Diachronic developments of voiceless nasals: the case of Ersu, Lizu, and related languages

Ersu, Lizu and Duoxu (collectively ELD) are three closely related, little-studied Tibeto-Burman [TB] languages of Sichuan Province in China. Their position within the broader TB family is a matter of dispute. Recent analyses variously link them to other lesser-known TB languages of Sichuan (known under the term Qiangic, Sūn 2001a) or consider them more closely related to the Naish languages (Bradley 2008, 2012; Jacques & Michaud 2011). This study presents one significant new finding for the reconstruction of Proto-ELD (going beyond the conclusions of Yu 2012): the existence of voiceless nasal onsets. This finding not only illuminates the broader problem of classification of the languages of the area, it also suggests the existence of a universal pathway of sequenced changes related to the development and loss of voiceless nasals in languages of the world. The study makes use of a significant amount of new data arising from recent fieldwork. The conclusions are based on a combination of (i) the techniques of the comparative method within the ELD cluster, (ii) external comparison with cognates elsewhere in Tibeto-Burman, and (iii) analysis of universal phonetic mechanisms and constraints. Voiceless nasals are posited based on cognate sets like the following for 'ripe', showing a correspondence between a voiceless fricative in Ersu, a voiceless nasal approximant in Lizu, and a nasal stop in Duoxu: Ersu /(dɛ³¹-)xe⁵¹/, Lizu /(de³³-)h ̃ e⁵¹/, Duoxu /me³⁴/ The reconstruction is supported by cognate forms elsewhere in TB (for example, Written Burmese /m̥ ɛ 1 / 'ripe' and Written Tibetan smyin 'ripe'). In combination with recent acoustic studies of voiceless nasals in several different TB languages, it is argued that ELD developments are part of a general pathway of change that can be schematized as: 2 *sN > *N̥ N > *N̥ > h̃ > x While the development of voiceless nasals within Tibeto-Burman is not uncommon, the high degree of consistency within ELD that allows for the regular reconstruction of Proto-ELD voiceless nasals in a particular subset of lexical items constitutes an innovation that suggests that the ELD cluster is a legitimate taxonomic node within TB that may not be as closely aligned with other TB languages of Sichuan as previously thought. We further suggest that the developments of nasal initials may be used as a general diagnostic tool to help sort out the relationships among lesser-known languages of Sichuan whose genetic and contact affiliations remain obscure. More generally, the study provides further insights into the synchronic and diachronic aspects of voiceless nasals, a type of sound that remains somewhat poorly described and poorly understood due to its relative rarity in languages of the world