| Tangut | |
|---|---|
| Xi-Xia | |
| 𗼇𗟲 | |
 Buddhist scripture written in Tangut | |
| Native to | Western Xia |
| Ethnicity | Tangut people |
| Era | AD 1036–1502 (attested) |
| Tangut script | |
| Official status | |
Official language in | Western Xia |
| Language codes | |
| ISO 639-3 | txg |
txg | |
| Glottolog | tang1334 |
Tangut (Tangut:𗼇𗟲;Chinese:西夏語;pinyin:Xīxià yǔ;lit. 'Western Xia language') is an extinct Sino‑Tibetan language, now argued to belong within the Horpa subgroup of West Gyalrongic.[1][2]
Tangut was one of the official languages of theWestern Xia dynasty, founded by theTangut people in northwestern China. The Western Xia was annihilated by theMongol Empire in 1227.[3] The Tangut language has its own script, theTangut script. The latest known text written in the Tangut language, theTangut dharani pillars, dates to 1502,[4] suggesting that the language was still in use nearly three hundred years after the collapse of Western Xia.
Since the 2010s,Tangutologists have commonly classified Tangut as aQiangic orGyalrongic language.[5][6] On the basis of both morphological and lexical evidence, Lai et al. (2020) classify Tangut as aWest Gyalrongic language.[7]
Beaudouin (2023a,b) showed that Tangut was aHorpa language, a subgroup of West Gyalrongic. He hypothesizes a position between the Northern (Stodsde) and Central (Stau, Geshiza) lects, proposing a tentativeUrheimat around the place where the Erkai variety is spoken today within Ngawa, Sichuan.[8][9]
Modern research into the Tangut languages began in the late 19th century and early 20th century whenS. W. Bushell,Gabriel Devéria, andGeorges Morisse separately published decipherments of a number of Tangut characters found onWestern Xia coins, in a Chinese–Tangut bilingual inscription on a stele atWuwei, Gansu, and in a copy of the Tangut translation of theLotus Sutra.
The majority of extant Tangut texts were excavated atKhara-Khoto in 1909 byPyotr Kozlov, and the script was identified as that of the Tangut state of Xixia. Such scholars asAleksei Ivanovich Ivanov,Ishihama Juntaro (石濱純太郎),Berthold Laufer,Luo Fuchang (羅福萇),Luo Fucheng (羅福成), andWang Jingru (王靜如) have contributed to research on the Tangut language. The most significant contribution was made by the Russian scholarNikolai Aleksandrovich Nevsky (1892–1937), who compiled the first Tangut dictionary and reconstructed the meaning of a number of Tangut grammatical particles, thus making it possible to actually read and understand Tangut texts. His scholarly achievements were published posthumously in 1960 under the titleTangutskaya Filologiya (Tangut Philology), and the scholar was eventually (and posthumously) awarded the SovietLenin Prize for his work. The understanding of the Tangut language is far from perfect: although certain aspects of the morphology (Ksenia Kepping,The Morphology of the Tangut Language, Moscow: Nauka, 1985) and grammar (Tatsuo Nishida,Seika go no kenkyū, etc.) are understood, the syntactic structure of Tangut remains largely unexplored.
The Khara-Khoto documents are at present preserved in theInstitute of Oriental Manuscripts of the Russian Academy of Sciences inSaint Petersburg. These survived theSiege of Leningrad, but a number of manuscripts in the possession of Nevsky at the time of his arrest by thePeople's Commissariat for Internal Affairs (NKVD) in 1937 went missing, and were returned, under mysterious circumstances, to the Institute of Oriental Manuscripts only in October 1991.[10] The collections amount to about 10,000 volumes, of mostly Buddhist texts, law codes, and legal documents dating from mid-11th up to early 13th centuries. Among the Buddhist texts, a number of unique compilations, not known either in Chinese or in Tibetan versions, were recently discovered. Furthermore, the Buddhist canon, theChinese classics, and a great number of indigenous texts written in Tangut have been preserved. These other major Tangut collections, though much smaller, belong to theBritish Library, the French National Library ('Bibliothèque nationale de France'), the National Library in Beijing, the Library of Beijing University, and other libraries.
