A language you could read in a morning

Claim (verbatim)

Beside Khotanese there was a sister Saka language, Tumshuqese, spoken around Tumshuq (Toqquz-Sarai) on the northern rim of the Tarim — and its entire surviving textual record is a small body of documents and religious fragments, few enough that a scholar can hold the whole corpus in view at once. It is a complete written language reduced to a quantity of text smaller than a single chapter of many of its neighbours' works: some administrative documents, a Buddhist ritual text or two, and the loanwords and names by which it also shows up in other people's writing. The mechanism is that a small oasis with a short literate life and no institution large enough to mass-copy anything leaves a corpus governed by pure chance of burial, so the language survives at the resolution of the accidental — present as a handful of physical objects, absent as a literature. Tumshuqese is the boundary case that calibrates the whole region's loss: it shows what survival looks like at the thin end, where the difference between a documented and an undocumented language is a few square metres of dry ground. Prediction: the complete published Tumshuqese corpus will comprise on the order of a few dozen texts and only a few thousand running words, with no single literary work preserved in full, and its ratio of loanword-and-name attestations (in Khotanese, Sanskrit, and Chinese contexts) to free-standing Tumshuqese texts will be high — the profile of a language better attested by mention than by use (primary clause: the corpus falling on the order of dozens of texts with no complete literary work; a substantial free-standing Tumshuqese literary corpus falsifies it). Kill: the Tumshuqese texts in the editions of Sten Konow and in Ronald Emmerick and P. O. Skjaervo, Studies in the Vocabulary of Khotanese, with H. W. Bailey's Khotanese and Tumshuqese materials as the corpus census.

Prediction clause (verbatim)

Prediction: the complete published Tumshuqese corpus will comprise on the order of a few dozen texts and only a few thousand running words, with no single literary work preserved in full, and its ratio of loanword-and-name attestations (in Khotanese, Sanskrit, and Chinese contexts) to free-standing Tumshuqese texts will be high — the profile of a language better attested by mention than by use (primary clause: the corpus falling on the order of dozens of texts with no complete literary work; a substantial free-standing Tumshuqese literary corpus falsifies it).

Kill-dataset (verbatim)

Kill: the Tumshuqese texts in the editions of Sten Konow and in Ronald Emmerick and P. O. Skjaervo, Studies in the Vocabulary of Khotanese, with H. W. Bailey's Khotanese and Tumshuqese materials as the corpus census.

Novelty / leakage triage

already answered in the literature

Tumshuqese is a textbook boundary case: the whole corpus is a small, effectively enumerable body of administrative documents and a couple of Buddhist and ritual fragments, with no complete literary work, as set out in Konow's editions, Emmerick and Skjaervo's vocabulary studies, and Bailey's materials. The primary clause - a corpus on the order of dozens of texts with no complete literary work - is thus already the stated size and shape of the record, not an open count. Only the finer mention-to-use ratio the prediction adds would be new arithmetic.

• Sten Konow, 'Ein neuer Saka-Dialekt' (Sitzungsberichte der Preussischen Akademie der Wissenschaften, 1935)
• Ronald E. Emmerick & P. O. Skjaervo, Studies in the Vocabulary of Khotanese, 3 vols. (Vienna: Verlag der OeAW, 1982-1997)
• H. W. Bailey, Dictionary of Khotan Saka (Cambridge University Press, 1979)

Predictions

No prediction registered yet.