^{page 133 note 2} A corresponding tonal analysis, relating the two-term Tonal system to the word unit, has been applied to the Verbal Phrase in my ‘Verbal Phrases in Lhasa Tibetan—I’, BSOAS, xvi, 1954,150–3.

^{page 133 note 3} ‘Phonetic Table for comparing the Different Dialects’ (H. A. Jaschke, Tibetan-English Dictionary, pp. XVI–XVII). An examination of Jaschke's usage with regard to the word ‘word’ makes it certain that he is not relating his tonal statement to the word in the sense in which it is used in this article (see below, Delimitation of the Word).

^{page 133 note 4} Love Songs of the Sixth Dalai Lama Tshangs-dbyangs rgyamtsho, translated into Chinese and English with notes and introduction by Yu Dawchyuan and transcribed by Dr. Jaw Yuanrenn (Y. R. Chao) (Academia Sinica, Series A, no. 5, Peiping, 1930); ‘The Phonemes of Tibetan (U-Tsang Dialect) with a practical Romanized Orthography for Tibetan-speaking Readers’, by Rev. Miller, P. M., B.S. (Journal of the Asiatic Society. Letters. Vol. XVII, No. 3, 1951, 191–216).Google Scholar

^{page 133 note 5} ‘Tone is a suprasegmental phoneme of the syllable’ (Miller, op. cit., 202); though not specifically stated, a corresponding analysis is implied by Dr. Jaw's syllabic system of tonemarking.

^{page 134 note 1} Rinzin Wangpo (rig-hdzin dban-po) (R) was employed by the School of Oriental and African Studies as a Research Assistant in London from December, 1948, until September, 1949. The material obtained from R. was checked against the utterances of other Lhasa-dialect speakers in Kalimpong and in Gyantse during the Session 1949–50.

^{page 134 note 2} Firth, J. R., ‘Sounds and Prosodies’ (TPS, 1948, 127–152).Google Scholar The titles of previous publications in which the ‘prosodic approach’ has been used are given in W. S. Allen, ‘Some Prosodic Aspects of Retroflexion and Aspiration in Sanskrit’, BSOAS, xin, 1951, 945. Subsequent publications include Mitchell, T. F., ‘The Active Participle in an Arabic Dialect of Cyrenaica’ (BSOAS, XIV, 1952);Google ScholarRobins, R. H., ‘The Phonology of the Nasalized Verbal Forms in Sundanese’ (BSOAS, xv, 1953);Google ScholarSharp, A. E., ‘A Tonal Analysis of the Disyllabic Noun in the Machame Dialect of Chaga’ (BSOAS, xvi, 1954);Google ScholarCarnochan, J., ‘Glottalization in Hausa’ (TPS, 1952);Google ScholarRobins, R. H., ‘Formal Divisions in Sundanese’ (TPS, 1953);Google ScholarMitchell, T. F., ‘Noun-Particle Complexes in a Berber Dialect’ (BSOAS, xv, 1953);Google ScholarAllen, W. S., ‘A Study in the Analysis of Hindi Sentence Structure’ (Acta Linguistica, vi, Fasc. 2–3, 1950);Google ScholarHenderson, E. J. A., ‘The Phonology of Loan-words in some South-East Asian Languages’ (TPS, 1951).Google Scholar

^{page 134 note 3} Jaw makes use of five pitch-levels (op. cit., p. 27) and Miller of four (op. cit., p. 204). Other difficulties attributed to the phonemic technique of analysis may be illustrated from such statements as the following: ‘The actual tones in connected speech follow the general principle of one tone being spread over two or more connected syllables. Thus, the high falling tone often becomes a high level tone, the following syllable or syllables, whatever its original tone, taking up a low or falling tone’ (Jaw, pp. 27–8). Miller makes use of the concept of ‘Perturbation of Natural Tone’ (p. 206) implied by the quotation from Jaw above: ‘a naturally high tone syllable may be perceptually lower than a naturally low tone syllable occurring outside the first named syllable's “intonation phrase”’ (p. 203); ‘syllables other than the first in a word, and specially postpositionals and terminatives … are more susceptible to intonation pressure, and hence do not conform so readily to their inherent tone’ (p. 204).