The connection between the writing and the pronunciation of the Tangut language is even more tenuous than that between Chinese writing and the modernChinese varieties. Thus although in Chinese more than 90% of the characters possess a phonetic element, this proportion is limited to about 10% in Tangut according to Sofronov. The reconstruction of Tangut pronunciation must resort to other sources.
The discovery of thePearl in the Palm, a Tangut–Chinese bilingual glossary, permitted Ivanov (1909) and Laufer (1916) to propose initial reconstructions and to undertake the comparative study of Tangut. This glossary in effect indicates the pronunciation of each Tangut character with one or several Chinese characters, and inversely each Chinese character with one or more Tangut characters. The second source is the corpus of Tibetan transcriptions of Tangut. These data were studied for the first time by Nevsky (Nevskij) (1925). Though these transcriptions were not written with the intention of representing with precision the pronunciation of Tangut, but instead simply to help foreigners to pronounce and memorize the words of one language with the words of another which they could understand.
The third source, which constitutes the basis of the modern reconstructions, consists of monolingual Tangut dictionaries: theWenhai (文海), two editions of theTongyin (同音), theWenhai zalei (文海雜類), and an untitled dictionary. The record of the pronunciation in these dictionaries is made using the principle offǎnqiè, borrowed from the Chinese lexicographic tradition. Although these dictionaries may differ on small details (e.g. theTongyin categorizes the characters according to syllable initial andrime without taking any account of tone), they all adopt the same system of 105 rimes. A certain number of rimes are in complementary distribution with respect to the place of articulation of the initials, e.g. rimes 10 and 11 or rimes 36 and 37.Fǎnqiè makes distinctions among the rhymes in a systematic and precise manner. Nonetheless, it is still necessary to compare the phonological system of the dictionaries with the other sources in order to "fill in" the categories with a phonetic value.
N. A. Nevsky reconstructed Tangut grammar and provided the first Tangut–Chinese–English–Russian dictionary, which together with the collection of his papers was published posthumously in 1960 under the titleTangut Philology (Moscow: 1960). Later, substantial contribution to the research of Tangut language was done byTatsuo Nishida (西田龍雄),Ksenia Kepping,Gong Hwang-cherng (龔煌城), M.V. Sofronov, andLi Fanwen (李範文).Marc Miyake has published on Tangut phonology and diachronics.[11] There are four Tangut dictionaries available: the one composed by N.A. Nevsky, one composed by Nishida (1966), one composed by Li Fanwen (1997, revised edition 2008), and one composed byYevgeny Kychanov (2006).
Modern refinements to Tangut reconstruction leverage a new type of data that was unavailable to previous scholars: the phonology of modern Gyalrongic languages.
The Tangut syllable has a CV structure and carries one of two distinctive tones, flat (平聲) or rising (上聲).Gong, X. (2025) recostructed the flat tone as a high-falling tone /HL/ and the rising tone as a mid-level tone /M/[12]. The paradoxical tone names and their respective contour are thought to have been the result of a tonal reversal process, and cognates between Tangut and otherQiangic languages are observed to correspond quite symmetrically, with the Tangut syllables usually in the opposite pitch of those in the Qiangic languages. Following the tradition of Chinese phonological analysis the Tangut syllable is divided into initial (聲母) and rhyme (韻母) (i.e. the remaining syllable minus the initial).
