A few words in Ragda-Sazgwa are onset-less, but Proto-Sazgwa tone derives in full from codas (mostly voiced vs. [or resp.] unvoiced, but also for sibilance), creaky voice, or long vowels, so it avoids that problem. Some P-Sazgwa examples of tonogenesis from sibilance over voicing, in case anyone is interested - 1. #"rAktA 2. Coda fricativisation - "rAxtA 3. Unstressed syllables take low tone - "rAxtA_L 4. High tone before /x/ (unvoiced) - "rA_HxtA_L 5. Coda /x/ becomes /h/ - "rA_HhtA_L 1. "pApkA 2. Voicing dissimilation on unstressed syllables (and voicing spread) - "pAbgA 3. Coda fricativisation - "pAvgA 4. Unstressed syllables take low tone - "pAvgA_L 5. Low tone before /v/ (voiced) - "pA_LvgA_L 1. #"siski 2. Voicing dissimilation on unstressed syllables (and voicing spread) - "sizgi 3. Unstressed syllables take low tone - "sizgi_L 4. High tone before sibilant - "si_Hzgi_L 1. #sAs 2. High tone before sibilant - sA_Hs 3. Coda /s/ to /K/ - sA_HK Does this work, tone-people? Also out of interest, does anyone else have distinction between velar and uvular fricatives? Natlangs? It doesn't seem unlikely, but I haven't stumbled across one so far.