On Sat, 9 Mar 2013 14:11:07 -0500, Dustfinger Batailleur <[log in to unmask]> wrote:

>[posted on the CBB, but no replies, so reposted here]

Well, Rosenfelderesque notation with a bunch of inscrutable one-letter category names is hard to read.  You coulda cleaned it up for human consumption some.  

Anyway, there's a bunch here that I don't find very likely...

>/i iː u uː/ <i ī u ū>
>/e eː o oː/ <e ē o ō> [ə]
>/a aː/ <a ā>

What is it that gets realised as the schwa, under what conditions?


This is a weird change.  It's most likely to be some sort of stress-based lenition, but then the conditions would probably invoke the stress structure of the word, making reference to nearby syllables and such, and not happening everywhere (e.g. it probably wouldn't happen word-initially, then, where lenitions are rare).  I suppose it could derive from a sort of syllable-length balancing (like the Finnic gradations) as I mention below, if the voiced consonants in the output were historically super-short.  Or just maybe (though this has been claimed never to happen by some) long vowels could bear intrinsic lower tone, and the low tone could be what's causing voicing changes.

The latter two of these solutions are likely to have other implications...


How universally do these apply?  Every resonant has an unvoiced variant?  Even resonants, and [?], induce this?  /tr/ [dr\], /r?/ [4_0?] ?


This'd be quite believable if the outcomes of these processes were _geminates_, that is there was place assimilation as well.  But fricativity, per se, is not the sort of feature that assimilates, in natural language: indeed, it more often _dissimilates_ in the cases I'm aware of, which suggests acoustic rather than articulatory motivation.  It might be believable that stops were lenited to (more audible) fricatives in contexts where they were hard to perceive otherwise; but why would this only be before fricatives?

On a larger view you have stops & affricates & fricatives all behaving kinda differently, and this is a little unusual (not too unusual, but a little).  Where did your affricates come from?  This might help you figure out what sort of processes they should engage in (and of course this advice is more generally applicable too).

In most languages that have them, affricates are _phonologically_ very similar in behaviour to stops, on account of that they were historically stops (of some PoA not represented among the stops anymore, mostly).  Given that you don't have a complete series of them, but just /tS dZ/, this is quite likely in your case (that they developed from a palatalization of one or more of the other series of stops, and are structurally like /c J\/).  Sometimes instead affricate-hood behaves more like a phonation or matter of articulation, but in this case there tend to be large series of them, at least initially (e.g. the second German consonant shift; some dialects retain /pf) ts) kx)/).


These are also implausible.  I can't think of a single case in which onset quality has affected vowel prosody.  And even if it was going to, on the grounds of the cases I know that go the other direction (like Finnic lenition), the tendency would be towards _uniformising syllable length_.  Since affricates take longer to produce than plosives, you'd expect them to occur with slightly shortened vowels.  


It's worth knowing that the contrast between alveolar and dental isn't actually the important one in most languages that are reported as having it; what's usually more important is that the alveolars are apical and the dentals are laminal.  


After _front_ vowels, is what makes sense here.  Close front vowels, if you like.  But not just close vowels: [k] > [c] after [u] is bizarre.


If a fricative goes to an affricate in some environment it's much likelier to be when it's _after_ something suitable.  Affricates differ from fricatives in how they begin, after all.


Odd but believable if it was some sort of historical complex coda simplification rule.  Historical /anl/ > [an] becoming synchronic /a~l/ > [a~], that sort of thing.

>#(S)(A)VC# <--roots

Reasonable; what about the phonotactics for _words_, though?

>It's supposed to be for a protolang by the way. 

That shouldn't make any difference to its structure, says the uniformitarian principle.  (Of course, might mean you design it to be especially prone to the sorts of changes you want to have happen along the way.)