The consonants are divided into the following categories:
| Chinese term | Translation | Modern term | Arakawa | Gong | Miyake |
|---|---|---|---|---|---|
| 重唇音類 | heavy lip | bilabials | p, ph, b, m | p, ph, b, m | p, ph, b, m |
| 輕唇音類 | light lip | labio-dentals | f, v, w | v | |
| 舌頭音類 | tongue tip | apicals (dentals) | t, th, d, n | t, th, d, n | t, th, d, n |
| 舌上音類 | tongue surface | laminals (alveolars) | ty', thy', dy', ny' | tʂ tʂh dʐ ɳ | |
| 牙音類 | molar | velars | k, kh, g, ng | k, kh, g, ŋ | k, kh, g, ŋ |
| 齒頭音類 | incisor tip | dental affricates and fricatives | ts, tsh, dz, s | ts, tsh, dz, s | ts, tsh, dz, s |
| 正齒音類 | incisor proper | palatal affricates and fricatives | c, ch, j, sh | tɕ, tɕh, dʑ, ɕ | |
| 喉音類 | throat | laryngeals | ', h | ., x, ɣ | ʔ, x, ɣ |
| 流風音類 | flowing air | resonants | l, lh, ld, z, r, zz | l, lh, z, r, ʑ | ɫ, ɬ, z, ɽr, r |
The rhyme books distinguish 105 rhyme classes, which are, in turn, classified in several ways: division/grade (等), type (環), and class (攝).
Tangut rhymes occur in three types (環). They are seen in the tradition of Nishida, followed by both Arakawa and Gong as 'normal' (普通母音), 'tense' (緊喉母音), and 'retroflex' (捲舌母音). Gong leaves normal vowels unmarked and places a dot under tense vowels and an -r after retroflex vowels. Arakawa differs only by indicating tense vowels with a final -q.
The rhyme books distinguish four vowel grades (等). In early phonetic reconstructions, all four were separately accounted for, but it has since been realized that grades three and four are in complementary distribution, depending on the initial. Consequently, the reconstructions of Arakawa and Gong do not account for this distinction. Gong represents these three grades as V, iV, and jV. Arakawa accounts for them as V, iV, and V.
In general, rhyme class (攝) corresponds to the set of all rhymes under the same rhyme type which have the same main vowel.
Gong further posits phonemic vowel length and points to evidence that indicates that Tangut had a distinction that Chinese lacked. There is no certainty that the distinction was vowel length and so other researchers have remained skeptical.
 | This articleshould specify the language of its non-English content using{{lang}} or{{langx}},{{transliteration}} for transliterated languages, and{{IPA}} for phonetic transcriptions, with an appropriateISO 639 code. Wikipedia'smultilingual support templates may also be used.See why.(June 2022) |
| Normal (普通母音) | Tense (緊喉母音) | Retroflex (捲舌母音) | |
|---|---|---|---|
| close | i I u | iq eq uq | ir Ir ur |
| mid | e o | eq2 oq | er or |
| open | a | aq | ar |
Miyake reconstructs the vowels differently. In his reconstruction, the 95 vowels of Tangut formed from a six-vowel system in Pre-Tangut because of preinitial loss. (The two vowels in parentheses appeared only in loanwords from Chinese, and many of the vowels in class III were in complementary distribution with their equivalents in class IV.)
