In the last episode, (On Tuesday 03 July 2007 17:23:16), Jeff Rollin wrote:
> In the last episode, (On Tuesday 03 July 2007 15:53:29), Dirk Elzinga wrote:
> > On 7/3/07, Jeff Rollin <[log in to unmask]> wrote:
> > > 1)  Is it credible to restrict the consonants that can appear in
> > > clusters to exclude the prenasalised variants?
> >
> > It doesn't seem likely to me, especially if the consonants really are
> > prenasalized rather than a sequence of nasal and homorganic stop. (Is
> > there a distinction between N+C and a prenasalized stop? That would
> > also be unusual.)
> Yes, there is. I'll explain below since you quoted a relevant example
> below.
> > > 2) Is it credible to restrict initial syllables to those beginning with
> > > a consonant, and have vowel-initial syllables internally?
> >
> > No. In fact, just the opposite pattern is typically found; that is,
> > typically you find only consonant-initial syllables in word-medial
> > position, with vowel-initial syllables allowed word-initially.
> Indeed. I may have to settle for a simple preponderance of
> consonant-initial words over vowel-initial. /a, , e, i, o, , u, y/ is,
> after all, only eight vowels (fewer if you group, e.g. /a/ and // together
> because of vowel harmony)
> > > kanta (nb cluster!)
> > >
> > > kanda (prenasalised "t")
> >
> > The distinction betwen /kanta/ and /kanda/ seems unlikely to me. How
> > is this distinction realized phonetically?
> Three ways. First, I should reiterate that although consonant clusters are
> not allowed word-initially or finally, so that "palast" could never be a
> word, nor "stapal", they are allowed medially, so "pastal", "taspal" etc
> are valid words (but not "lastap" or "stalap", since "-p" isn't allowed
> word-finally either.)
> We do find:
> palant / ntava
> and even
> ntwandwa / myalant'
> (where ' marks a final, palatalised consonant)
> Secondly, the voiceless plosives /p, t, k, c?, q/ and their aspirated,
> palatalised and labialised variants have neither allophonically nor
> phonemically voiced variants, whereas the nasalised plosives do (in medial
> position).
> -- This is also the source of the usage of (I) "nt" &c to mark (a) a medial
> cluster, (b) an initial nasalised plosive, and (c) a final nasalised
> plosive, and of (II) "nd" &c to mark the corresponding nasalised plosive in
> medial position.
> (I could also have used "nt" to mark the cluster and just "d" to mark the
> nasalised plosive - but I don't think the latter is very intuitive. Plus it
> reintroduces/increases the use of plain "b", "d", "g", which make the
> language look more like Estonian than Finnish.)
> Thirdly (and lastly), the clusters are subject to consonant gradation,
> whereas the nasalised variants are not. For example, the genitive of
> "ranta" "speech" is "rannan", whereas the genitive of "Nanda" "Nanda
> (woman's name) is "Nandan". Similarly "mamma" "mother" is "maman" in the
> genitive, and "mamba" (criminal) is "mamban" in the gen.
> This is probably because the prenasalised consonants seem only to occur in
> the onset of a syllable (or the coda of a final syllable).
> > > 3) Anyone know of a conlang that has two (or more) tones and has to use
> > > different diacritics to represent them over different letters (e.g.
> > > high and low tone over front and back vowels?)
> >
> > I have no idea. But this seems to be an orthographic rather than a
> > phonological decision.
> Yeah. I just put it there to avoid starting another thread.
> > Bear in mind that this is *your* language, and you should do what
> > feels right to you in constructing it. This may include flouting
> > proposed universals of human language.
> >
> > Dirk
> That's true. Thanks for that.
> Jeff

Now that I think about it, actually, a better analysis might be:

"Vn has plain [can anyone think of a better word than that?], aspirated, 
palatalised, and labialised consonants, and allows consonant clusters of up 
to two consonants (including asp, pal, and lab variants). Initially, and 
finally, only clusters of nasal+stop are allowed. Medially, stops following 
nasals tend to be voiced." 

Of course that would mean that I would have to get rid of the distinction 

ranta -> rannan

Nanda -> Nandan

plumping for either:

ranta -> rantan / Nanta -> Nantan OR ranta -> rannan / Nanta -> Nannan 

But it might make more sense phonologically. OTOH maybe there are voiceless 
(but not voiced) plosives, but /both/ voiced and voiceless nasalised 



PS I just remembered: IIRC, in Finnish, consonant gradation:

1. ignores the presence of plural -i- before case endings

2. doesn't apply to proper names or borrowings (thus the genitive 
of "auto" "car" is "auton", AIIRC, not the expected "audon". Since it seems 
unlikely that there is some "phonomagical" process going on here which can 
distinguish between the /t/ of "auto" and that of, say, katu "street", and 
assuming the allegation in:

That "prestopped nasals" exist not only in some Australian languages but also 
in Slavic languages, it seems natlangs are a lot weirder than any conlang 
I've ever created! I might stick with my original system.

"Please understand that there are small        
European principalities devoted to debating   
Tcl vs. Perl as a tourist attraction."

                            -- Cameron Laird