Lijiazui Village is located in Wujiao Township, Muli County, Sichuan, adjacent to Yongning Township of Yunnan Province. The village has a population of around 400. The language spoken by these people, whose endonym is /nɑ/, is referred to here as the Na language (following Lidz 2010); it is also know as ‘Mosuo' and ‘Narua' (ISO 639-3 code: NRU).
Traditionally, scholars working on tonal languages in China used to record any tone they encountered with the five-point scale devised for Chinese dialects, which has shown its usefulness in the notation of contour tones in the Sinosphere. Recently, some new models and perspectives have been applied for the analysis on tone system of Sino-Tibetan languages of Sichuan and abutting areas, e.g. Southern Qiang (Evans 2008), Pumi (Ding 2001; Jacques 2011), Naxi (Michaud & He Xueguang 2007), and Yongning Na (Michaud 2008).
Some phenomena in the initial fieldwork on Lijiazui Na inspired me to analyze its tonemes as constituting a level tone system. For example, different phonetic behaviors are observed for one word, which hints that these surface tones belong to one underlying toneme. Fig 1 shows four tokens of /lɑ˧/ ‘tiger', colored red, yellow, blue, and green from the first to the fourth. The 1st, 3rd, 4th token have close mean F0 value, while the 2nd token has a difference of about 2 ½ semitones vs. the other three. With the five point scale, the 1st, 3rd, 4th token would be recorded as high to mid falling tone, and the 2nd as mid level tone. However, the distinction of F0 of one word can also be explained as belonging to mid tone in the view of level tone system. For another instance, the toneless possessive particle /bv/ gets its tone from the former syllable, which suggests that the tone of nouns in possessive constructions can be ‘split' or ‘spread', the pattern extending over the noun itself and the following possessive particle.
Fig.1 Tiger (unable to be pasted here)
A set of experiments placing lexical words in carrier sentences (frames) have proved that the surface tonemes of Lijiazui Na map to various underlying tonemes. The experiments attempted so far aim to cover the following categories: (i) isolated words, (ii) compound nouns, (iii) object-verb combinations and (iv) numeral-plus-classifier phrases. Phonetic data (F0 tracings) are also compared in order to advance further towards understanding the connection between the phonetic behavior and phonological identity of the various tone categories.
Provisional conclusions will be set out: at the present stage, the system is analyzed as comprising (i) four surface tonemes: M-mid tone (33/53); L-low tone (31); LM-low to mid tone(13); MH-mid to high tone (35); (ii) eight underlying tonemes for nouns, and six for verbs.
