^{page 116 note 2} Love songs of the sixth Dalai Lama Tshangs-dbyangs=rgya-mtsho, translated into Chinese and English with notes and introduction by Dawchyuan, Yu, and transcribed by DrYuanrenn, Jaw (Y. R. Chao) (Academia Sinica, Series A, No. 5, Peiping, 1930), 8–9Google Scholar. I am indebted to my colleague Mr. H. Simon for a translation of the relevant passages.

^{page 116 note 3} R. was born and educated in Lhasa. He was employed as Research Assistant by the School of Oriental and African Studies for the period December 1948–September 1949, in London. The material obtained from R. was checked against the utterances of other speakers of the Lhasa dialect in Kalimpong (West Bengal) and in Gyantse (Tsang Province, Tibet) during the session 1949–50.

^{page 117 note 1} For ‘prosodic system’ see Firth, J. R., ‘Sounds and prosodies’, TPS, 1948, 127–52Google Scholar. Subsequent publications that distinguish prosodic and phonematic categories are listed at BSOAS, XIII, 4, 1951, 945Google Scholar, and BSOAS, XVII, 1, 1955, 134Google Scholar.

^{page 117 note 2} See also Palmer, , ‘“Openness”’, 561Google Scholar, and Carnochan, J., ‘Vowel harmony in Igbo’, African language studies, I, 1960, 156Google Scholar.

^{page 117 note 3} When used as phonological terms Syllable, Word, Piece, Closure, etc., are distinguished by capital letters.

^{page 117 note 4} Palmer, , ‘“Openness”’, 577Google Scholar.

^{page 118 note 1} Since the Piece is either coextensive with, or contained in, the Word, it follows that more than one grammatical category can be exemplified in the Word as well as the Piece: the examples given at (b), (c), and (e) below can also serve as examples of Noun- + -Particle, Adjective-+-Particle, and Verb-+-Particle, Words. The examples bzhes-kyi-yod-pas and rgyab-kyi-yin-pas at (d) are also examples of the Verb-+-Particle Word. The examples at (a), on the other hand, exemplify a single grammatical category, the Noun, and are termed Noun Words; the second Word of the examples at (e, ii), gnang, exemplifies the Verb category only, and is termed a Verb Word.

^{page 118 note 2} It has been assumed that it is legitimate to treat a given phonetic feature as available for statement as an exponent of more than one term, whether prosodic or phonematic. In this particular instance it is the feature half-openness (ε) that is shared by the V term of the Syllable skad with the exponent of the Open term of the Closure system, which embraces vowel features of both Syllables of the Piece (ε — ә). For a discussion of whether such an overlap is permissible, see Palmer, , ‘“Openness”’, 576–7Google Scholar.

^{page 119 note 1} The phonetic transcriptions, in the International Phonetic Alphabet but with the addition of (non-velarity), show no more detail than is relevant to this account of vowel harmony; pitch features, therefore, have not been symbolized.

^{page 119 note 2} The forms in round brackets are phonetic spellings, put immediately after the regular orthographic form with which they alternate.

^{page 119 note 3} The spaces between words, for which there is no warrant in Tibetan orthography, reflect the delimitation of words on formal (phonological and grammatical) grounds.

^{page 119 note 4} Referring to Syllables by their phonological formulae rather than by their orthographic forms, though theoretically preferable, would in practice result in unjustified complications, and would pose more questions than it answered, as may be seen from a comparison of gnang, the orthographic form of one of the Verb Syllables, with its phonological formula ^lŋәvNV, in which the indices indicate the prosodic classification of this syllable as Tone One (of a two-term Tone system, referable to the Word), as ŋ (of an eight-term Quality system, referable to a monosyllabic or disyllabic Piece), as ә (of a three-term Labiality system, referable to a monosyllabic Piece), and as v (of a three-term Glottality system, referable to a monosyllabic Piece), while N specifies the N term of a C phonematic system, and V the sole member of a V system.

^{page 120 note 1} Prosodic analysis enables two or more phonetic forms, which may or may not be differently symbolized in the orthography, to be associated with a single phonological structure, as exponents of units of that structure under differing prosodic conditions. Thus, the phonetic forms be:, ge:, ŋe:, and Ie: can all be stated as exponents of the C and the V terms of the Interrogative-Particle Syllable variously symbolized in Tibetan orthography as pas, gas, ngas, or ras, in one or other of eight prosodically differing types of disyllabic Piece covering not only the relevant features of this Particle Syllable, but also associated features of the preceding (Verb) Syllable. Similar prosodic statements provide grounds for associating various phonetic forms with the single phonological structures symbolized as ma/mi (Negative Particle), pa/ba/ga/nga/ra (Past Particle, Nominalizing Particle), and gi/gyi/kyi.

^{page 120 note 2} The term closeness is used for a tongue position higher than half-close; it therefore includes ι and ω.

^{page 120 note 3} Bell, C. A., Grammar of colloquial Tibetan, Alipore, 1939, 136Google Scholar.

^{page 121 note 1} ibid., 165.

^{page 121 note 2} Based on SirGould, Basil and Richardson, Hugh Edward, Tibetan sentences, O.U.P., 1943, 11Google Scholar.

^{page 121 note 3} ibid., 102.

^{page 121 note 4} Gould, and Richardson, , Tibetan language records, etc., Kalimpong, 1949, C 29Google Scholar.

