One to Twenty in Jurchen, Khitan and Lute


[Mirrored from]

The Jurchen number system, and its relationship to Mongolian and Khitan, has been well-studied (e.g. Janhunen 2003 pages 399–400; Kiyose 2000 page 184; Laufer 1967). What is less well understood is the relationship between Jurchen number characters and the tablature signs used for writing music for the Chinese lute (pipa 琵琶).

Detail from a Liao dynasty (907–1125) tomb mural in Inner Mongolia showing Khitan women playing the flute and lute

Source : Zhongguo Meishu Quanji 中國美術全集 (繪畫編) vol.12 plate 163

Khitan Numbers

Number Chinese Khitan Large Script Khitan Small Script Phonetic Gloss Mongolian
Undotted Dotted (♂)
1 nigen
2 qoyar
3 ɣurban
4 or dörben
5 (tǎo) = *tau tabun
6 or or ǰirɣuɣan
7 doloɣan
8 naiman
9 yisün
10 arban
20 二十 or or qorin
30 三十 ɣučin
40 四十 or or döčin
50 五十 tabin
60 六十 ǰiran
70 七十 dalan
80 八十 nayan
90 九十 yeren
100 (zhǎo) = *jau ǰaɣun
1,000 mingɣan
10,000 or tümen

The two Khitan scripts, the Khitan Large Script [KLS] and the Khitan Small Script [KSS], each have their own completely unrelated sets of logographic characters for numbers. Many of the KLS numerals are derived from the corresponding Chinese number characters, but the derivations of other KLS numerals and all the KSS numerals are not obvious, and in many cases the glyph forms seem to have been arbitrarily chosen. The KSS numerals 1–9 occur in undotted and dotted forms, where the dotted form is thought to have represented the grammatically masculine form of the word. Oddly, if it is true that the two Khitan scripts were used to write the same Khitan language, the KLS numerals 1–9 only occur in a single form with no distinction for grammatical gender.

Words for 11 through 19 are constructed regularly from "ten + unit" (as in Chinese) in both Khitan scripts, which suggests that there were no special words for these numbers (as in Jurchen). However, there are special characters for the decade numbers from twenty to ninety in both KLS and KSS, which implies that there were special Khitan words for 20 through 90 analagous to, and probably cognate with, the Mongolian numbers 20 through 90. Half of the KLS decade numbers incorporate the same component (仒), which may represent a common phonetic element (presumably the -ty part of the word), but it is not clear whether the other component in these characters has a phonetic role or not. In contrast, the glyphs for the KSS numbers 20 through 90 are very simple, comprising only one or two strokes, making them noticably less complex than the glyphs for 1 through 10.

As Khitan numbers are represented logographically in both KLS and KSS, there is little direct evidence for the pronunciation of Khitan numbers. However, phonetic transcriptions in Chinese characters for the numbers 'five' (tǎo 討) and 'hundred' (zhǎo 爪) are given in the History of the Liao Dynasty 遼史, which does allow for the reconstruction of these two numbers:

Both numbers are clearly cognate with the corresponding Classical Mongolian numbers, and they both show the same phonetic features relative to Mongolian: a) absence of final -n; and b) loss of a medial consonant with accompanying diphthongization. This may give us some clues as to how to reconstruct some of the other Khitan numbers for which there is no direct phonetic evidence. For example, we might suggest by analogy with the Khitan and Mongolian forms for 'five' that:

The KSS character for 'fifty' is identical to the Chinese character 乙 , the second of the ten heavenly stems, which is notably used in the astronomical and Daoist word tàiyǐ 太乙 "Great One". Is it possible that 乙 was chosen to represent 'fifty' precisely because of the association of 乙 with 太乙, and because the Khitan pronunciation *tai for 'fifty' was homophonous with Chinese 太? This sounds like a very circuitous and contrived derivation, but perhaps it was precisely because of this sort of indirect derivation of characters that it is so hard to explain KLS and KSS logograms.

As an example of circuitous derivation of a Khitan character from Chinese, the native Khitan word for 'person', reconstructed as *ku, is written identically to the Chinese character 仁 rén "benevolent". There is no phonetic or semantic relationship between the Chinese and Khitan characters, but it is evident that the Khitan character is written as 仁 because the Chinese character is a homophone of the Chinese character 人 rén "person". But why write the Khitan character for 'person' with a homophone of the Chinese character for 'person' instead of simply writing the Khitan character as 人?

We can suggest similar indirect derivations of other Khitan logograms from Chinese characters. According to the History of the Liao Dynasty, "a male, when an infant is 'yellow', when four years old is 'little', when sixteen years old is 'middle', and when twenty-one years old is an 'adult'" 男幼為黃,四歲為小,十六為中,二十一為丁. The KSS logogram for 'twenty' is identical to the Chinese character 丁 dīng 'adult male' (originally meaning 'nail'). Is this a coincidence, or was 丁 deliberately chosen to represent 'twenty' because of its association with the age of 'twenty-one'?

