Re: Revising my consonant inventory

On Thu, 9 Aug 2007 06:41:35 -0400, Nick Scholten <nick.scholten@...>
wrote:

>The past couple of days I've had quite some ideas. A part of this is, after
>I revised the allophony in the vowel system of Kitanic, I figured maybe my
>consonant inventory could use some work also. The language currently makes
>these distinctions (er, how do you make this comprehensible in monospaced
>text... I'll just make a list). It uses only 3 POA's: bilabial, alveolar and
>velar. So, the phonology (in X-SAMPA):
>
>......................................
>unvoiced stops: p t k
>unvoiced fricatives: p\ s x
>voiced fricatives: B z G
>nasal stops: m n N
>alveolar lateral fricatives: K and K\
>alveolar lateral approximant: l
>
>possible clusters:
>initial: tK ts ps ks
I wonder whether /tK ts/ are actually unitary affricates /tK) ts)/?  If not
for the /ps ks/ that would be a tempting analysis. But then it would be kind
of unexpected that they only show up initially.

>medial: ml nl Nl, Bl zl Gl, mp nt Nk, + finals.
>finals: p\k xp xk, p\N xm, Bm zn GN.
>
>besides clusters it's basically CV(C) but words can start in vowels.
>Possible last phones are: p t k, n m N, p\ B x.
>
>to add vowels: /a [A] E I i o [O] u/ + /ai [AE] Ei [E:] oi [OE] au [AO] Eu
>[EO] IU [IO]/ (the allophones occur in unstressed sylliables)
>........................................
>
>On clusters: I thought that, with consonant clusters, languages allow
>different ones between vowels and at the ends of words. Then I read
>something about the whole onset and rhyme thing which didn't mention such a
>thing at all. Is it true that all possible rhymes can appear either between
>vowels or in a final position?
As far as I know, languages which allow different cluster sets internally
and finally will make the internal set a superset of the final set.
(Counterexample, anyone?)

In terms of onsets and rimes, each internal cluster occurs at the juncture
of two syllables, so will consist of the coda of one syllable plus the onset
of the next.  If you assume there's nothing special about rimes in absolute
final position, this looks like a valid final (cluster) followed by a valid
initial (cluster).  There may of course be further restrictions so that not
every final + initial is a valid internal.  If you assume furthermore that
you can always make the onset part of an internal cluster zero, then what
remains is just a final cluster; but that's a big if (and you might have the
maximum onset principle to contend with anyway).

That said I don't know the answer to your actual question, as to whether
final syllables more commonly allow more, or fewer, or the same set of rimes
as nonfinal ones.

>Anyway, on to my real issue. I'm stuck on what to do with this. I feel that
>it's too regular, and frequently when I'm trying to make up words [h] and
>[j] and voiced stops keep showing up. Also, I think the phoneme inventory is
>too small, and is so regular it forces me to use sounds I don't like much
>anyway. The changes I would like to make:
>
>adding: /h/ /j/ /w/ /?/ /5/ or /L\/ or both as allophones. Adding a
>voiced/ejective/aspiration distinction in stops or a combination of those.
>Palatalization. Maybe /X/ and /S/ or /s`/. I'm not sure if I want a rhotic,
>maybe /r/ is a good idea but I'm unable to pronounce it.
>
>losing: the voice distinction in fricatives, and losing bilabial maybe
>replacing it with labiodental, maybe /x G/ also. /K\/, maybe /p/ also.
All these additions and deletions look entirely fine.

>I know that the choices you make for a sound system are largely subjective,
>but I want it to be somewhat realistic. I have no idea whatsoever what
>consonant clusters are realistic, which is the main reason I'm asking these
>questions.
Well, one hears much less about universals or tendencies regarding cluster
inventories than those regarding phoneme inventories.  I suspect there
aren't many such, and you can probably justify any questionable parts of
your cluster system with the right sound changes.

Your final clusters /p\k xp p\N xm/ are interesting for not being homorganic
when the homorganic equivalents are mostly disallowed; usually it's
homorganic clusters that are favoured, due to assimilation.  It's really odd
that voiceless fric + nasal should be antihomorganic while voiced fric +
nasal is homorganic.

And if you can have /ml nl Nl Bl zl Gl/ you might also have /l/ after other
things, /pl tl kl p\l sl xl/.  Well, unless that's because the voicing has
to agree, which I see it mostly does, except in /mp nt Nk/ which are begging
as your system stands now to be realized [mb nd Ng], and in /p\N xm/.  So,
as an idea, maybe generalize to all clusters a (phonetic) constraint that
voicing must agree?

Alex