumed that J bitxəši should have been a -si word even though it was historically a -ši word.

The Chinese transcription 筆帖式 bitieshi implies a nonstandard Manchu *bitheši = *bitxəši which may be from a dialect that had palatalized *si. I would have expected something like *筆帖西 bitiexi (pronounced *[pithjesi] during the early Qing; there was no Mandarin [b] in spite of the romanization) as a Chinese transcription of Manchu bithesi.

Whoever came up with 筆帖式 had no idea that he had taken the word full circle from 筆 'brush' back to 筆 'brush':

*pit > Turkic bitigçiː > Para-Mongolic *bitkesi > Jurchen/nonstandard Manchu bitxəši, transcribed as 筆帖式

Appendix: Post-Early Middle Chinese 'brush'

I wanted to focus on the descendants of Turkic bitig after section 4, so I left out other later descendants of 筆:

Early Middle Chinese 筆 *pit became *pir in northern Late Middle Chinese. This *-r is reflected in Written Tibetan pir 'brush'and Middle Korean pirʔ 'brush' (modern Sino-Korean phil with an irregular aspirate).

Northern Chinese eventually lost *-r. Jurchen

pi 'brush' (with a graph derived from 聿 [Jin 1984: 121])

is from northern post-Middle Chinese *pi(ʔ). The initial *p- had weakened to f- in late Jurchen and standard Manchu fi.

Southern Chinese retains *-t to this day in Cantonese 筆 pat. (6.25.00:10: Sino-Vietnamese should have bất < *pət from a Cantonese-like southern Chinese dialect, but there is no such reading for 筆; if it ever existed, it was displaced by bút.)

Next: More Personal?

Tangut fonts by Mojikyo.org
Tangut radical and Khitan fonts by Andrew West
Jurchen font by Jason Glavy
All other content copyright © 2002-2012 Amritavision