| Pre-Tangut vowel | Class 1 | Class 2 | Class 3 | Class 4 |
|---|---|---|---|---|
| *u | əu | o | ɨu | iu |
| əəu | oo | ɨuu | iuu | |
| (əũ) | ||||
| əụ | ɨụ | iụ | ||
| əuʳ | iuʳ | |||
| *i | əi | ɪ | ɨi | i |
| əəi | ɪɪ | ɨii | ii | |
| əĩ | ɨĩ | ĩ | ||
| əị | ɨị | ị | ||
| əiʳ | ɪʳ | ɨiʳ | iʳ | |
| əəiʳ | ɪɪʳ | ɨiiʳ | iiʳ | |
| *a | a | æ | ɨa | ia |
| aa | ææ | ɨaa | iaa | |
| ã | æ̃ | ɨã | iã | |
| ạ | ɨạ | iạ | ||
| aʳ | æʳ | ɨaʳ | iaʳ | |
| aaʳ | ɨaaʳ | iaaʳ | ||
| (ya) | ||||
| *ə | ə | ʌ | ɨə | iə |
| əə | ɨəə | iəə | ||
| ə̣ | ɨə̣ | iə̣ | ||
| əʳ | ʌʳ | ɨəʳ | iəʳ | |
| ɨəəʳ | iəəʳ | |||
| *e | e | ɛ | ɨe | i.e. |
| ee | ɛ | ɨee | iee | |
| ẽ | ɛ̃ | ɨẽ | iẽ | |
| ɛ̣̃ | ɨẹ̃ | iẹ̃ | ||
| ɛ̣ | ɨẹ | iẹ | ||
| eʳ | ɛʳ | ɨeʳ | ieʳ | |
| *ik *ek *uk | ew | ɛw | ɨew | iew |
| ɨiw | iw | |||
| eʳw | i(e)ʳw | |||
| *o | o | ɔ | ɨo | io |
| wɨo | ||||
| oo | ɔɔ | ɨoo | ioo | |
| õ | ɔ̃ | ɨõ | iõ | |
| ɔ̃ɔ̃ | ɨõõ | iõõ | ||
| ọ | ɔ̣ | ɨọ | iọ | |
| oʳ | ɔʳ | ɨoʳ | ioʳ | |
| ooʳ | iooʳ | |||
| õʳ | iõʳ |
The classes here are related to those of Chineserime tables.
Tangut clause syntax prefers thesubject–object–verb order. Like Chinese, the Tangut NP places numeral and classifier before the noun.
Like other Gyalrongic languages, Tangut verbs are highly synthetic with many different morphological slots. The general verb template is shown in the table below:
| +6 | +5 | +4 | +3 | +2 | +1 | core | -1 | -2 | -3 | -4 | |
|---|---|---|---|---|---|---|---|---|---|---|---|
| ja- | TAM/ORIENT | mood | NEG | modal | valency | Incorporated noun | verb stem | person | -jij¹ (Other suffixes) | -sji² | -djij² |
𗥑𗭴
xu1·jow1
Fu.Rong
𗤄𘒣
·jɨr1dạ2
ask
𗥑𗭴𗤄𘒣𗍳𘊐𗉘𘈞𘉞𘘣
xu1·jow1 ·jɨr1dạ2 nji2kjij1-tśhjɨ1-mjịj1-nja2 ·jɨ2
Fu.Rong ask 2.HON Q.PFV:IN-EXP-dream-2SG QUOT
Fu Rong asked, "did you dream?" (Leilin, 06.16B.4)
In Tangut texts, only few instances of syntacticnoun incorporation are attested: the head is final, since it doesn’t move, the directional marker serves as adverb;transitive verbs can absorb the object, but not the subject. In otherQiangic languages that possess high levels of pronominalization such asJaphug andKhroskyabs, NI is still a more syntactically productive process with widespread uses.[13]
𘈷𗗙𘙌𗦳𗠔𘉞
gji2=jij1 kjɨ1-dzju2-phjo2-nja2
son=ANTIERG PFV.IN-lord-cause-2SG.P
You made your son lord (of Zhongshan). (Leilin, 03.10.B.4)
𗒯
khji1
Ji
𘄽𘄽
ŋạ2.ŋạ2
very
𗒯𘄽𘄽𗞞𗤶𘅎
khji1 ŋạ2.ŋạ2 dja2-njɨ̣j1-ljɨ̣j2
Ji very PFV-heart-be.happy
Ji rejoiced with all his heart. (Leilin, 07.12.B.4)
Like other Gyalrongic languages, agreement in Tangut is sensitive to both the subject and object.