Jurchen Numbers

Number Chinese Jurchen Manchu Mongolian
Glyph Gloss Reconstruction
1 or 厄木 *əmu ᡝᠮᡠ emu nigen
2 or *ʤo ᠵᡠᠸᡝ juwe qoyar
3 or 以蘭 *ilɑn ᡳᠯᠠᠨ ilan ɣurban
4 or 都因 *duin ᡩᡠᡳ᠌ᠨ duin dörben
5 or or or 順扎 *ʃunʤɑ ᠰᡠᠨᠵᠠ sunja tabun
6 寧古 *niŋɡu ᠨᡳᠩᡤᡠᠨ ninggun ǰirɣuɣan
7 or 納丹 *nɑdɑn ᠨᠠᡩᠠᠨ nadan doloɣan
8 or 扎困 *ʤɑkun ᠵᠠᡴᡡᠨ jakūn naiman
9 兀也温 *ujewun ᡠᠶᡠᠨ uyun yisün
10 *ʤuɑ ᠵᡠᠸᠠᠨ juwan arban
11 十一 or 安朔 *omʃo ᠣᠮᡧᠣᠨ omšon
12 十二 只兒歡 *ʤïrxon ᠵᠣᡵᡤᠣᠨ jorgon
13 十三 or 戈兒歡 *ɡorxon
14 十四 or 獨兒歡 *durxon
15 十五 脱卜歡 *toboxon ᡨᠣᡶᠣᡥᠣᠨ tofohon
16 十六 泥渾 *nixun ᠨᡳᠣᠯᡥᡠᠨ niolhun
17 十七 or 答兒歡 *dɑrxon
18 十八 or 女渾 *niuxun
19 十九 斡女歡 *onioxon
20 二十 or or 倭林 *orin ᠣᡵᡳᠨ orin qorin
30 三十 or 古申 *ɡuʃïn ᡤᡡᠰᡳᠨ gūsin ɣučin
40 四十 忒希 *dəçi ᡩᡝᡥᡳ dehi döčin
50 五十 or 速撒一 *susɑi ᠰᡠᠰᠠᡳ susai tabin
60 六十 寧住 *niŋʤu ᠨᡳᠨᠵᡠ ninju ǰiran
70 七十 納丹住 *nɑdɑnʤu ᠨᠠᡩᠠᠨᠵᡠ nadanju dalan
80 八十 扎困住 *ʤɑkunʤu ᠵᠠᡴᡡᠨᠵᡠ jakūnju nayan
90 九十 or 兀也温住 *ujewunʤu ᡠᠶᡠᠨᠵᡠ uyunju yeren
100 or 湯古 *tɑŋɡu ᡨᠠᠩᡤᡡ tanggū ǰaɣun
1000 or 皿干 *miŋɡɑn ᠮᡳᠩᡤᠠᠨ minggan mingɣan
10000 萬(万) or or 土满 *tumən ᡨᡠᠮᡝᠨ tumen tümen

† = only in calendrical usage

As with Khitan, Jurchen has special words for the decade numbers from 20 to 90. Of these, the numbers 60 through 90 are formed regularly from "unit + ten", 50 is irregular of unknown derivation, and 20 through 40 appear to be borrowings from Mongolian, possibly by way of Khitan.

Jurchen also has special words for 11 through 19 (only some of which survive in Manchu). These are of particular interest as the words for 12 through 18 are constructed from an element cognate to the root of the corresponding Mongolian unit number plus a "-teen" suffix -xon or -xun (perhaps related to Old Turkic on "ten"):

The anomalous forms for 12 and 16 are explained as being due to Mongolian qoyar "two" and ǰirɣuɣan "six" having replaced the original Proto-Mongolic forms of the numbers. Mongolian ǰirɣuɣan has been explained by Wilhelm Schott (1807–1889) as ǰir + ɣu[rban] + ɣan "two times three". On the basis of the Jurchen forms for 12 and 16, Proto-Mongolic 2 and 6 have been reconstructed as *ǰir and *nil repectively. The Jurchen forms for 2 through 18 have also been used to conjecturally reconstruct the Khitan numbers 2 through 8.

The glyph forms of Jurchen numbers are not obviously derived from either the large or small Khitan numbers, although, as with the Khitan Large Script, some of the glyph forms are based on the corresponding Chinese number characters. There are, however, some interesting similarities between Jurchen number characters and tablature signs used for writing lute music.

