On Oct 9, 2009, at 11:30 PM, Roger Mills wrote:

> --- On Fri, 10/9/09, Kelvin Jackson <[log in to unmask]> wrote:
>
>> I got this idea a couple days ago:
>> have a maximum syllable structure of CVC, but make roots
>> mostly CVCC. Then, you would delete one of the final
>> consonants
>
> My first feeling about this was that it was contradictory. If  
> "roots" (i.e. the underlying, abstract, forms) are mostly CVCC, then  
> that has to be an allowable "maximum syllable structure". Then, as  
> you say, there would be rules that delete one or another of the  
> final C in absolute final position.
>
> Then David P reminded me of the Polynesian languages that have lost  
> original final C, but those C (or at least some of them) reappear in  
> suffixed forms. Various Melanesian langs. (see esp. Fijian, in  
> Capell's good dictionary) and some Indonesian langs. show similar  
> developments.

I recall seeing a paper recently saying that the endings in some  
Polynesian languages (though I don't remember which ones) were at  
least partly predictable based on the phonological material of the root.
I don't remember where I saw that, but at the time I was looking for a  
paper which I think was called "Why Ainu isn't like Maori" or vice  
versa. It compared the Maori system of endings to Sakhalin Ainu, which  
neutralizes syllable-final /p t k r/. The original consonant  
resurfaces in suffixed forms there, too. I really wish I could find  
that paper again!

Apart from that, I've seen the same sort of segment resurgence  
phenomenon in Wampanoag.

>
> In every case I know of, there is a llmited number of C that can  
> reappear in suffixed forms-- sometimes they reflect the original/ 
> proto C, often not. Clearly a lot of analogical shuffling has gone  
> on, as speakers generalize, or forget, what the original final was.  
> So that could be a development in your proposed language.
>
> In the language family I worked on (South Sulawesi, Indonesia) the  
> inventory of original final C (*p t k m n N r l s) had been severely  
> reduced--
>
> in one subgroup, *ptrls > all ?, *k was preserved; of the nasals *m  
> n > n, N retained. In suffixed forms, ? could be replaced by t, r,  
> or s;  k n N were of course retained.
>
> a second, major, language (Buginese) merged all ptkrls > ?, all  
> nasals > N; when suffixed, ? > k,r or s. You just had to know, or to  
> have been taught, which C to choose....  (N unaffected).  There was  
> a very good dictionary available (1853), from which I'd culled a lot  
> of bases + derivatives. Generally, the k-derivs. reflected original  
> *ptk;

Interesting that all three would get /k/!

> r reflected orignal *r,l; s reflected origianl *s (though IIRC even  
> in 1853 some anomalies had crept in).  Middle-aged/older informants  
> tended to agree with the dict.; but some of my younger (20s)  
> informants were quite adamant that e.g. the "correct" derivative of / 
> nipi?/ 'thin' (*nipis) was _nipiri_, not the dictionary's _nipisi_--  
> so there you see the beginning of the system's collapse.
>
>> So, for example:
>>
>> kasr /kas/ or /kar/, but kasra /kasra/
>> temn /tem/ or /ten/, but temnov /temnov/
>>
>> Comments? Is this ANADEW?
>
> Well I guess it is.... It looks like Alex's Korean ex. (talk  
> 'chicken', pron. [tak_}] (I'm unsure what that "_}" represents)

Unreleased.

> -- I'd assume that the written form probably retains the "l", no?  
> (Just as Engl. damn/damnation does).  Quite likely your language's  
> spelling system would have to retain "silent" final C like French,  
> so speakers would know what to add on, otherwise you're heading for  
> a lot of analogical re-formations a la Polynesian/Fijian. 
>
> In Dirk's Uto-Aztecan exs., I'd assume that possessed forms are  
> sufficient common (as opposed to the bare, vowel-less forms) that  
> speakers still "know" what the final V is.
>
> Interestingly, some of my Buginese informants held that the  
> "correct" forms of the various suffixes were "-ki, -si, -ri" et al.   
> That may be because in the native (and I think romanized)  
> orthographies, final -? is not indicated. I wonder if that is  
> happening in the Uto-Azt. case-- are ("-in, -on, -ün, -an" etc.)  
> considered (natively) "variants/allomorphs" of the 1sg. poss.  
> suffix???
>
>
>