Re: Consonants as source of vowels

Henrik Theiling wrote:

> In order to achieve part of this, I thought that for each of the three
> interesting vowels, /a, i, u/ and possibly for /@/, too, I should find
> one or two corresponding consonants for each one.  In certain
> phonological contexts, the vowel will appear, on others, the
> consonant, and yet in others, the consonant will generate interesting
> mutations in clusters.  That's the basic idea.  One thought is to
> split the current diphthongs /ai/, /ui/, /au/ and /ua/ (all falling)
> so that one group has a phonemic vowel at the end, while the other has
> a phonemic consonant.
Interesting ideas. But mightn't the V > C part result in some unwieldy Cons.
clusters??
>
> An obvious vowel-consonant correspondance might be something like:
>
>    /i/ ~ /j/
>    /u/ ~ /w/
>    /a/ ~ /?\/
as well as /@/ or better /1/ -- /G/
>
> Now, a more interesting correspondence seems to occur in some frequent
> mutations, like in French, where e.g. |ct| -> |it| (as in |fait| <
> |factum|).  This also happened in Portuguese (|leitor| < |lector|).
I suspect this had to do with (1) gemination (factu- > fattu-), then loss of
gemination but the lost mora is replaced by a high vocalic glide viz. [j]
occasioned by the upward movement of the tongue. One might generallize this
idea to: geminate labial > [wp], geminate velar > [@k]
>
> I might, therefore, use
>    /i/ ~ /k/
>
> And I could also have
>    /u/ ~ /k/
> as in Portuguese |doutor| < |doctor|.
It's a possibility, but I wonder if that particular word is in the same
category. It may be a late learned loan, since every other *-Vkt- sequence
> -Vit- AFAIK; muito < multu-, oito < octo, feito < factu- etc.
>
>> Question 2: What other vowel-consonant pairs are feasible in such a
> way?  I'm searching for any not-so-common and interesting
> correspondence of a vowel and a consonant.  E.g., as here, in shifts.
> Example: (Pre-)Ancient Greek's shift /a/ < /n=/.
Not sure this is relevant, but I've seen written forms where "e" represents
the modern reflex of *r, *l; it implies, I guess, a palatalized approximant
stage.  There's also the familiar /l/ > [u] change via velarization. So you
could have a [-aw] ~[-al-] alternation.
Or [r\] > @ > length, so [-a:] ~[-ar\-]
>
> Question 4: More generally: what interesting phonological phenomena
> have you seen in languages?
Some of the changes just mentioned have been observed in various Austonesian
languages, especially -am/-ap > -om, op and -at/-an > -et/en; also -ut
> -uit, and -ang > -aJ. Another case-- i/j by later sound change could
become associated with [z], as mcoi [mt_soj], suffixed mcoz/a /mcoi+a/ the
word for 'dead, die' in some Formosan language (AN *matay).

Reply