¹ Also called Salween Basin Group.

² An elementary Palaung grammar (Oxford); A dictionary of English-Palaung and Palaung-English (Rangoon). MrsMilne, also published an account of Palaung culture: The home of an Eastern clan (Oxford, 1924).Google Scholar

³ Formerly also called Katur.

⁴ For the terms ‘structure’ and ‘system’ cf. Robins, R. H., ‘Formal divisions in Sundanese’, TPS, 1953, 109Google Scholar, n. 2, and passim. ‘Level’ throughout this paper refers to levels of abstraction of the kind indicated, and is not used in the sense made current by J. R. Firth; it thus corresponds to Simon, H. F.'s ‘stage of analysis’ (‘Some remarks on the structure of the verb complex in Standard Chinese’, BSOAS, XXI, 3, 1958, 554–6).Google Scholar

Structural statement ‘at’ a given level always entails reference to at least one ‘lower’ level, abstraction at which provides the units of the ‘higher’ structures. Thus no structural statement is or can be made at the lowest level, that of the sound.

¹ The writing of ə in minor syllables may be theoretically justified as an indication of syllabicity. The first syllable of e.g. ime ‘male’, which appears to invalidate this statement, is interpreted as /y/; cf. p. 553.

² I use the term ‘pattern’, at the suggestion of my colleague Mr. R. H. Robins, to denote the total manifold of structures and systems. ‘Primary pattern’ thus corresponds to the term ‘primary system’ introduced by Henderson, E. J. A., TPS, 1951, 132Google Scholar, and ‘structure’ to Henderson's ‘pattern’.

In a sample of 2,800 words of running text in formal narrative style words referable to secondary patterns accounted for 1·8 per cent of occurrences.

¹ Burmese words are cited in the transcription used in Stewart, J. A., Manual of colloquial Burmese (1955).Google Scholar The Shan transcription is that devised by my colleague Mr. E. H. S. Simmonds; the tone marks ′ (rising), ˇ (low), (falling), ' (high), (short falling) correspond to Cushing's tones 1–5 respectively.

² Stress is to be understood as including correlated features of intensity, relative pitch level, and rhythm.

³ Since the Palaung vocabulary has not been exhaustively recorded, it is likely that many such words would prove wholly analysable on further investigation.

¹ The use of the symbols C, V for ‘consonant’ and ‘vowel’ is not to be taken to imply equivalence of the classes represented by them wherever they occur. The value of a symbol in each case is given by the choice of symbol together with its context.

² For the interpretation of [s'i], etc., see p. 549.

¹ cf. Shan l n shi. Of 39 words in which a sibilant was followed by a front vowel or central-to-front dihthong, 27 were ronounced with s, 7 with SW, 5 with s or sw.

¹ In rapid speech the distinction ai ≠ ah is one of vowel quality only; cf. . 553.

¹ The correlation between slow speech and stressed pronunciations is general for tempo-linked features.

A pronunciation of 9n ‘he’ as [], [n], heard in certain unstressed contexts, is best attributed to a distinct weak form.

² The opposition eə ≠ iə has little distinctive value. eə, in 26 out of the 32 words in which it was recorded, followed a consonant with dental or alveolar articulation—t(h), d, n, (C)r, (C)1, SW—while iə followed such a consonant in only 13 out of 57 words, against an expectation on random distribution of 14 and 25 words respectively.

Apart from this limited opposition it is possible to regard the ə-vowels as constituting a phonological series /ə/, /yə/, /wə/, in which ə is the locus of a ternary prosodie system. Diphthongal /y/ and /w/ are differentiated from y and w as terms of the secondary system described on p. 549 (e.g. in kyŋ, kwat) by their occurrence following all types of initial, as opposed to the limited C₁ possibilities of the secondary pattern: hreəŋ, kluəy are possible, but not e.g. *hrya, *klwa. They are transcribed as i, u in order to make clear this distinction between elements analysed as vocalic and consonantal respectively.

¹ cf. p. 545, n. 1.

¹ The only exponents common to all rapid speech contexts are voicelessness and continuance; transcription by r is recommended by the alveolar fricative articulation in slow speech contexts, and the wider distribution of this articulation in rapid speech. But issues of cross-identification are raised when this fricative is compared with the lingual roll represented by the same symbol in contexts other than major syllable final position.

¹ This feature does not extend to structures C(C) + lhV((V)C).

² There is one exception: pəbo ‘onion, garlic’. i- does not occur before palatals in such structures, nor S before S, but the statement of a general ‘heterorganic’ feature is precluded by the occurrence of sequences r + r, ? + ?: rərəp ‘net’, ə?O? ‘dog’. The restriction does not apply to two-word structures C CV((V)C): kə guəy ‘there is not’.

¹ Paw Shwe Kya spoke Burmese without tonal distinctions. I cannot answer for his Shan.