PDF《inventory, and a short word list》

1 must be added some phonemes that are so restricted in their distribution that it is simpler to present them separately. These are (i) a neutral vowel, /?/, appearing in some function words such as the negation, /m??/, and taking on its quality from the vowel that follows;

and (ii) as many as nine nasal rhymes, shown in Table 2. syllable hw?a h? h?? h h? hw? h? h?e example 绿豆听,金毛走慢人站, 面包 syllable ?? v?? ? example 呛螳螂,自,驴 Table 2. Laze syllables containing nasal rhymes (in phonemic notation). Most syllables with nasal vowels are made up of an initial [h?] and a nasal vowel: nasality runs throughout the syllable. They can be analysed synchronically in either of two ways: as /h /, where / / is a nasal vowel, in which case a large set of nasal vowels must be granted phonemic status;

or as /h?V/, recognising an additional consonant: a nasal fricative, /h?/. The latter solution, adopted by Bradley 2003:224 for similar examples in Lisu, is more economical in terms of number of phonemes. These sounds originate diachronically in initial consonantal clusters CN-, where C is an obstruent and N a nasal consonant. There are also three syllables without an initial /h/, consisting simply of a nasal vowel: [?v?and [? ?] . Examples include: (i)brass'

(ii) [?v???] '

to irritate (e.g. smoke irritates the eye)'

;

Huang Bufan also reports [?v???] (original transcription:for '

goose'

, a word for which the language consultants in the valley of Xiangjiao now use a Chinese borrowing. (iii) [l?mie?lpraying mantis'

, [? ??ts ??] '

-self (in: myself, yourself)'

, and two likely borrowings from Pumi/Prinmi, which up until the middle of the 20th century was the lingua franca of Muli (about the history of Muli, see Wellens 2006): [t ??bv??] '

donkey, ass'

and [n ??bv??t ?i?] '

elephant'

. These syllables raise an issue for phonemicisation. Those with an initial glottal stop can be interpreted phonemically as simple vowels, the phonetic glottal stop being an empty- onset filler, on the analogy of other cases such as /i/, realised as [?i]. Pushing further the logic of descriptive economy, in the absence of any [??] or [? ] syllables, the syllable [???] could be interpreted as the phonetic realisation of a simple /?/, and [? ?] as / /;

but this option, which would be costly in terms of phonetic plausibility, is not open for [?v??], since that syllable contrasts with [v]. We propose to analyse [?v?and [? ?] as /v? and / ?/, a decision which is consistent with the recognition of the phonemic nasal rhymes w?a,?/, /w? /, /?/ and /?e/ after /h/. 2. A note about vowel harmony and sandhi phenomena in Laze Some syllables in Table

1 carry the mention '

VH'

, indicating that they are probably the result of vowel harmony. Some explanations about this phenomenon are in order. Sporadic anticipatory vowel harmony ('

right-to-left'

harmony) is common in Laze in connected speech: for instance, /i?dy?/ '

family'

is sometimes realised close to [y?dy]. Phenomena such as vowel harmony and the voicing of intervocalic voiceless consonants are cross-linguistically common in connected speech;

however, the extent to which these phonetic tendencies manifest themselves and eventually become lexicalised or even generalised as phonological rules is highly language-specific. Of the three Naish languages studied here (Laze, Na and Naxi), Naxi is least prone to the lexicalisation of such phenomena, and Laze most prone to it. Vowel harmony is lexicalised in some disyllabic words, along with other processes such as the voicing of intervocalic voiceless consonants (and tone sandhi, about which see, again, Michaud 2009). A typical example is / ie?lie?mie?/ '