^{page 134 note 4} For the technical specialization of the terms ‘structure’ and ‘system’, see R. H. Robins, ‘Formal Divisions’, 109.

^{page 134 note 5} The Tibetan script recognizes the syllable, terminated by the ts'eg; and a unit, terminated by the sad, comprising an indefinite number of syllables but approximating to the sentence. No intermediate unit is recognized.

^{page 134 note 6} Two structures are recognized for the syllable irrespective of grammatical category: CV, CVC. In my ‘Verbal Phrases—II’, 320, only one structure was recognized for members of the verb category, viz. CVC; this statement has proved unsatisfactory.

^{page 135 note 1} Cf. the prosodic statement of Tempo in relation to the Verbal Phrase in ‘Verbal Phrases—I’, 149: ‘Rate of Utterance’.

^{page 136 note 1} The phonetic examples are given in the International Phonetic Alphabet but subject to the same conventions as those stated in ‘Verbal Phrases—I’, 142. Two additional symbols have been used in both the phonetic transcriptions and the transliterated text: (I) —/— Pause. (II) - - - Sentence incomplete or interrupted.

^{page 136 note 2} Where possible, inter- and intraverbal junction have been illustrated from words having a common constituent in both of these two prosodic contexts (‘lhod’ in this instance), so that the exponents of either term may be given the maximum prominence.

^{page 136 note 3} The task of writing down one of the modern spoken dialects of Tibetan raises problems of some difficulty, for none of the contemporary spoken dialects appears to have an orthography: they all make use of classical Tibetan as their written medium. Thus I have been informed by H. E. Richardson, C.I.E., O.B.E., formerly in charge of the Indian Mission, Lhasa, that ‘it is perhaps not strictly correct to say that utterances in Lhasa Tibetan are written down at all. Tibetans do not write what they say except for special purposes such as your research and our sentences’ [i.e. Tibetan Sentences, by Sir Basil Gould, C.M.G., C.I.E., and Hugh Edward Richardson, O.B.E. (O.U.P., 1943)]. The Tibetan spellings given in this article are either those of R. himself or of dPal-hbyor P'un-ts'ogs. In the main they do not differ from Classical-Tibetan orthography, but the writers have sometimes indulged in phonetic spellings, especially where the phonetic implications of the traditional spelling are markedly different from the phonetic form heard from the recording. Some spellings will therefore look unusual, but then making recordings of spoken Tibetan is not one of the more usual Tibetan activities.

^{page 136 note 4} Since marker exponents of inter- and intra-verbal junction have already been stated for certain verb forms, ‘Verbal Phrases—I’, 146–9, the examples given in this article have been restricted to words analysable into (I) Noun + Particle, (II) Adjective + Particle.

^{page 137 note 1} When co-articulated with voice friction has been recorded as in free variation with plosion in Intraverbal Junction: g/γ, b/β.

^{page 137 note 2} An advantage of stating a prosodic system of junction is that it enables one to relate the various phonetic forms of a given constituent to a single invariable phonological formula; differences that there may be between the phonetic forms are stated as exponents of (I) Interverbal, or (II) Intraverbal, junction. There is thus no need to give one phonetic form precedence over the others, or dignify it with the title of ‘norm’. Cf. A. E. Sharp, ‘A Tonal Analysis …’, 169. Similar advantages are to be gained from a prosodic statement of tempo. From the examples in which a given constituent has been recorded in both Junction contexts it will be seen that Tibetan orthographic usage is not unduly phonetic: the syllable in question in pedze and tchabi (dpe-c'a, p'yag-dpe) is given the same symbolization in both (dpe) despite considerable phonetic differences.

^{page 138 note 1} See p. 136, note 3.

^{page 138 note 2} When co-articulated with voicelessness + non-aspiration plosion in the Intraverbal-Junction context is sometimes distinguishable from plosion in the Interverbal-Junction context: it is accompanied by a lax articulation, and might be symbolized in greater detail as g, d, b.

^{page 140 note 1} Labial friction has been recorded in interverbal junction, but only when final in the sentence, e.g. kheraile 'dukse 'lrβ duksa Miff (k'o-ran-Ia hdug-se lab), and is probably best considered as a sentencefinal feature.

