PIE e/o Ablaut

[Stanley Friesen]

At 05:27 AM 3/16/00 +0000, Patrick C. Ryan wrote:

>As for "(true) absolute unversal(s)", I do not know of any in linguistics or
>any other field. I enjoyed Plato myself but I think it is a mistake to take
>him seriously.

Let's put it this way: the rarer something is among living languages, the
more evidence I would require to accept it in a reconstructed language.

>I believe that in the earliest Indian, /e:/ must have, at least
>transitorily, have been pronounced like English /ey/ (Trager-Smith), and
>/o:/ like English /ow/ (T-S); and the early Indian grammarians clearly treat
>these sounds as diphthongal.

>But where is the simple (uncompounded) /e/ in Old Indian? Or the simple /o/.

>It does not exist so far as we can determine.

>I think it is obvious that /e/ is an allophone of /a/ in an environment
>preceding /j/ and /o/ is an allophone of /a/ in an environment preceding
>/w/; and, as they have no existence outside of these conditioned
>environments, they cannot be considered phonemes, whether they retained
>their earliest diphthongal character or not.

When the diphthongal character is lost, so is the *conditioning*
*environment*.  Thus at that point they do indeed become independent
phonemes, namely /e:/ and /o:/.   This is standard phonology.  New sounds
become phonemic when the conditioning factor is lost.  (Now this does
produce an unusual situation of a language with more contrasts in it long
vowel system than its short, but that is not unheard of, and Sanskrit is
far better attested than many living languages, so it is hardly
reconstructed anyhow].

>[SF]

>> The fact that the vowel in 'leapt' (in English) is obviously *derived* from
>> the same vowel as in 'leap' does not make the distinction in vowel quality
>> any less phonemic in the current language.

>[PR]

>Yes, but /e/ is not an allophone of /i/, is it?

I am not sure what your point here is.  I was just pointing out that being
*derived* diachronically is not sufficient reason to deny *synchronic*
phoneme status, and gave an example of that.

>[PR]

>Regardless of its acceptance, I find it very strange. Pre- means 'before';
>if a form occurred *before* PIE came into existence, then it is non-PIE. If
>it is non-PIE, why not call it something else --- like Nostratic?

Because that implies it was ancestral to other languages as well.  Terms of
the form Pre-PX refer to *internally* reconstructed stages with no separate
descendent languages known.  Thus Pre-PIE is assumed to be *later* than
Nostratic (or whatever one calls it), but earlier than PIE,  Moreover it
has no known descendent languages that are not also IE languages, so no
other name is available for it.

>[SF]

>> Umm, what sort of example would you expect?

>[PR]

>True minimal pairs, a paltry requirement for phonemicity that would be
>undisputed in any other language.

Not at all.  Many phonemes are accepted as such *without* minimal pairs
even in living languages.  To show some sound difference is not phonemic
you have to show that it occurs in a *strictly* conditioned fashion.  If it
is not *uniformly* due to some identifiable set of conditioning factor,
then it is left as a phoneme.  This is how it is presented in all of the
best texts on phonology.

The *origin* of */o/ can be argued for PIE, but the very fact that it *can*
be argued is strong evidence that at *that* time it was already a distinct
phoneme.  If it were an allophone of */a/ then the conditioning factors
*still* should be visible, and apply uniformly to all cases.  It is the
fact that there are too many environments in which */o/ occurs, with no
identifiable commonality, that makes the sound a phoneme.

>[SF]

>> The shift **a: > *o would have
>> been a regular phonetic change, and thus would have been essentially
>> universal.  By the very nature of the hypothesis no *direct* trace would be
>> left - ALL of the old a:s would have gone over into o's, resulting in an
>> alternation between *e and *o where there had formerly been one between **a
>> and **a:.  So, in a sense the e/o alternation *is* the example.

>[PR]

>I do not think I am the only reader who will find this dizzyingly circular.

>The fact of e/o alternation "proves" an a/a: alternation?

I NEVER said that.  I was just pointing out that the lack of a direct
example in PIE does not *refute* the hypothesis, as such is not expected.

To put it another way: we have in PIE what can be viewed as the expected
result of a regular sound change, so it regularity is hardly evidence
*against* a sound change.

Certainly other evidence is needed.  My preference for that hypothesis is
based simply on the fact that it is the only origin model for /o/ in PIE I
have yet seen that has actually been *observed* to occur in other languages
(English, for example).

Other alternatives include that the /o/ is ancient, and inherited from the
preceding proto-language (e.g. Nostratic).  If it turns out that the
/e/-/o/ distinction in PIE corresponds regularly to some vowel quality
distinction in a wider group of languages, then inheritance is supported.
(Now, at present no such evidence is forthcoming, and I would be surprised
if it were, as /o/ looks recent to me in PIE as reconstructed).

>And why is that preferable to considering /o/ an allophone of /e/ under
>certain accentual/tonal conditions?

Because nobody has come up with a consistently applicable set of such
conditions that can explain all of the occurrences of /o/ in PIE.

>[PR]

>Certainly very interesting. But your example certainly does not mean that we
>should expect OE *sa:wian for ME sew (/so:/) when seowian is attested, does
>it?

Umm, where do you get that?

You seem to keep turning what I say around and making the implication go
the opposite direction from anything I ever said.

--------------
May the peace of God be with you.         sarima at ix.netcom.com