At 02:06 PM 11/20/2004, "suzmccarth" <suzmccarth@...> wrote:

>--- In qalam@yahoogroups.com, "Peter T. Daniels" <grammatim@...> wrote:
>
> > Tamil disallows inital (high front?) vowels, and Tamil-
> > speakers prefix y- to any English word starting with e-.
>
>I have never noticed that Tamil disallows initial vowels of any kind -
>they seem to be quite common, maybe we are thinking of two different
>phenomena.

Tamil actually does do what Peter is talking about. Written
Tamil(Tamizh, since there're no aspirates in Tamil) has initial vowel
onsets, i.e. <iran.t.u> 'two', and <eel[u> 'seven', <on[patu> 'eight, but
in spoken/colloquial Tamil the initial vowels are preceded by either [j]
"y" or [v], depending on context: <iran.t.u> [jirV~du-], <eel_u> [je:lu-],
<on_patu> [vo~bVdu-],etc. (I'm using Kirshenbaum's ASCii IPA notation
(http://www.kirshenbaum.net/IPA/ascii-ipa.pdf))...

Peter's contention that Tamil is using high vowels for breaking of
disallowed clusters is probably more right than wrong, given the very very
extensive literature on loan phonology and the tendency for languages
phonologically similar to Tamil to do so.
It is highly unlikely that one will find an inserted "unvoiced"
vowel in cases where the surrounding environment is not unvoiced....
Take for instance the case of Japanese... 1. Japanese inserts high
vowels (usually <u> [u-]) for loan words that have initial consonant
clusters that are not native, 2.Japanese also has voiceless vowels.
Are these two necessarily correlated? No... (1) happens to be a
phonological process that takes place when loanwords have to be adapted,
while (2) takes place in Japanese as a language universal.
Taking two words found loaned into Japanese: "beefsteak" and
"blue(jeans)" we can examine the two processes at work.

<beefsteak>
/bi:fstejk/ (Eng.)
*bi Pu- su- te: ki
[biPu-<o>su-<o>te:ki<o>]
/bihuste:ki/

-=high vowels in Japanese /i, u/ [i, u-] are devoiced when surrounded by
two unvoiced segments, where word final boundaries (which basically equal
zero) are included.

It would then seem that Japanese speakers, whose unvoiced high vowels seem
to mimic clustering in languages like English, must then use unvoiced
vowels to represent clusters... _wrong_.

"Bluejeans" provides a nice counter... we expect that there will be an
unvoiced vowel where we find the English bl- cluster...

<bluejeans>
/bluwdZijnz/
*bu ru-: ji: n zu-
[bu- ru-: ji: n zu-]
/bu ru: ji: n zu/

>Possibly, but the argument from the Tamil seems to be that
>they don't alter the pronunciation, they simply perceive
>the near-voiceless vowel which Hindi and English speakers
>do not perceive.

Why should one posit a segment (near-voiceless vowel) in two
languages for which they are not phonemic/phonetic in native words?
It's much more likely that speakers of a language, whose structure
disallows clusters, will insert segments to allow for pronunciation, rather
than vice versa.

>The Tamil have considered but more or less rejected using
>all the grantha characters for consonant conjuncts. It is
>an ongoing debate. They won't use the pulli or consonant
>plus inherent vowel form.

That's largely due to inherent intuition... however, grantha
conjuncts were exactly that... without any required vowel intervening
between the letters... since Tamil doesn't allow for CC- onsets, of course
they'd disprefer any such analysis... no?

cheers,
-Patrick

[Non-text portions of this message have been removed]