Theiling Online    Sitemap    Conlang Mailing List HQ   

Shady Austronesian Linguistics (longish)

From:Roger Mills <romilly@...>
Date:Thursday, July 20, 2000, 3:38
I though it might be useful for all those interested in this project (?) to
have a little info on early Polynesian phonology, to give an idea of the
constraints we're up against. Apologies for the length.

Polynesian Mini-seminar
Proto PN had this sound system:    p t  k ?,m  n N,f  s  h, l r, w
Vowels:     i, e, a, u, o;   Syl.Structure    (C)V...

*p reflected merger of AN *mp/mb; *t < t/nt; *k < k/g/Nk/Ng; *f < p/b; *h <
s/z/c/j; *s < ns/nz/nc/nj; *l < l/r, some d; *r < d/nd. *? < q (uvular?),
evidently >glottal early on.  0 reflexes of AN *R (velar fric.) and *h.
Vowels *i, a, u generally < AN *i, a, u. *e regularly only from AN *-ay, but
often < *iCa/aCi sequences; *o < AN schwa and *-aw, sometimes, like *e, from
*uCa/aCu sequences.  A lot of PN *e (vis-a-vis AN) is of uncertain origin.

Proto Oceanic is the probable ancestor of all Melanesian, Polynesian and
most Micronesian languages, not yet clearly subgrouped except for PN: The
immediate ancestor of PPN seems to have included Fijian and Rotuman (an
aberrant and very interesting language– “Wier’s Law” operated).  POC still
retained final *p
t k r l s m n N, maybe *q,  and some of these are still reflected  in Fi/PN
derivatives, though not always etymologically correct (“centuries of
analogical re-shuffling”, as Hockett puts it).

An interesting feature of Fi/PN is the large number of doublets, where one
will have the “oral” consonant, the other the “nasal” (e.g. *f ~ p, *h ~ s
in either initial or medial position)-- generally the “nasal grade” form has
a semantically “heavier” or “intensified” meaning.

Tonga and 2 others apparently separated first; they retain reflexes of *?
and the *l/r distinction.  Then there is “Nuclear PN”, Samoan and 4 others
including Easter Isl.-- this group loses *h, merges *l/r > l, retains *?.
Out of this came “Eastern PN”-- 6 langs. including Rarotonga, Maori,
Tahitian, Marquesan and
Hawaiian.  These have lost all trace of *?, shifted *s > h, and tend to mix
up reflexes of *f with those of *w or *h (probable dialect borrowing).

Hockett in 1975 assigned dates of approx. 1200-1000 BCE to Proto Fiji-PN,
500-200 BCE for Proto PN and NPN, 300 BCE-700 CE for EP, with Easter
breaking off a bit before that. I think more recent archeological research
has pushed some of these back by several centuries at least.

SO...our Roman galley, arriving in the neighborhood of Tahiti sometime
around 0 BCE/CE, would most likely have encountered speakers of Eastern
Polynesian.   Latin /p t k f w l/ would surely survive the contact. /s/
might be reintroduced (also < Greek theta), or it might follow the natives
and > /h/.  Latin /h/, I’d like to think, would be adapted to /?/ (or
perhaps /k/ could > ?, as happens in Tahitian and Haw. I’m partial to
glottal stops).   /r/, as we’ve apparently decided, gets lost, as do all
final C.  Simplification of consonant clusters could get interesting.  Long
vowels could be retained– PN langs. have them too, phonetically
(phonemically /VV/).  Accent? PN is strictly penultimate; so Lat. cantare >
katáa?? káta?? (or
would cano/canere still have been used– káne??  Actually, which form of the
verb would be chosen-- 1sg,, 3sg, infinitive less -re?, and which of nouns--
nom. or oblique?).  Hmm– perhaps verbs could have the long vowel, and be
nominalized by shortening the V and shifting the accent, katáa ‘sing’, káta
‘song’? (Or, loss of the nasal in the stressed syllable could cause
compensatory lengthening here, káata?)

