"syllabicity"

[CONNOLLY at LATTE.MEMPHIS.EDU]

Pat Ryan wrote, concerning whether PIE [e] etc. were phonemic:

>>> Unless it contrasts with another vowel, which produces a *semantic*
>>> difference, I believe it is correct to say that the phone is not phonemic.

I replied:

>> More precisely, we must say that it *could indicate* a semantic difference.
>> In English, /i:/ and /ai/ are phonemically distinct, even though _either_
>> means the same no matter how it is pronounced in that word.  Nor do they
>> indicate a semantic difference in the context /l_t/, there being no word
>> *_leet_ to contrast with _light_.  It's the semantic difference between
>> _meet_ and _might_ that shows that /i:/ and /ai/ are phonemically distinct;
>> we cannot say that phonemes ever *produce* a semantic difference.

Pat responded:

>Well, there is no set of relationships in IE wherein *CeC is semantically
>different from *Ce{sub}C. That, IMHO, demonstrates conclusively that
>[e/e{sub}] is not phonemic.

With Lehmann, I agree that [e e{sub}] were originally allophones of a single
phoneme, however that phoneme must be analyzed.  But the stress system of
(pre)-PIE changed, and that changed things greatly.  At one point, [e{sub}] was
a predictable variant of [e] under weak stress.  But later, the occurrence of
[e] and [e{sub}] was no longer predictable.  More than meaning, it is the
*unpredictability* of one form or another that requires them to be analyzed as
separate phonemes.  Once speakers could no longer know from the stress pattern
of a word which to say, it was one more fact to be remembered about the word --
which in systems which understand phonemes as *psychological* entities,
(essentially, all post-structuralist systems), is sufficient. -- Needless to
say, the contrast was not instantly exploited for semantic purposes, at least
in the sense that new roots were introduced which differed from alreathy
existing roots precisely because one had [e], one [e{sub}].  But quantitative
ablaut came to play a very large role in the verb system, as well as in the
declension of certain nominal types.  Germanic strong verbs (called "irregular"
in English) exploit ablaut to this day.  The reflexes of the difference do not
differentiate roots, but they still differentiate verb tenses.  That would be
impossible if they weren't separate phonemes!

(I should add, by the way, that most Indo-Europeanists reject /e{sub}/, and
apparantly also [e{sub}].  But the reason is what they see as lack of evidence
for it, not some theoretical problem.)

>As for [e:], I do not believe it is an allophone
>but rather the product of *He/e{sub}.

This just plain won't work.  Consider the word for 'father' in Greek and
Sanstrit:
			Greek		Sanskrit
	Nominative	pate:'r		pita:
	Genitive	patro's		pituh.
	Dative		patri'		pitre:
	Accusative	pate'ra		pitaram
	Vocative	pa'ter		pitah.

Surely the long vowel of the nominative  cannot derive from /He/: if so, at
least the accusative and vocative should have shown the same development.  The
[e:] of which Lehmann speaks is traditional PIE lengthened grade.  It occurs in
specific morphological forms, in part old, in part through later extension
(verb tenses in many languages).  It has nothing to do with laryngeals;
although I grant that sequences VH yield V: in many forms in many IE languages.
(HV normally yields a short vowel, not a long one, except in Sanskrit, which
has an abnormal number of long vowel reflexes.  "Laryngeal metathesis" seems
likely for Sanskrit, but (peace, brethren and sisteren!) not elsewhere.

[my stuff on Lehmann's /^/ omitted]

>Leo, c'mon. Give Lehmann a break. He used "phoneme" for "syllabicity" because
>there really was not a recognized term then for his analyzed parts of /"^/,
>realized as [e].

I didn't object to Lehmann's calling /^/ a phoneme.  Like any good
structuralist or post-structuralist, I accept the idea that some phonemes are
non-segmental.  My objection is only to his claim that /^/ was, for some
reason, non-segmental.  Only his notion that there were originally no other
vowel phonemes could possibly support such a claim, and even then it's
unacceptable for all sorts of other reasons.

Someone wrote:

>>>> Thus, since even an extremist monovocalic IE phonology would oppose a 3sg
>>>> in *-t to a 2pl in *-te, it must have a phoneme /e/.  This of course does
>>>> not detract from the stimulating effect of the book - just look at us!

Pat replied:

>>> With the best attempt to see this, I confess I cannot.  The difference
>>> between *-t and -*te is simply explained by paying attention to the
>>> stress-accentuation: *"-t(e) and *-"te.

I responded:

>> Won't work.  Consider the perfect active, where the 3.sg. desinence is PIE
>> -e, but the stress was on the root (as demonstrated by accent in Greek and
>> Sanskrit and failure of Verner's Law to operate on Germanic preterite
>> singular forms, though the plurals were affected; ...

Pat suggested:

>As for the 3rd sg. perfect active, -e is understandable on the same principle
>if we consider the earlier form to have been -e: (from *-He), reduced to -e in
>a stress-unaccented syllable.

I know of no evidence for a laryngeal there -- and one piece of evidence
against.  Skt. _cakara_ 'I have done' reflects PIE _*kwe-kwor-Ae_, where A is
an a-coloring laryngeal.  The short -a- of the root, where -a:- would be
expected by Brugmann's Law, is often taken as evidence for a laryngeal in the
desinence: Brugmann's Law operated only on -o- in open syllables.  But the
3.sg. is _caka:ra_ < *_kew-kwor-e_ -- Brugmann's law operated in the open
syllable.  (Not all accept this argument: the 1.sg. can have -a:-, presumably
by analogy, and some deny Brugmann's Law altogether.  But in any event I know
of no evidence *for* a laryngeal there.  Do you have any?)

>>> IMHO, the morpheme for the second and third persons, containing <t>, has
>>> a unitary origin: *T{H}O, 'tribe-member'.

>> Given the tone of recent discussions, I'd best not say what I think of that
>> proposal.

>If you have *reasons* for rejecting this as a possibility, do not hesitate to
>let us all know.

Indeed I do.  You are, in effect, positing a nominal stem (not pronominal, if
it has the meaning you attribute to it) which is shorter than any other in the
language.  Or do you mean the pronoun underlying Lat. _is-tu-d_, Gk. _to-d_,
English _tha-t_ (the dental is a pronominal desinence)?  And what is {H} in
your reconstruction?  Without a lot of explanation, your reconstruction (and in
particular your gloss) cannot be accepted.

Leo

Leo A. Connolly                         Foreign Languages & Literatures
connolly at latte.memphis.edu              University of Memphis