^{page 122 note 1} For a corresponding alternation in vowel duration to the e(:) and ι of the V term of red, with or without a correlated alternation in vowel quality, cf. du(:), 'dug, and dωge:, 'dug-gas (U:/ω); jø(:), yod, and jøβε:, yod-pas (ø:/ø); mε:, med(mad), and mεβε, med(mad)-pas (ε:/ε), all of them examples of the Complement sub-category of Verb (red, 'dug, yod, med(mad), yin, min(man), byung, yong).

^{page 122 note 2} The difference in exponency of the V term of ma/mi in Words in which the Verb Complement is represented by yong and red is associated with the presence of initial palatality (-εj-) and non-palatality (-ai-) in the Verb Syllable, and would require a further prosodic statement.

^{page 123 note 1} Bell, , Grammar, 53Google Scholar.

^{page 123 note 2} With the possible exception of the Nominalizing Particle dwogs, or rdog, e.g. yin-dwogs kha-po red ‘he may perhaps be’, 'gro-rdog kha-po red ‘I might perhaps have to go’ (Gould, and Richardson, , Records, C 42)Google Scholar, and the Perfective Particle med(mad), e.g. (oP) 'byor-mad-pa-no (dẓ:mε:-) ‘have you not received’, (cP) bzhugs-mad-pa-no ([?] *cu:mε:- or *cu:me:-) ‘is he not staying’, for neither of which does the available material provide sufficient evidence.

^{page 124 note 1} Only the difference in degree of vowel closure is relevant here; but, since there are no generally accepted symbols for e.g. closeness, half-closeness, other features (frontness, spreading, etc.) are unavoidably symbolized as well.

^{page 126 note 1} A phonetic form cιgι- has been noted for bzhes in addition to ce:gι-, bzhes-kyi-yod-pas ‘do you drink’; the latter form is perhaps to be regarded as a spelling pronunciation. Alternative phonetic forms chi:gι-, khyer-gyi-'dug, and chιa-, khyer-ba(ra)-red, have also been noted that would require khyer to be classed as cP (p. 128).

^{page 126 note 2} P. not only pronounced mthong-gi-'dug ‘he sees’, thωŋgydu:, but insisted that it and 'thung-gi-'dug ‘he drinks’ were homophonous.

^{page 127 note 1} Gould, and Richardson, , Sentences, 97Google Scholar.

^{page 127 note 2} ibid., 69.

^{page 128 note 1} Fast-tempo alternative to tsha:nε, tshar-nas.

^{page 128 note 2} Bell, , Grammar, 174Google Scholar.

^{page 128 note 3} ci:gι- (bskyed-hyi-'dug) provides the only example of i (p. 126).

^{page 128 note 4} Features thought to have been influenced by the orthography (pp. 126–7) are enclosed in square brackets throughout.

^{page 128 note 5} Except for a-yod (?ajø), a-yong (?aj:, ?ajѽ), ma-red (mare), mi-yong (mεj), the first Syllable is also the Verb Syllable.

^{page 129 note 1} Bell, , Grammar, 134Google Scholar.

^{page 130 note 1} Gould, and Richardson, , Sentences, 51Google Scholar.

^{page 131 note 1} Gould, and Richardson, , Sentences, 67Google Scholar.

^{page 131 note 2} Bell, , Grammar, 154Google Scholar.

^{page 132 note 1} symbolizes non-velarity.

^{page 132 note 2} ι is not preceded by D.

^{page 133 note 1} The V terms of some cP Verb Syllables do have alternative exponents, but in relation to differences in tempo, e.g. (i) bzhugs-pa(ga)-red; cu:ge (fast tempo), cωgbәie (slow tempo), cu:bәie (common tempo); (ii) btsir-ba(ra)-yin: tsi:bәjĩ (common tempo), tsliajĩ (fast tempo).

^{page 133 note 2} These features refer to the Particle Syllable a, e.g. ?Ajĩ. The first Syllable is a Particle Syllable in a-yin, mi-'dug.

^{page 134 note 1} Bell, , Grammar, 43Google Scholar: ‘The Tibetan Verb… is in effect a Verbal Noun. Thus: khos lug-sha za-gi-'dug… he is eating mutton, lit., by him, as regards mutton, an eating is’. Hannah, Herbert Bruce, A grammar of the Tibetan language, Calcutta, 1912, 239Google Scholar: ‘the so-called Tibetan Verb is rather a kind of Noun, modified in its signification by the Verb to be, according to the mood or tense of the latter’.

Even orthographically it is not possible to identify the Verbal Particle with the Nominal: the form of the Verbal Particle gi/gyi/kyi that follows a vowel letter is gi; but the form of the Nominal Particle gi/gyi/kyi/-'i that follows a vowel letter is not gi but -'i.

^{page 135 note 1} The analysis implied by Tibetan orthography is, incidentally, in agreement with the prosodic analysis in this respect (ε/ι is symbolized by e, /ω, ø/y, D/ω, and /O by o, and ε/e, a/A, and a/A by a), and therefore seems superior to the phonemic analysis of this material.

^{page 136 note 1} Bell, , Grammar, 174Google Scholar.

^{page 136 note 2} On the inadmissibility in phonemic analysis of assigning successive occurrences of the same sound under the same phonetic conditions to different phonemes, see Bloch, Bernard, ‘Phonemic overlapping’, American Speech, XVI, 278–84Google Scholar, and Readings in linguistics, Baltimore, 1957, 93–7Google Scholar.

^{page 137 note 1} Allen, W. S., ‘Some prosodic aspects of retroflexion and aspiration in Sanskrit’, BSOAS, XIII, 4, 1951, 939CrossRefGoogle Scholar.