'Brightening' in Ersu, Lizu, Duoxu and neighboring languages

The Qiangic subgroup, the hypothesis of a common origin of thirteen Tibeto-Burman languages of Southwest China (e.g. Sun 2001), has come under much scrutiny in the past decade as more data on these languages have become available for fine-grained analysis and comparison. While it is now generally accepted that these thirteen languages mostly share only typological features, which does not constitute sufficient evidence for placing them together in a separate branch, there remains one phonological innovation proposed for the entire “Qiangic” subgroup by James A. Matisoff (2004): “brightening”, or a strong tendency for *-a, the best-attested rhyme in Proto-Sino-Tibetan (PST), to be raised and fronted to -i or -e. In the Sino-Tibetan context, this phonological tendency is quite exceptional, hence making it an interesting common trait of these languages and, if it really is a common innovation, evidence for the existence of the Qiangic subgroup. However, the phonetic basis of this unusual change, and the reasons for its inter- and intra-language variability, all remain without explanation. This paper presents a critical assessment of the phenomenon of “brightening”. We start with a small subset of Qiangic languages for which we have in-depth data: Ersu (Ganluo), Lizu (Kala, Ga’er, Mianning), and Duoxu (Mianning) (all “ELD” languages). We take as the starting point the original list of “brightened” forms presented in Matisoff (2004). We focus on cognate reflexes showing vowel raising in ELD languages (see 1-4), and provisionally set aside those forms where no brightening is evident in ELD (as in 5). No.GlossPTBErsuKala Lizu Ga’er LizuMianning LizuDuoxu1‘bee/bird’*byabz̩̀bǐbi³³`bibi²¹2‘wear’*gwaʒź̩[dê-]wu[da³³-]vu⁴⁴`[de-](ɣ)we, `[de-]vʉve²²3‘nose’*s-na ~ *s-na:rsú[mbú]tô[mbu]ke³³[mɐ⁴⁴]`ʃti[mbʉ]nja³¹[ku⁵³]4‘frog’*balpź̩[má]pi[mæ̂]pe³³mje⁴⁴pimæpa³³[ma⁵³]5‘come’*lalàlǎlæ³³læla²¹Based on this data, we argue that the “brightened” reflexes essentially fall into two groups: (1) Vowel raising can be correlated with the ...