Corumayas wrote:I have a question about the Proto-Western to Gezoro sound changes: there are a number of changes conditioned by stress, and then some changes to stress assignment in rule 24 toward the end; but no mention of how stress is assigned before that late rule. And looking closely at the rules and some of the outcomes in the lexicon I suspect that some tricky stuff may be going on. For example, my best guess is that PW γʷeye 'hill' > Gezoro gwei 'mountain' went through these stages: γʷeye > ˈɣʷeje > ˈɣʷeji > ɣʷeˈji > ɣʷji > ɣwji > gwji > gwei – that is, the final e must be unstressed when it gets raised to i by rule 4, but then (if I'm guessing correctly) the first syllable seems to become unstressed so that that e gets deleted in rule 8 (otherwise it seems like the word would become ɣʷi at rule 10, and end up as Gezoro **gui). Can you shed some light on this?
Most of the PW > Early Gezoro changes were originally written by kodé. When I started working with them, I had to figure out how to get the right outcome for his example words, and I came to the conclusion that the stress system he assumed for PW was most likely not fully phonologically predictable, but with lexical exceptions. My sound change file does have a phonological stress rule, with the option to override it for individual words: Primary stress on the first syllable, unless otherwise specified. Secondary stress on every second syllable after that, except that syllables beginning with
*ʔ cannot receive secondary stress.
The vowel syncope rules themselves are by far the most complicated part of the sound change file. In general though, quality changes target both unstressed and secondarily stressed syllables, whereas deletions target secondarily stressed syllables only if no adjacent unstressed syllable can be syncopated instead. For
*ɣʷeje, the development is in fact simply
*ˈɣʷeje > ˈɣʷeji > ˈɣʷej > ˈɣwej > ˈɣwei > ˈɡwei; I forgot to specify on the wiki that the
*ej > i change in rule 10 actually only applies in unstressed non-final syllables.
And now I've noticed another little thing: in rule 8, unstressed vowels are lost before a resonant; but then you comment that ɣ is treated like an approximant here. Does that mean vowels are lost before ɣ too? And then what about ɣʷ?
Both
*ɣ and
*ɣʷ count as resonants here, so fully unstressed vowels are indeed deleted before them. (However, rule 14 later re-inserts an epenthetic vowel in the same position, so that unstressed
*CVɣ *CVɣʷ usually end up as
Cig Cug.)
I think there should be room for a small Plateau Western family, and anyone making a sister language would need to duplicate your changes up the branching-off point. I'm imagining Gezoro and Tjakori forming one branch – it seems they must be quite closely related if they share everything up to the l > ɾ shift in rule 20; I'd put Proto-Plateau further back, say around rule 10, and I'd probably have a southern branch spoken more around the giant lake.)
Yes, I think there could easily be some additional Plateau Western languages besides Gezoro and Tjakori. I imagine Tjakori as definitely sharing everything up to rule 13, plus rule 16 (but not necessarily the loss of labialisation in rules 14/15) and the l > ɾ shift in rule 20 (but not necessarily including rules 17-19 and the rest of rule 20). Proto-Plateau, if it includes more languages, should probably share at least rules 1-8 and maybe 11, and may differ in the details of rules 9 and 10. (Note that in both of these situations, the exact timing of the "skipped" rule groups is for the most part not crucial for the Gezoro outcome.)