¹ My field work in spoken Tibetan, in the course of which I collected most of the data here presented, was carried out in Darjeeling, West Bengal, from August 1953 to March 1954. It was made possible by a grant from the Board on Overseas Training and Research of the Ford Foundation.

² See for example Sir C. A. Bell, Grammar of colloquial Tibetan ³ 15 (Calcutta, 1939): ‘There are also other irregular sounds, but it seems unnecessary to enumerate them all here. They will be picked up more easily in the course of conversation.’ In P. M. Miller's treatment they are to some extent recognized (see note 3); but his ‘prenasalization’ analysis, for which there is no formal basis, artificially segments the materials, and prevents him from facing the problem.

³ Abbreviations of languages cited: CT = Central Tibetan, WT = Written Tibetan, Lh. = Lhasa dialect, Ld. = Ladakh dialect; Ar. Ch. = Archaic Chinese, An. Ch. = Ancient Chinese; WB = Written Burmese, Bu. = spoken Burmese, Rangoon dialect. All citations in this paragraph are CT; below, citations are identified. The CT, Lh., and Ld. materials are from my field notes (see fn. 1); the transcriptions are phonemic.

Symbols employed have their usual values, but the following may be noted. Acute = high level tone, grave = low or in certain environments low-mid-low tone; ṭ, ṭh, ḏ = palatal stops. Final stops are unexploded; in CT and Lh. -b and -g are [p] and [k] with high level, [b] and [g] with low; in Ld., -b -d -g are unexploded [p t k]; all these stops have homorganic voiceless spirantal releases before pause, ö is mid-front rounded, ü is high-front rounded.

My main informant for Lh. was a former Lhasa lay-official now residing in Darjeeling and Kalimpong; his speech agrees well with the dialect recorded by Chao (Y. R. Chao and Yu Dawchyuan, Love songs of the sixth Dalai lama = Academia sinica monographs, Series A, No. 5; Peiping, 1930). My informant for Ld. was an educated layman from Leh. My main informant for CT was a lama residing in Darjeeling. His dialect closely resembles that described by P. M. Miller, The phonemes of Tibetan (U-Tsang dialect) with a practical romanized orthography for Tibetan-speaking readers, Journal of the Asiatic Society: Letters, 17:3.191-216 (1951), at least as far as the phonetic data are concerned; but Miller's analysis, of which I find important parts unacceptable, leaves him with a phonemic transcription differing considerably from that which I employ here. (Georges de Roerich, Modern Tibetan phonetics, with special reference to the dialect of Central Tibet, JASB NS 27.285-312 [1931], is more an attempt to explain the writing system than to describe a dialect; as far as it goes, it deals with a dialect not unlike our CT.) WT is a literal transcription of the conventional Tibetan orthography, but distinguishing subscript i̭, here written j, from prefix plus i̭ or initial i̭, here written y, on the basis of what was evidently a phonemic distinction in the dialect reflected in the WT orthography. Ar. Ch. and An. Ch. are reconstructed by B. Karlgren, Études sur la phonologie chinoise = Archives d'études orientales, Vols. 12, 13, 19, 24 (1915-26), and conveniently available in his Analytic dictionary of Chinese and Sino-Japanese (Paris, 1923); for Ar. Ch., see his Grammata serica = BMFEA 12.1-471 (1940). WB is a literal transcription of the orthography; Bu. follows W. Cornyn, Outline of Burmese grammar (Language dissertation 38; Baltimore, 1944), with a few changes in symbols but not in analysis.

⁴ -džù < tšhú- ‘beak’; cf. tšhútó ‘lip’. The morphophonemic change here, though sporadic in its operation in CT, is otherwise identical with that described for Bu. by Mary R. Haas, The use of numeral classifiers in Burmese, Semitic and oriental studies presented to William Popper 11.191-200 (University of California publications in Semitic philology, 1951).

⁵ A small subclass of verbs shares this formal mark with these nouns.

⁶ For bibliography of these problems see S. N. Wolfenden, Outlines of Tibeto-Burman linguistic morphology 203-14 (London, 1929); R. A. D. Forrest, The Chinese language (London, 1948).

⁷ For this transcription, and those below, see my remarks in Wennti 5.14-6 (1953).

⁸ I.e. final /d/ and /g/, perhaps better interpreted phonetically as spirants; cf. my discussion, A note to Karlgren's Phonologie, The ΦΘ annual: Papers of the University of California oriental languages honor society, Vol. 3 (Berkeley, 1952).

⁹ The complete statement for this allomorph in the forties of the Ld. number system is as follows (numbers give the expression where the particular allomorph is found): ža- 45, 46 ~ žib- 40 ~ žag- 41, 42, 43 ~ žab- 44, 48 ~ žeb- 47 ~ žer- 49. In CT we find: šè- 41, 42, 43, 44, 45, 46 ~ šìb- 40 ~ šèb- 47, 48 ~ šèr- 49. Lh. offers no materials here.

¹⁰ The complete statement for this allomorph in the nineties of the Ld. number system is as follows: ko- 95, 96 ~ kub- 90 ~ kob- 94, 97, 98 ~ kog- 91, 92, 93 ~ kor- 99. In CT the morph appears in different shapes: gó- 91-99 ~ gùb- 90, not useful here.

¹¹ Again, perhaps better [-b]; cf. fn. 8.

¹² The complete statement for ‘ten’ in the teens of the number system for CT, Lh., and Ld. is as follows: CT tšúg- 11, 13, 16 ~ tšúŋ 12 ~ tšúb- 14, 17 ~ tšúr- 19 ~ tšób-18 ~ tšō- 15; Lh. tšúg- 11, 12 ~ tšóg- 13 ~ tšúb- 14, 17 ~ tšób- 18 ~ tšú- 16, 19 ~ tšṓ-15; Ld. tšug-11, 12, 13 ~ tšub - 14, 17, 18 ~ tšur- 19 ~ tšu-16 ~ tšo- 15.

¹³ This is not to say, of course, that such morpheme variants are not of old standing in the Tibetan dialects. The 9th-century inscriptions clearly show them in operation even at that date; cf. Paul Pelliot, Quelques transcriptions chinoises de noms tibétains, TP 16.1-26 (1915), and see P. Demiéville in Pelliot's posthumous work, Les débuts de l'imprimerie en Chine 136-7 (Paris, 1953).