Re: Some Sound Changes

> >>  1. i, u > j, w / __ V     i, u > j, w / V __
> >
> >This seems a bit ambiguous wrt. /iu ui/. If the first rule takes
>precedence
> >(so > /ju wi/) I'd think it were better to include the two rules as
>separate
> >ones. You might also want to specify the 1st as applying right-to-left to
> >avoid eg. /uiu/ > /wju/.
>
>I'll try those.
It occured to me that you have /iw/ > /ju/ a bit later anyway; in this
light, the easiest choice would be to have i > j / __V__ first, and u > w /
__V__ 2nd. That would handle /Viu uiV/ too, tho not /Vui/ /iuV/. But 3V
clusters probably aren't all that common anyway - possibly not even extant?

> >>  3. V > Vj / __ Cj
> >>     V > Vw / __ C_w
> >>  5. ij, ej, aj, oj, uj > i, e, æ, ø, y
> >>     iw, ew, aw, ow, uw > ju, jo, Au, o, u
> >
> >This looks a bit like it would be better analyzed as umlaut, tho /Vj Vw/
> >coming from /Vi Vu/ would complicate that. Also, if the /j/ duplicates
> >across all consonants, why doesn't /w/ duplicate across non-velars too?
>
>I don't know how to write the rules for umlaut, so I did it in two parts,
>and also wanted to handle diphthongs. Good question about /w/, although I
>don't have an answer. I'll take a look and see what happens if I change
>that.
If all difthongs come from former bisyllabic V + (u i), the most concise
representation I can think of is:
1. a o u > æ ø y / __CiV, __iC
    i e a > ju jo Au / __CuV, __uC
2. i u > j w / __V
3.     > 0 / V__ (except /Au/)
You could simplify the first step a bit - to "/ __(C)i/u" - if you allowed
"regular" umlaut too, say /peku/ > /pjoku/, and disallowed any pesky /Vui
Viu/ clusters.

....no, wait, the uvulars & /r/ mess this up. Eg. /teqio/ should, along your
original rules, become /teqjo/, but the umlaut interpretation givs /tæqjo/
with the uvular coloring applying on top of the umlaut. Hmmmm. I can't think
of anything better than your original epenthesis to deal with this issue.

Another note, along the original rules you'd also have to specify that
simplifications only happen with difthongs; unless you *want*, say, /niue/ >
/niwe/ > /njue/. Pretty pedantic, yeah, but if you plan on implementing
sound changes automatically you'll have to be.

> >>  7. tj, dj > ts), dz)     sj, zj > S, Z
> >
> >....I take /t d/ are dental but /s z/ alveolar, then?
>
>I don't see why /t d/ can't be alveolar at this stage.
Well, because they don't palatalize to /tS dZ/ analogously to /s z/...
dentals palatalizing to alveolar /ts dz/ directly is all fine and attested
IIRC, but alveolars doing the same (i.e. without any POA shift), not so
much. /th/ > /T/ also suggests the series being dental.

I suppose you could invoke a direcly follo'ing /tS dZ/ > /ts dz/, maybe by
push of /c J\ C/ > /ts\ dz\ s\/ > /tS dZ Z/, but that would seem to imply
the /S Z/ from /sj zj/ also going back to /s z/.

> >>10. c, J\, C > tS), dZ), S 11. hj, hw > C, x_w
> >
> >A new /C/ drifting in just after the old one merged into /S/? Looks a bit
> >unlikely to me... but an easy fix would be to switch the order of palatal
> >breikdown and aspirate spirantization (i.e: ch > tSh > S)
>
>I think 11 could just occur later.
Especially since Alex noted that fricativization (oops, not spirantization
if there's an /S/ there) would have to happen before affrication for /th/ to
work out properly. Then you could also have both of your affrication steps
occuring simultaneously.

> >dongoresu > duNgworer
>duNgurer, I think
I'm going along the vowel-breikage-before-r rule: or > oar > wAr > wor

> >Looks like you'll get a contrast between /kw k_w/ there too... or is the
> >labialization rule meant to remain live?
>
>The latter, I think. I'm not sure how to specify that.
>
>Jeff
Since you seem to also have /w/ occuring before other stops, and no "sink"
for it, you could just reassign this process as allophonic or delay it until
after the newest /w/'s are live too.

> >John Vertical
_________________________________________________________________
Uutiset ja kasvot uutisten takaa. MSN Search, täyden palvelun hakukone.
http://search.msn.fi