PIE e/o Ablaut

[Stanley Friesen]

At 02:12 AM 3/27/00 +0000, proto-language wrote:

>We are talking about whether Old Indian *at one point* had basically a one
>vowel system.

>I think the facts make obvious that there was a time, however brief, when
>Old Indian had only /a/ as a vowel, with [ay] and [aw] on the way to
>becoming /ai/ -> /e:/ and /au/ -> /o:/.

I maintain that at that time it *also* had /u/ and /i/, making it a *three*
vowel system.  Indeed, in Sanskrit it is probably harder to maintain that
/u/ and /w/ are allophones than it was in PIE (and likewise /i/ and /y/).

>Above you wrote: "the rarer something is among living languages, the more
>evidence I would require to accept it in a reconstructed language."

>If that is your position, why not put it in practice on all questions under
>consideration?

>Can you name any "living language" that has a vowel system of /a/, /e:/, /i/
>(giving you the benefit of the doubt by counting /i/ as a vowel rather than
>a vocalic allophone of /j/ in an avocalic position), /o:/, /u/, (giving you
>the benefit of the doubt by counting /u/ as a vowel rather than a vocalic
>allophone of /w/ in an avocalic position)?

1) Sanskrit is NOT RECONSTRUCTED, so the rule does not apply.

2) I believe it is more adequately described as: /a/, /a:/, /e:/, /i/,
/i:/, /o:/, /u/, /u:/.

3) I am fairly certain there are other somewhat similar languages.  (My
reference that has the vowel system survey is buried, and not currently
accessible, so I cannot immediately name any).

>If this is not "unheard of", how about letting us hear about it?

If I manage to find my vowel survey text, I will.

>>> [SFp]

>>>> 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.

>>> [PRp]

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

The vowel difference between those words *started* as an allophonic one,
and developed separately later.

>I am talking about a *synchronic* situation during which Old Indian had /a/,
>/ay/, and /aw/.

and /i/ and /u/ and /a:/ and /i:/ and /u:/.

[PR]

>Well, something has been lost in translation here. You mentioned Pre-IE, I
>thought, not Pre-PIE.

If so, then it was a typo.  The term is indeed Pre-PIE.  (Check some of
Lehman's work, for instance, which even uses Pre-Pre-PIE).

[SF]

>> Not at all.  Many phonemes are accepted as such *without* minimal pairs
>> even in living languages.

>[PR]

>You assertion by itself does not convince me.

>Would you mind citing an example of any phoneme in any language that is not
>in a minimal pair?

Most of my old basic linguistics texts are also inaccessible. But if I
remember correctly English is particularly rife with that situation.

Remember, there is a combinatorial problem: to fulfill your requirement you
must have a minimal pair for *every* pair of comparable sounds.  Even
allowing that vowels and consonants cannot really be paired, this still
makes for an amazing number of necessary pairs for English, or even High
German.

English, depending on the dialect, has between 9 and 11 simple vowels (not
counting diphthongs), requiring around 100 minimal pairs just for the basic
vowels alone!  Then there are at least 21 consonants, 23 if one counts
/dz^/ and /tc^/, 24 if one adds the glottal stop, requiring 441 to 576
minimal pairs.  So, English would require at least 522 minimal pairs (605
in some dialects), even keeping vowels and consonants separate.  If one has
to distinguish each vowel from each consonant with a minimal pair (which
your strict rule would require), one would need 900 to more than a thousand!

I seriously doubt one can come up with even 500 minimal pairs in English.

>[PR]

>Could you name a "best text on phonology", and cite a relevant definition of
>phoneme from it?

If you will volunteer to come over and sort my library :-)

>[PR]

>So, you consider that a chain shift was going on in IE?

The change of /a:/ to /o/ is not necessarily a chain shift!

Indeed, if there had previously been no mid-back vowel, there is no need
for any further shifts at all.

>[PRp]

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

Because:

a) nobody has ever produced a set of tonal and accentual conditions that
actually explains all of the /o/'s in PIE.  (I.e. no sufficient set of
conditioning factors has ever been proposed).

b) it is phonetically unlikely: I cannot think of any other well-attested
case of rounding being conditioned strictly be tone or accent.

[Note, a phonemic /o/, even originating mainly from older /a:/, need no
longer have a  visible set of conditioning factors, due to phonetic change
and analogical extension].

>[SF]

>> 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]

>I am under the impression that a consistent explanation ofIE /o/ has been
>formulated: namely, that /e'/ becomes /o/ when the stress-accent is
>transferred to another syllable.

It fails to cover some of the cases, especially in the perfect and deverbal
nouns, and it requires some variances in the "reconstructed" accent in
regard to the endings to make it work.

[Believe me, I have tried to make a complete inflectional model in which
this works, but the moment one gets into details, it breaks down: unless
one uses circular reasoning to postulate a shift in accent when no evidence
except the /o/ point to it].

>[SF]

>> 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.

>[PR]

>I believe it was your point that OE /a:/ shows up as Modern English /o:/.

Yes, but I *never* said that *all* ME /o:/ come from OE /a:/!
[Any more than all PIE /o/ would come from /a:/, some would come from H3,
and some might come from loss of nearby -u or ^w (aka umlaut)].

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