--- In cybalist@..., "michael_donne" <michael_donne@...> wrote:
>
> > Greek has three phonemically independent short vowels 'a', 'e'
> > and 'o', and three corresponding long vowels.
>
> Are the corresponding long vowels also written like 'a', 'e'
and 'o'
> but pronounced longer?
Greek has only the letter alpha for both long and short a , but
epsilon for short e , eta for long e , omikron for short o ,
and omega for long o. The long vowels differ in both length and
quality from the short. Ancient Greek's short e was the spread
close-mid front vowel, long e the spread open-mid front vowel,
short o the rounded close-mid back vowel, long o the rounded
open-mid back vowel, short a the chart's lowest central vowel
(upside-down a ), and long a the open front cardinal vowel.
Besides these differences in quality the long vowels are also
pronounced longer, that is they are of greater duration. On
those occasions where I need to actually pronounce Sanskrit
or Greek aloud, rather than just read or write them, I give the
long vowels precisely twice as much time as the short. I know
that this is probably somewhat artificial, but it's easy to mentally
hear a steady beat in your head and then hold the short syllables
for one beat and the long syllables for two. I'm not sure what the
exact difference in duration was in either language.
> So my point is: if Greek had a short 'e' and a long 'e' and
> Sanskrit's were in between, then can you really say with certainty
> that they weren't more like the short form?
If you mean "more like the short form" in terms of vowel quality
(the vowel's position on the vowel chart), then I personally can't
guarantee you of the ancient pronunciation. What I tell you about
that, I've taken from what the experts have written. But remember
that for I.E., Sanskrit, and most (all?) of the early dialects,
vowel length was the major feature distinguishing the two sets,
not the vowel quality. So in terms which are important to I.E.
you can't have a long e that's more like a short e . The vowel
is either long or short.
> The point about the meters is very telling. The Greeks had short
> and long versions of 'e' and 'o' right? How were they pronounced?
> Do we have an examples of them being used metrically like the
> Sanskrit?
Yes, much of the ancient Greek material was written in meters that
were based on specific sequences of long and short syllables, just
like Sanskrit.
> Oh, I see what you mean now. I thought you meant something else.
> You're talking about this:
> > na-ya-ti (CV-CV-CV) 'he leads, guides'
> > *nay-tram (CVC-CCVC) > netram 'eye'
>
> My problem with this, is that I think it's backwards. The root 'ni'
> which means 'to lead' is lengthened to become 'netram' or 'eye'.
> *nay-tram is an analysis after the fact.
That will work with 'ni-', but what about 'svap-' "sleep"?
zero-grade guna vrddhi
ni- ne-/nay- nai-/na:y-
sup- svap- sva:p-
'Ni-' follows the pattern i > e > ai, like you say, but notice that
'sup-' does not go u > o > au. Now reanalyze the whole thing
as modern linguists do.
zero-grade guna vrddhi
ni- *nay- *na:y-
sup- *svap- *sva:p-
Now zero-grade is that without an a , guna grade that with an a ,
and vrddhi grade that with a long a . If we start with guna as
the basic grade, we simply remove the a to get the zero-grade or
lengthen the a to get the vrddhi grade, but if we start with the
zero-grade as the basic form, then we don't know to which side
of the semivowel to add the a for guna and the a: for vrddhi.
The ancient Indian grammarians had to devise less elegant rules
to explain forms like 'svap-' which to them seemed irregular.
> Um, why is this distinction between long and short vowels relevant
to
> the discussion about how PIE merged e, a, o into Sanskrit 'a' ?
> I was going to post more about this but I think I'll wait until I
> understand better why it matters. :-)
>
> I gather that maybe the long versions of e,a,o didn't collapse into
> short 'a' in Sanskrit but behaved differently somehow?
Yes, the long versions of *a, *e, *o merge in long a in Sanskrit,
and the short versions of *a, *e, *o merge in short a in Sanskrit.
> Yes, this is the crux of the matter. Could you upload to the
> library a video of yourself explaining it with all the hand-waving?
Sorry, I don't do internet videos anymore since my nieces and
nephews got computers. (-:
> OK, so the long versions of the Greek collapse to the long
> Sanskrit 'a:'
Yes.
> And their dipthongs collapse to corresponding Sanskrit dipthongs
> (ai, au) or long vowels (e, o); is this correct?
The diphthongs with a short *a, *e, *o collapse to Sanskrit e:
or o: in a closed syllable and to ay or av in an open
syllable, but not ai or au . The diphthongs with a long *a,
*e, *o collapse to Sanskrit ai or au in a closed syllable
and to a:y or a:v in an open syllable.
> I guess the central point of this whole argument is when Hock
> says: "the correspondence of Sanskrit a- vowels to Latin and
> Greek e-, a-, and o- vowels cannot be explained by means of a
> regular sound change that "split" the 'a' of Sanskrit into the
> e, a, o of the other languages."
>
> My question is: Why not? Apart from the fact that it's really
> messy.
> 3 > 1 is certainly preferable in a theoretical sense than 1 > 3.
> Are there any other reasons? Are there any environment where this
> does not happen or where it is explainable in other ways?
Because changes in a language's sound system are never random.
Why would a become e in one word o in another, or stay a
in still a third? There should be specific conditions causing
the change to one or another, but we find all three vowels in
cognates of Sanskrit with an a before and after the same
consonants, in all positions within a word, etc. In other words
there is no discernible rule for converting a hypothetical original
a into an e here, or an o there. From this pattern we can
tell that *a, *e and *o were seperate phonemes in the parent
language. Also, since all the daughter languages in which the
three vowel distinction is still discernible show the same vowel
in the same place (cognates), they would all have had to have
branched off after this three-way split, and Sanskrit would have
to have branched off before this split. This produces a family
tree that fails to account for almost all other observed
relationships between the I.E. dialects. Also Sanskrit shows ca
where the other dialects show evidence of *kWe, and ka where the
other dialects show evidence of *kWa or *kWo. Now from studying
many languages we know that the fronting and/or fronting and
affrication of a velar, like kW, before a front vowel, like e
or i , is one of the most common sound changes known.
> Do you know of any really long lists of where this happens plus any
> exceptions?
I imagine there is some such list somewhere. My problem with trying
to make such a list myself is that I have no Sanskrit dictionary and
no Greek dictionary necessary to compile one, and I don't have enough
competence in any of the other dialects either.
> > But what relevance does the modern pronunciation have? Vowel
length
> > is not distinguished in the modern pronunciation of Latin either.
> > ...
This one begins Miguel's reply.
> OK, so now the question has changed: what about Sankrit vowels:
> i and u? What happens with them in other IE languages like Greek?
I.E. long i becomes Greek long i which is the spread close
front cardinal vowel. I.E. short i becomes Greek short i
which is slightly more open than the spread close front cardinal
vowel. I.E. long u becomes Greek long ü which I always
thought was the rounded close front vowel, but which Beekes gives
as a rounded close central vowel. I.E. short u becomes Greek
short ü which I always thought was slightly more open than the
rounded close front vowel, but which Beekes likewise gives as
central. Of course long i and ü are also longer in duration
than their short counterparts. Hopefully one of the Greek
experts on the list can clarify the pronunciation of the ü
for us.
David