Re: syllable-word nonalignment

On Sun, 14 Dec 2008 15:27:55 +0000, Eugene Oh <un.doing@...> wrote:

>rDzongkha does something similar, IIRC. But it went a different way in that
>the consonant clusters simplified, so instead of having CVC.CVC, you have
>CV.CV.
>See Wikipedia! It can explain so much better than I can.
Actually I don't see that much on WP: is it somewhere other than the main
Dzongkha article?  But, yeah, I bet the most likely thing to happen if most
of your words have onset clusters is for these to simplify in place.

On Sun, 14 Dec 2008 14:34:40 -0500, Roger Mills <rfmilly@...> wrote:

>This is vaguely reminiscent of Leti, a language of the Indonesian Lesser
>Sundas, except there it operates on final -CVC [...]
>There's also a recent book-length study (2004) "Leti: a language of
>Southwest Maluku" by Aone van Englenhoven, a native speaker, that ought to
>be in the UC library by now.  Fascinating.
Ah, yes, you've mentioned Leti quite a lot onlist; maybe I'll have to go
check it out.

>The other thing that occurred to me:  Perhaps at an earlier stage, initial
>clusters required a vowel/schwa to be inserted; then your stage of this
>language lost that rule, with your outcome as a result.
Plausible enough.  Or this rule took hold immediately after the loss of a
diachronically original vowel in those clusters.

>But the question
>remains, how would your outcome differ from an input of /bat gudem daskal
>ti/ assuming those are possible words????
Probably it wouldn't, unless prosody depended on it maybe?  So if stress
assignment preceded resyllabification and fell on the rightmost closed
syllable then the inputs /daska lti/ and /daskal ti/ would be distinguished
as ['das.kal ti] and [das.'kal ti].

>Or, what would happen with /baC tgudeC mdaska.../??
Interesthing things ;-)  I haven't sat down and thought out anything
particular.  The boring thing would be for one or the other coda C to drop.
 Or maybe they could coalesce (/ba? tgu.../ [bat_> gu...]) or one could
reduce to a vowel feature or some kind of supersegment (/ban tgu.../ [ba~t
gu...]) or ...

>Vaguely similar are the cases in Engl. and I think elsewhere, where the
>initial C of a word gets reassigned to a preceding article-- Engl. "(an)
>adder" vs. other Germanic _Nader_ etc. [...]
Oh, yes, I should've remembered that, given I was just merging the Wikipedia
articles on all that stuff.

Which also reminds me, tantalisingly: I've heard of an instance of not quite
this but something else similar in some native North American language:
there was a clitic that phonologically bound to the previous word despite
being syntactically associated to the next one.  But I remember nothing else
about this except having seen it in Mithun's book...

Alex