Backwards conlanging (repost)

This is for the benefit of B.P.Jonsson and H.S.Teoh, and anyone else who
received a garbled version.

Nik Taylor wrote:
>Or something simpler that could have been elaborated into a 5-vowel
>system.  Perhaps an earlier stage had /i a u/, and /i/ and /u/ split
>under certain conditions, open/closed syllables is one way.  Quechua has
>[e] and [o] as allophones of /i/ and /u/ when adjacent to a uvular
>sound.  Or, perhaps earlier /aj/ and /aw/ became /e/ and /o/, and the
>diphthongs in the modern language came from a later source, perhaps lost
>consonants (e.g., /ahi/ -> /aj/)>
That's all quite similar to developments in the Austronesian family; of
course they're all quite common changes.  The protolang. had just 4 vowels,
*i a u @; *i a u tend to be remarkably stable, though i/u often lower
sporadically to e/o.  But *@ goes all over the map:  retained in Malay,
Javanese, Buginese et al.; /a/ in many, /i/ in Tagalog, /u/ in Bisayan; /o/
in Batak and all Oceanic; /e/ or /o/ in various Moluccan langs.  Sequences
with one of the so-called laryngeals tend to simplify, *aqu > o or O; *aqi >
e or E (the open varieties often when there was a final C). Similarly for
the original diphthongs *aj, aw (only occurred word-final).  In well-known
langs. like Malay and Jav., contrastive /e/ and /o/ were fairly clearly
introduced by borrowing from Indic languages, and later Portuguese-- Port.
even introduced a marginal /e:E/ and /o:O/ contrast into Malay:  pEsta
'party', bonEka 'doll'; rOda 'wheel'.

An interesting situation in some Moluccan langs. I've been studying:  *@ >
e, but higher [e] if the next V is i/u, lower [E] if it's a (far more
common)-- but then there are minimal pairs, with ...E-i and ...e-a-- where
do they come from?  Sloppy informants? Careless investigator?  Loanwords?
(Probably.) A marginal, vanishing contrast?  What may have happened in the
70-plus years since the description was written?
(The very interesting lang. is Yamdena (Jamdeensch), well described by a RC
missionary named Drabbe; publ. in 1932 in the Verhandelingen series of the
Bataviaasch Genootschap if you want to take a look.)

ObConlang:  Kash has /i e a u o/, and I'm in the same fix as Yoon-Ha-- no
proto-language, and very few dialects to compare-- but I've made a start.
The present variety of Kash has quite a symmetrical sound system:
    Voicless stops/afficate:   p t c[tS] k
    Voiced  ditto:                 b d j      g (phonetically mb, nd, ndZ,
Ng)
    Fricatives:                       f  s  S    x
    Nasals:                           m n ñ    (N only in final position)
    Other:                            v  r l y    ("y" is IPA j)

Basically, vl.stops < original vl.stops; the vd/prenas. series < merger of
original *mp/mb, nt/nd, Nk/Ng.  Modern /s/ probably < *T. (Was there also
*s? Perhaps apical, the source of /S/?)  The affricates/palatals were
probably not present in the proto-lang.  Original *b > v,  *-d- and -d > r;
*g > undecided......
The nearest dialect has contrastive /N/ in all positions; phonetically, the
voiced stops are not prenasalized, and /x/ is uvular, not velar, and will
probably produce lowered allophones of /i u/.

(Added note:  after watching my cats "talk", I've decided that the Kash
(much bigger than the average cat and of course quite different) probably
have wider mouths than we, and can't actually round /u o/.  So the
difference is that front vowels have spread lips, back vowels don't.)