I just got these examplea of N- > n- and N- > k- from Ross Clark via
http://groups.yahoo.com/group/austronesian/message/1929 . Apologies
to those who are also members of Austronesian (I think at least 5 of
us are); unfortunately, its archives are not open to the public.

--- In austronesian@yahoogroups.com, "Ross Clark (FOA DALSL)"
<r.clark@a...> wrote:
> Subject: [@ustronesian network] Oceanic Reptiles (was: Apple,
> Apples of the Garden of Hesperides and)
> > > > However, in Fiji and Samoa the word for "snake" is "ngata",
> > > > Proto-Oceanic *mwata
> > Hawaiian /naka/ is exactly cognate with /ngata/. It's the
> old "snake" word,
> > used by people who don't have snakes in their environment. Can
> refer to
> > a land shell. The Marquesan word is /nana'a/ or /kaka'a/
> on
> > dialect, cognate with Maori /ngaarara/, an East Polynesian word
> lizard.

Marquesan has k (northwest dialect) or n (southeast dialect) from
*ng, and all dialects have glottal stop from original liquid. So the
original Marquesan form would be *ngangara. As you can see, this is
not an
exact match for the Maori form. The other half dozen languages that
cognates agree with Maori in having ng-r-r rather than ng-ng-r. So
I'd go
for PEP *ngarara, with a minor innovation in Marquesan.

Ross Clark
--- End forwarded message ---