It might be useful, to avoid too much homonymy, if Lat. /d-/ > t rather than
l or 0 < **r– so tike < dicere.      QUESTION:  had Latin at this time begun
to affricate /k/ and /t/ before /i,e/? tsise?  tsitse?  sise? (t> ts before
i is found in PN).  Ego/aku might blend > eku or ako?  Porcus/fuaka > pOka
rather like exaggerated NYC pronunciation of "porker", or fOka etc.
(assuming /o/ was [O] before /r/.)  It’s possible a seven vowel system could
develop, i e E a u o O.

Latin/Greek/Aramaic? influence might introduce M/F distinctions (though
natural gender only, I think), in which case we can have articles le/la; lu
or lo for non-M/F.  At the same time, Poly-Latin could adopt the PN
alienable/inalienable possession contrast, perhaps even using the PN poss.
suffixes: Latinate calque “lo i le ?afe  kaha (or, tomu)” ‘his house’ = {the
he habet house}, or maybe blended, lo ?afee/na kaha {the habere-his house}
vs. lo manu/na ~ lo manu/ku ‘his/my hand’.

Lat. _in_ would reduce nicely to /i/ – matching  EPN *i ‘locative particle’.

I like _haruspex_ > ?alufe.  There’s a similar form in eastern IN, some MN–
but it means ‘rat’. Uh-oh. (I'm not sure it survived into PN)

_animal-_ > animau (that velar /l/) or maybe, nimáu, nimÓO (unstressed
antepenults might tend > 0).  Do we want /l/ to be velarized all the way to
/w/ in every position, or would final only suffice?

For the grammar mavens: Tongan (though not others, really) is said to show
signs of being (or having been) an ergative language.  Since the Latin
passive probably won’t survive, maybe....?  Actually PN has something called
"passive", but it's more like what we've been referring to as the
anti-passive, I think.  More research is needed here (I need to find a
grammar of a/any PN language).

About the only morphology that survives in PN languages is: *faka-
‘causative pfx.’, *fei- ‘reciprocal pfx.’; *-(C)i or *-aki ‘transitive sfx.,
often with locative sense (do VERB to...)’; full and partial reduplication;
compounding.  There are semi-fossilized prefixes *ta- ‘accidental’ and *ma-
‘stative/adjectival’ which
could perhaps be revived.  Of course Latin could contribute fe- (per- or
prae-), ko- (cum-), a- (ad-), af(a) or a(f)- (ab-); e(k)- or ek(e)- (ex-).

Surely PN niu ‘coconut’, fala ‘pandanus’, talo ‘taro’, (?)ufi ‘yam’, tofu
(!) ‘sugarcane’ would survive.
Toss up between ahu/kane ‘dog’, au/apou (arbol) ‘tree’?   Perhaps the ship
had a couple  kato aboard– though kato often turn out to be disastrous to
island ecology  in the real world.

Suggestions for some particles, postposed:  pa (< -ba-) past; tu (-tus)
participial, adjective; to (-tor) agent nouns; e, fu, or ia (est, fuit, iam)
perfective; e, i or vou (est again, ire, volo) future (assuming afe(e) <
habere is for possession).  Presumably esse as copula would be lost.  There
are PN particles that could also fill these roles.  Pronouns:  blend of
aku/ego as above; tu or wo 2d per. superior, native ko inferior; i le/i la/i
lo 3d pers. could be viewed as native i(a) + (il)le. Plurals: PN
distinguishes inclusive/exclusive we, also IIRC dual and trial; this would
need work.

Minor questions:  Were Roman galleys sufficiently sea-worthy to make it so
far? Arab ships of this period probably were; perhaps an adventurous/wealthy
owner took an Arab ship from a Red Sea port?  But what were they doing in
the middle of the Pacific?   Testing out the hypothesis of that Greek, whose
name escapes me, that the world was indeed round (he was ignored at the
(And perhaps after all that time at sea, the crew were speaking Polari.)