KANJAKI or Kančaki, language mentioned in the 11th-century Turkish lexicon of Maḥmud al-Kāšḡari (tr. Dankoff and Kelly, I, p. 84) as being spoken in the villages near Kāšḡar (see KASHGHAR). The underlying place name Kanjak or Kančak seems to derive from an old name of Kāšḡar, which is attested in the form Gahjag in Tibetan texts of about the 8th century (Emmerick, pp. 44-45, 70-71; for discussion of the Tibetan form and its relation to Kanjak, see Pelliot, pp. 210-11). The words cited by Kāšḡari show that Kanjaki was essentially a Turkic dialect (Tremblay, p. 74), though it also contained non-Turkic elements, many of which probably derive from the language spoken in the region in earlier times. Following H. W. Bailey (1958, p. 131; 1985, pp. 50-54), X. Tremblay identifies this substrate language as Iranian, possibly a “Saka” language similar to Khotanese, supporting his view with a detailed etymological study of the words identified as Kanjaki by Kāšḡari, as well as of names and titles attested in Chinese and other sources and attributable to the pre-Turkic language of Kāšḡar.

From a historical and geographical point of view, it is quite plausible that an Iranian language closely connected with Khotanese and Tumshuqese was formerly spoken in Kāšḡar. However, many of the etymologies proposed in support of this view are highly speculative. Kanjaki or Kāšḡari words with clear cognates in other Iranian languages mostly belong to categories of words which are easily borrowed, such as terms for vessels (e.g., kändük “flour jar” = Middle Persian kandūg, Persian kandu) and names of plants (e.g., känbä = Khotanese kuṃbā-, Sogdian kynpʾ “flax”; cf. Akkadian qunnapu, Syriac qnpʾ, Greek kánnabis). One of the few forms which may display a specifically ‘Saka’ character is ʾAlyoṅj´a (variant ʾAlyohjah), which is given in a Tibetan text as the name of a queen of Gahjag who became the consort of a king of Khotan (Emmerick, pp. 70-71). Bailey (1958, p. 131) compared this with Khotanese alsāṃgyā- “girl, lady,” feminine to alysānaa- “boy, prince.” However, even if this etymology is accepted (despite doubts concerning Tibetan -ly- as equivalent of Khotanese -lys- [lz]; see Emmerick, p. 107), it seems quite possible that ʾalyoṅj´a “princess” is not really the queen’s personal name but a Khotanese title which she received as a result of her marriage to the king of Khotan. Overall, the material does not seem to justify more than the cautious conclusion (Sims-Williams, p. 166) that the language contains an Iranian element.


H. W. Bailey, “Languages of the Saka,” in HO I/IV/1, Leiden, 1958, pp. 131-54.

Idem, Khotanese Texts VII, Cambridge, 1985.

R. E. Emmerick, Tibetan Texts concerning Khotan, London, 1967.

Maḥmud al-Kāšḡari, Compendium of the Turkic Dialects (Dīwān Luγāt at-Turk), ed. and tr. R. Dankoff and J. Kelly, 3 vols., Harvard, 1982-85.

P. Pelliot, Notes on Marco Polo I, Paris, 1959.

N. Sims-Williams, “Eastern Middle Iranian,” in Compendium Linguarum Iranicarum, ed. R. Schmitt, Wiesbaden, 1989, pp. 165-72.

X. Tremblay, “Kanjakī and Kāšγarian Sakan. Contributions towards a Comparative Grammar of Iranian Languages, XI,” Central Asiatic Journal, 51/1, 2007, pp. 63-76.

(Nicholas Sims-Williams)

Originally Published: December 15, 2010

Last Updated: April 20, 2012

This article is available in print.
Vol. XV, Fasc. 5, p. 502