Dirk Elzinga wrote:

>At 10:25 PM -0500 10/19/02, Anthony M. Miles wrote:
>>Malay has an interesting feature in the prefix me [[log in to unmask](C)means the
>>me-nganga /ng/ = [N]
>>me-nyanyi /ny/ = [n^]
>>I wonder why the voiceless plosives disappear rather than the voiced? And
>>maybe the [s] of simpar is etymologically a palatal voiceless plosive [c]?
>I don't know about the [s] being an etymological *c; Roger might have some
ideas on that.
Unfortunately, it's just one of the mysteries of the Austronesian family.
Dempwolff originally reconstructed *t' (palatalized t) where most of the
languages have /s/-- some few have a distinct reflex, e.g. in one case
(lang. forgotten offhand) this *t' merged with *t; in one Melanesian lang.
*t > /?/ but *t' (along with all the other "palatals") > /t/.

Dempwolff had these voiceless stops:
*p t t.(retro.) t' (pal.) k' (pal.) k , each with a voiced counterpart--
*b d d.         d'         g'           g.
  One of his reasons for doing so was that, with this system, he could state
a rule that "only voiceless stops undergo Nasal Substitution; and all stops
can undergo "Nasal Accretion) [i.e. > NC] in medial position" (the official
term for the processes seen in Anthony Miles' list).  When the stop system
was "adjusted" and made more typographically familiar to US linguists, we
ended up with:
* p t   t. s  c (pal.affr.) k, each with a voiced counterpart --
* b d d. z  j                g
(a rather out-of-whack system, since *z now represented a pal. affr. [dZ]
and *j became something of a Mystery Phoneme with no clear phonetic
correlates (it's reflected variously as /g r l d y z 0/ inter alia.  Well,
they're just symbols after all)

The nasal subst. rule can still be stated, in Feature terms, a little more
elegantly, as "Only true (voiceless) consonants (or Obstruents) undergo
Nasal Subst.".

Original AN *s (or whatever you call it) must have had some palatal element
to it-- at least in the Oceanic languges, all 4 of the reconstructed
"palatals" merge, usually > /s/; similarly, the possible "accretion" NCs all
merged to a single sound, reconstructed as OC *ns but seldom reflected as
any kind of nasal or cluster in any OC language. (Fijian for ex. has
prenasalized /mb, nd, Ng/ for original N plus stop, but the reflex of *ns is
/s/-- the reflex of *s, curiously is [] written "c")

Interestingly, languages of Taiwan show almost no evidence for the palatal
affricates *d' or *k' (*z, *c,= Malay [dZ, tS] )-- one of the reasons for
positing a distinct "Malay-Polynesian" level, which includes essentially all
the languages outside Taiwan-- and implies that the affricates must have
developed post-"break-up of the ProtoLang." (In the 30s, Dempwolff had
little reliable data from Taiwanese langs.)

In those Indonesian languages that still retain a /s/::/c/ contrast (Malay,
Javanese, Sundanese, Madurese and a few others) the nasal substitute for /s/
is indeed //; only Javanese IIRC also has meny-(c).... in at least some
cases.  In the many langs. where *s and *c have merged (invariably > **s),
the nasal subst. is /n/ (e.g. Tagalog surat , manurat 'write', Ml. surat,