In Tangut, two parts of the verb are sensitive to agreement, the person suffix (slot -1) and the verb stem itself (verbal core).[14] For intransitive verbs, only the person suffix is relevant where it agrees with the subject of the verb. As for transitive verbs, verbs generally agree with the absolutive argument except if the absolutive argument is 3rd person and the ergative is 1st or 2nd person. In these situations, the suffix instead agrees with the ergative argument.
| Subject or Agent | Patient | Intransitive | ||||||
|---|---|---|---|---|---|---|---|---|
| 1sg | 1du | 1pl | 2sg | 2du | 2pl | 3 | ||
| 1sg | Σ-𘉞 Σ-nja2 | Σ-𘂆 Σ-tsjɨ1 | Σ-𗐱 Σ-nji2 | Σ-𗧓 Σ-ŋa2 | Σ-𗧓 Σ-ŋa2 | |||
| 1du | Σ-𘙌 Σ-kjɨ1 | Σ-𘙌 Σ-kjɨ1 | ||||||
| 1pl | Σ-𗐱 Σ-nji2 | Σ-𗐱 Σ-nji2 | ||||||
| 2sg | Σ-𗧓 Σ-ŋa2 | Σ-𘙌 Σ-kjɨ1 | Σ-𗐱 Σ-nji2 | Σ-𘉞 Σ-nja2 | Σ-𘉞 Σ-nja2 | |||
| 2du | Σ-𘂆 Σ-tsjɨ1 | Σ-𘂆 Σ-tsjɨ1 | ||||||
| 2pl | Σ-𗐱 Σ-nji2 | Σ-𗐱 Σ-nji2 | ||||||
| 3 | Σ-𘉞 Σ-nja2 | Σ-𘂆 Σ-tsjɨ1 | Σ-𗐱 Σ-nji2 | Σ | Σ | |||
Cells coloured in green not only involve the person suffix but also involve alternations of the stems from the basic stem A to stem B. This stem alternation pattern originates from a 3rd person object suffix of the form*-w as is also found in otherSino-Tibetan languages.[15] In general, stem alternation involves changing the vowel of the stem in a pattern shown as below.[16]
| Stem A | Stem B | Example |
|---|---|---|
| -i/e | -o | 𗡅dzji1 →𗠈dzjo1 "eat" |
| -u | -o | 𗕼lju2 →𗬘ljo2 "to throw" |
| -ej/ij | -o | 𗿷dźjij2 →𗲉dźjo2 "to possess" |
| -ej/ij | -i/e | 𘟀ljij2 →𗐵lji2 "to see" |
| -a | -ɨ/ə | 𗴒kjạ1 →𗕐kjɨ̣1 "to fear" |
Some verbs display an alternation unrelated to person, encoding a non-past/past distinction. The non-past stems can be used in non-finite contexts as action nominals, functional infinitives, or in clause chaining.[17][2] The past stem is always prefixed by a perfective preverb, except for "to come", and post-verbal occurrences of "to go", which may indicate a different part of speech.
| Stem 1 (non-past) | Stem 2 (past) | Meaning |
|---|---|---|
| 𗶷śjɨ1 | 𗶹śji2 | "to ɡo" |
| 𗄼lja1 | 𗆐ljịj2 | "to come" |
| 𗈶sjɨ1 | 𗢏/𗏋sji2 | "to die" |
| 𘐩phjɨ1 | 𘜉phji2 | "to abandon, to lose" |
Sino-Tibetan branches | |||||
|---|---|---|---|---|---|
| WesternHimalayas (Himachal, Uttarakhand,Nepal,Sikkim) |
|  | |||
| EasternHimalayas (Tibet,Bhutan,Arunachal) | |||||
| Myanmar and Indo- Burmese border |
| ||||
| East andSoutheast Asia |
| ||||
| Dubious (possible isolates,Arunachal) |
| ||||
| Proposed groupings | |||||
| Proto-languages | |||||
Italics indicates single languages that are also considered to be separate branches. | |||||