^{page 141 note 1} Velar and dental nasality, voiceless velar occlusion, and apical friction (r), may be held to mark the absence of word boundaries in some styles of utterance but not in others: in the ‘reading’ and ‘spelling’ styles these four types of articulation may be final in the word, and cannot therefore be used as criteria for the delimitation of words in these two styles, whereas they can be used as criteria for the style considered here (‘speaking style’ ).

^{page 141 note 2} The exponents appropriate to st are symbolized in brackets.

^{page 142 note 1} op. cit., pp. 323–8; 340–2. ‘Prosodic System of Closure.’

^{page 142 note 2} A ‘word’ unit may be delimited in spoken Burmese by the same technique of analysis as has been applied here to Lhasa Tibetan. Indeed, the similarity even extends to particular marker exponents of inter- and intra-verbal junction.

^{page 144 note 1} Some of these particles have differing phonetic forms, according as the structure of the syllable with which they are in junction is CV or CVC. This phonetic difference is often, and overphonetically, reflected in the spelling:— CVC Junction CV Junction I. kyi/gyi/gi hi, e.g. dehi, dbyin-jihi, skor-gyi, bod-pahi. II. kyis/gyis/gis -s, e.g. nas, hdi-hdras, cf. sba-bu-lags-kyis. III. la la/-r/hi, e.g. snas-la; cf. lun-par, snas-mahi. IV. yan yan/hi/-s, e.g. da-yan, suhi, bsam-las.

^{page 144 note 2} See ‘Verbal Phrases in Spoken Lhasa Tibetan’, p. 134.

^{page 144 note 3} Two distinctive pitch levels are recognized. The symbols used are as follows:—: I. High level^-; II. Low level;_-; III. Fall /; IV. Rise-fall ^; V. Rise /.

^{page 145 note 1} See p. 144, note 1.

^{page 145 note 2} The Particle ni cannot be included in this list: although in some instances the criteria for denning the Nominal-Phrase Particle are applicable to ni, e.g. dehi rkyen-gyi de-ni na-ranhts'o bod-yig de-hdras-se bris byas …: Noun (de) + Particle (ni), other examples have been recorded in which they are not, e.g. med zer-na-ni t'on-yod-red-pa: Verb (zer) + Verb Particle (na).

^{page 147 note 1} See p. 144, note 1.

^{page 147 note 2} The intonation-patterns given are in every case exhaustive for the material studied.

^{page 147 note 3} In the Intonation-III context therefore a given word cannot be identified as a Tone One or a Tone Two word from exponents of the first order, pitch, but it may well be possible to identify it prosodically from other criteria, i.e. from the third order of exponent, word-initial features (word-initial voice + plosion, voicelessness + non-aspiration + plosion, etc.).

^{page 148 note 1} By using the intonation-pattern given, I have avoided associating the fall in pitch with either the former or the latter of the two syllables. In practice the fall in pitch has been perceived as associated with the latter syllable when that syllable is of a structure CVC, and with the former syllable when the latter syllable is of a structure CV, e.g. teagAIU (-CVC): ^-\, but ?yibe (-CV): \-.

^{page 148 note 2} For the sort of statement that would be required for an account of the relevant prosodies of the word (Tone) and the Sentence (Intonation), cf. A. E. Sharp, ‘A Tonal Analysis…’.

^{page 152 note 1} As stated above (p. 144, note 1), the exponents of this particle are dependent on the prosodic context, i.e. whether the syllable with which it is in junction is CVC in structure (e.g. sman-k'an, ka-sbug, hbras-ljons) or CV (e.g. lhas-sa, rgyal-rtse, dgon-pa). For these latter examples alternative phonetic forms to those given above have been recorded:— In ft, in fact, these alternative phonetic forms have been more often recorded than those (leisele, etc.) given at (II). The alternative forms are sometimes symbolized in the Tibetan text as lhas-sa-fa, rgyal-rtse-la, etc., and sometimes specifically indicated by the use of such phonetic spellings as lhas-sar/lhas-sahi, rgyal-rtser, etc.