On Thu, Nov 09, 2000 at 01:43:30AM -0500, Roger Mills wrote:
> Eric Christopherson wrote:
> >On Tue, Nov 07, 2000 at 01:23:06AM -0500, Muke Tever wrote:
> >> Are "regular" metathesis changes common?
> >> Hadwan A has *[affricate][liquid] -> [sibilant][stop][liquid], basically
> >> meaning stuff like */tsr/ becomes /str/.
> >
> >I really like that change :)>
>         But a little unnatural, unless the /ts/ is a cluster, not a unit.
> (Imagine Engl. /c/ [tS] "undoing" itself to become [St].....?)  Still, it
> could result from historical changes:  suppose proto-clusters *t+s+vowel
> and *t+s+r+vowel; (Rule 1) the cluster *t+s metathesizes /__r;  (Rule 2) the
> cluster *t+s > unit /ts/.

But we've talked before (at length!) about the old Greek /zd/, arising from
sources such as */dZ/, so you never know...

> >So far I've been thinking of a view different kinds of regular metathesis
> >for Dhak, one of which I've told you (Muke) about already:
> >
> > #CCwV > #CuCV, #CCjV > #CiCV, where CC is a combination deemed to be
> >unpronounceable (or impronounceable, or maybe impronuntiable!) at the
> >beginning of a breath group (I haven't worked out yet which groups will be
> >permissible).
> >
> > Examples: ksabu > ksObu > ksO:b > kswOb > kusOb
> >           btari > btEri > btE:r > btjEr > bitEr
> >
> >I think it's pretty cool, but would anyone care to comment on
> >how realistic it is?>
>         Nice.  It also admits of several analyses.  One possibility:  ksabu
> > ksaub (met.!!) > ksO:b then / kswOb > (2d met.!!) kusOb, as you have it.

I think the a>O change is going to be due to assimilation, not metathesis;
however, this allows for more interesting metathesis with glides at that

ksabwi > ksawbi > (option 1) ksO:bi (the glide blocks the -i from
influencing the a)
                  (option 2) ksO:bi > ks9:bi > ks9:b (9 being X-SAMPA for
{oe} ligature)

> Or:  ....ksO:b > k@sO:b (cluster breaking) > kusob (assimilation)
> Or:  ...ksO:b > k[w]sO:b (anticipatory rounding) > kusob

Good work... I've thought of similar possibilities before, and yes, there
is definitely more than one way to get the result I'm talking about. Perhaps
I could leave the exact details up to conlinguists :)

> Both these strike me as quite natural possibilities, and are similar to an
> apparent metathesis in a group of langs. in eastern Indonesia that I've
> mentioned before. E.g. ku- '1st sg.' + dávar (made-up verb, can't find
> dictionary) > kdwávar 'I (whatever)'.  Presumably the rounding of the
> prefix's /u/ carries over onto the initial /d/, then the pretonic /u/
> deletes.  Similarly, mi- '2nd pl' + davar > mdjavar.

That's really cool to hear. Interesting that it goes the opposite way. I
seem to recall something in Navajo involving somewhat regular morphological
metathesis also, and I remember reading on the list that Georgian had a
metathesis of /o/, which later became a consonant or semivowel (spelled <v>,
not sure how it's pronounced).

> ObConlang:   Kash has quite a bit of metasethis ;-), mostly involving /r/ +
> cons. e.g. /axar/ 'pain' + -mi 'my' > /axamri/ pron/written "ahambri" 'my
> pain'; /amar/ 'age, era; eon' + /-kale/ 'adj. formant' > /amakrale/
> 'eternal'.   As well as some fossilized forms like /liweNgi/ 'throughout,
> for a period of...'  < /liwek/ 'long' + -ni  '3d sg. poss.', lit. 'its
> length'.


Eric Christopherson / *Aiworegs Ghristobhorosyo