Lute Tablature

Detail of a lutenist in the 10th century painting "Night Revels of Han Xizai" 韓熙載夜宴圖 by Gu Hongzhong 顧閎中

Source : Zhongguo Meishu Quanji 中國美術全集 (繪畫編) vol.2 plate 65

The four-stringed Chinese lute (pipa 琵琶) was a very widely played musical instrument during the Tang dynasty (618–907) and the Song dynasty (960–1279), and several manuscripts using a special system of tablature signs for writing lute scores have survived in China and Japan:

Part of the Tang dynasty Dunhuang manuscript Pelliot chinois 3808 showing Lute Tunes 20–23 (in two different hands)

The four-stringed lute had four strings and four frets, allowing a total of twenty finger positions (open string and four fret positions for each string). There are twenty basic tablature signs in lute scores, each corresponding to one of the twenty possible finger positions. The absolute value of the notes represented by the tablature signs depends on the tuning of the lute required for the mode of the piece to be played. Different modes require different sets of notes, and so only a subset of the twenty tablature signs are ever used in any given tune.

Fret String I
(gōng 宮)
String II
(shāng 商)
String III
(jué 角)
String IV
( 羽)
(sàndǎ 散打)
Open String
(tóuzhǐ 頭指)
Index Finger
(zhōngzhǐ 中指)
Middle Finger
(míngzhǐ 名指)
Ring Finger
(xiǎozhǐ 小指)
Little Finger

The table below shows the twenty tablature signs arranged in Fret/String order.

Number Chinese Jurchen Lute Tablature
String/Fret P. 3808
P. 3808
P. 3537* P. 3719 Tempyō
Biwa Fu
Biwa Fu
Gogen Kinfu Japanese Reading†
1 I-0 ichi
2 II-0 otsu
3 III-0 gyō
4 IV-0
5 I-1 ku
6 II-1 ge
7 III-1 shichi
8 IV-1 hachi
9 I-2
10 II-2
11 十一 III-2 ni
12 十二 IV-2 boku 丿
13 十三 I-3 shū
14 十四 II-3 bi
15 十五 III-3 gon
16 十六 IV-3 sen
17 十七 I-4
18 十八 II-4 ko
19 十九 III-4 shi
20 二十 IV-4 ya

* Signs IV-2 (#12) and III-2 (#15) are swapped in P. 3537, but for ease of comparison they have been swapped back to their correct positions in this table.

† The Japanese readings are late reanalyses of the tablature sign derivations, and are provided for completeness only.

The first ten tablature signs in the above table (except for IV-0) are clearly derived from the Chinese characters for 1 through 10:

The fourth tablature sign is not derived from Chinese 四, as would be expected, but is similar to the Jurchen character for 14 , with which it may share a common origin.

As the first ten tablature signs correspond to the numbers one through ten, it is reasonable to assume that the last ten signs correspond to the numbers eleven through twenty. There are no single Chinese characters for the numbers 11 through 20, but there are single Jurchen characters for these numbers, and there are some interesting correspondences between the Jurchen numbers and tablature signs:

The correspondences are far from conclusive, and we have to assume that the assignments of lute tablature signs in extant manuscripts do not accurately reflect their original sequential order. Nevertheless, I think that the similarities are more than just coincidence, and that there must be a common origin for at least some of the lute tablature signs and Jurchen number characters.

The extant lute score manuscripts date to the Tang dynasty (618–907), but the Jurchen script was not invented until 1120, several hundred years later, so the lute tablature numbers were obviously not derived from the Jurchen numbers, but equally the Jurchen numbers are unlikely to have been derived directly from lute tablature signs. However, I think it possible that there is a common origin for lute tablature signs and the Jurchen script. The origins of the pipa lie in Central Asia, and the instrument was particularly associated with the peripheral areas of China and non-Han nationalities. It is therefore possible that lute notation was developed by non-Chinese speakers who used a script based on simplifications and cursive forms of Chinese characters for writing their own language. This may not have been a fully devloped writing system with a fixed orthography, but may have been more like the use of Man'yōgana to write Japanese, with multiple different Chinese characters being used to represent the same sound. As the pipa was mostly played by women, it is also a possibility that the source for lute tablature was a script used only by women, similar to the Nüshu script of southern China. This, now lost, urscript could have been ancestral to both the Khitan Large Script and the Jurchen script, which would explain why Jurchen has many characters that are similar to KLS characters, although Jurchen does not seem to have been directly derived from KLS. The poorly-documented Bohai script was posited by Janhunen in 1994 as an ancestor to the Khitan and Jurchen scripts, and although the evidence for this hypothesis is still lacking, I think that the lute tablature signs do add some support for the idea of a lost ancestral script for Khitan and Jurchen.


Index of Blog Posts