Re: The Monovocalic PIE Myth (was Germans have no /w/, ...)

Ray Brown wrote:

> On Tuesday, June 8, 2004, at 11:28 , Jörg Rhiemeier wrote:
>
>> Hallo!
>>
>> On Tue, 8 Jun 2004 17:39:29 -0400,
>> Roger Mills <rfmilly@...> wrote:
>>
>>> This matter is being discussed on Cybalist at the moment!! For IE it
>>> seems
>>> to require a big stretch.  But one writer has posted a very convincing
>>> argument for Sanskrit as monovocalic. See msg. 33008 in the yahoogroup
>>> archive for a text in the 1-vowel system.
>>
>>
>> Well, it is a matter of definition whether, in a language like Sanskrit
>> or Latin, one defines [i] and [u] as syllabic allophones of /j/ and /w/,
>> or conversely [j] and [w] as non-syllabic allophones of /i/ and /u/.
>
>
> Whaaat???
>
> I can't speak for Sanskrit, but it ain't so for Latin, especially with
> regard to /w/ and /u/ ~ /u:/. It is simply _not_ predictable when the
> V in
> the combo V+vowel is a consonant or a vowel, i.e. [w] and [u] are simply
> not in complementary distribution.
>
In Sanskrit, some instances of /u/ can definitely be counted as
allophones of /w/(or, more accurately, /F/, I believe) for instance
'svapati'(he sleeps) has a past participle 'supta'.  An analogous thing
happens to roots with 'r' and 'l' in them - smarati 'he remembers' vs.
smr=ta 'remembered(ppp)'.  PIE *n= and *m= were reduced to <a>/@/, so,
for instance manyate'he thinks' but mata'thought'(ppp).