Re: nasalless and stopless language

>> The Austronesian langs. of South Sulawesi (ex-Celebes) abound in geminate
>> stops, both voiced and voiceless.  The latter, of course, are relatively
>> easy to produce.  For the voiced gem. stops, most of the languages resort
to
>> a cluster glottal stop + stop; one informant even had an brief intrusive
>> echo of the preceding vowel between the /?/ and the stop.

Dirk Elzinga wrote:
>So they really aren't voiced geminates after all! (whatever they may
>be underlyingly).
        Indeed.  Historically in this group, all C were geminated following
*schwa, which is the main source.  Synchronically, it makes sense to view
all the geminates as ?+C clusters-- then you can describe Buginese syllable
structure as CV (N,?).  BTW for many speakers, voiceless geminates are
pronounced either C: or ?C, pretty much indiscriminately.  Van der Veen's
Tae' (Toraja) dictionary gives some forms with CC, others with ?C, but my
informants didn't distinguish them.  Possibly VDV's informant(s), or the
language in the teens-20s, did.  Historically, it didn't seem to make any
difference.

>> Interestingly, in Buginese, historic *_nasal+vl.stop_ > geminate vl.
stop,
>> while *_nasal +vd.stop_ > nasal+vl. stop.   With a distressing number of
>> exceptions, presumably inter-language or Malay/Jav. loans.
>
>Is this only morpheme-internally or also at morpheme boundaries?
>
There are some examples for morpheme boundaries, which may be fossilized:
gau? 'work' :: mangkau?, jaji 'become' :: mancaji,  (w)ul@ng 'moon' (*bulan)
:: ul@mpul@ng:/@eng 'menses', watang 'trunk, main part' + bone 'name of a
region' ::  watampone (the main town of that region).  From a student's
thesis (on another dialect) I culled the odd:  jambajjoli? 'diarrhea' <
jambang 'toilet' + joli? 'fast'.  (jambang evidently < Ml. jamban id.) Aside
from compounds like these, word-final N + vd.stop would be rare, there's
only one suffix I can think of that begins with a vd.stop and I don't recall
what happens in that case-- I wasn't concentrating on the grammar!

Word-initial is another matter.  Historically there were two main verbal
prefixes, *maN- and *mar-;  both fell together as /ma?-/ before vl. stops;
*mar- would have > /ma?-/ before a vd.stop, and it seems that *maN- has too
(apparently at least by 150 years ago, when Matthes did his Bug.
dictionary)-- and yes, it produces a ? + vd. stop cluster. (Mangkau? and
mancaji are the only forms I can think of that still show the *maN- prefix.)

Other sound changes got in the way too:  generally *b > Bug. w, *d > r, *nd
> nr [ndr], while *j and *g were not common.  Present day, there are lots of
alternations /w.../ ~ /ma?b..../ and /r.../ ~/ma?d..../ (or /marr.../-- but
/?w/ or /ww/ are verboden)