Re: Radical-Metathesis

Mr. Veoler wrote:
>
>To feed the list a little... My sketch 8.1.5 uses radical-metathesis,
>that is, the root have the shape CCC, and the order of the consonants
>indicate
>a morphological category.
>
>For example:
>
>rbl > rlb > railabi "know"
>rbl > brl > bairali "learn"
>
>Does it exist any natlang that exhibit this feature?
Not quite the same, but several languages of E. Indonesia (islands east of
Timor) use metathesis, though it seems to have more to do with _preferred
sequences of sounds_ (euphony?).  There's an article on Met. in Leti in the
Rutgers Optimality Archive; if you have access to a good Univ. library,
there's also an excellent recent book "Leti, a language of SW Maluku" by
Aone van Engelenhoven, of Leiden and a native speaker. If you search his
name on Google you will turn up more of his articles.

Another lang. of Timor-- Atoni also called Dawan and Timorese, perhaps
distantly related  to Leti & Co.-- may use met. (of vowels) for grammatical
purposes, but there's very little in print. I seem to recall a paper by
Nicolaus Himmelmann-- I can give you a ref. later.  From my own notes I
rrecall only teun 'three' vs. tenu 'third' (or vice versa??)-- the base
should be /tenu/ < AN *telu. But almost every form has CVCV ~CVVC
alternants, and the -VV- sequences undergo changes.

That is similar to what happens in Rotuman, apparently an early off-shoot of
the Polynesian family; you may find info online. IIRC there's an old grammar
by Churchward.

Otherwise, metathesis seems to be a sporadic process, usually motivated by
phonological considerations. AN *limaw 'citrus' > Oceanic *limo, but pops up
as "moli" in some languages. Then IIRC there's Engl. "horse", originally
hros (?); Engl. wart, Du. wrat and others.

My newest conlang, Prevli, does use metathesis in the way you mention, I
think: CVCVC is the base form, final -VC met. > CVCCV forms realis of verbs,
and nouns, initial CV- met. > VCCVC forms irrealis. I'm still trying to
puzzle it out...........;-(

>If not, do you judge it to be unlearnable*?
>*as a productive pattern for a native speaker
In the case of the natlangs I cited, apparently not :-)))) And so far, I
don't seem to have trouble with it in Prevli.............