Syllabicity

[CONNOLLY at LATTE.MEMPHIS.EDU]

>Leo writes:

>>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.

>Pat responds:

>>I think a useful distinction to be made is whether we *can* predict
>>the occurences --- limited by our imperfect understanding of the
>>stress-accentual fluctuations  ---  and --- if we had a better
>>understanding of them, whether we could predict them.

I concede that the real question is whether speakers of some particular time
could select the correct form solely on the basis of phonological environment
(including stress and whatever other suprasegmentals may have been relevant).
If they could, then [e e{sub}] were allophones of a single phoneme, and any
semantic difference would be attributable to the difference in the phonetic
environment.  Whether we could know the details is irrelevant; it's their
phonemes, not ours, and they were the ones who had to know.  And while we're
the ones with the theory, they're the ones who really know.

Yes, our knowledge comes from detective work and is necessarily uncertain.  But
we must agree that it's worth doing, else we wouldn't be at it like this.  So
the distinction would seem useful only as a way of weaseling out of the
consequences of a proper understanding of phonological principles which (if
they are to have any validity at all) must apply to *all* languages, modern and
ancient, attested or reconstructed.

[stuff omitted]

Pat responds to remarks on ablaut in Germanic::

>This is one of those questions in which it is hard to decide if the
>dog is chasing his tail or the tail is teasing the dog.

>On a non-IE scenario, a phoneme can easily be identified as marking a
>semantic difference: English cat / cot.

>In IE, we simply do not find that CeC is a semantically different
>morpheme from CoC. Now, as you have rightly identified, these
>variations *do* mark grammatical distinctions. You obviously prefer to
>define forms indicating different grammatical employments as
>semantically different; I do not because, if we did, we would be
>forced to say that cat / cats  are *semantically* different. I do not
>think you would be willing to go this far, would you?

Look, if you can't find a semantic difference between _cat_ and _cats_, I
shudder to think what your kitty-litter bill must be.  Yes, I do claim that
many (not all) so-called "grammatical morphemes" such as plural -s have
semantic meaning.  And you should too, if you think about it.  Below you make
an extraordinary claim: that all verb agreement markers were originally nouns
(there being no original pronouns).  If this were so, wouldn't these markers
have had *exactly* the same sort of semantic meaning that you postulate for
roots?

 <snip>

Pat commented earlier:

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

[Leo called attention to the long vowel in the nominative singular of Gk.
_pate:r_, Skt. _pita_ 'father.]

Pat argues:

>I have no reason to think that the analysis of this word is other
>than as a compound of the root plus the suffix of the nomina agentis
>-ter (not **te:r).

	[stuff omitted]

>I do not believe that the e: of this word derives from /He/; I also do
>not believe it is original. In view of *ma:te{'}r, and in view of the
>fact that we have no IE root of the form *p6-, considering that
>*p6te:{'}r is obviously an analogous formation, I believe the
>likeliest scenario for the long vowel is a metathesis of laryngeal
>(or, just simply, the feature length): *pV:(H)-ter -> *p(6)-te:r.  On
>the basis, the /e:/ is simply an allophone of /e/, hence, cannot be
>phonemic.

Several comments:

First, the idea that the family words contain an agent suffix, though old, is
without basis.

Second, the idea that sequences of the type VHCV (where C is a stop, not a
resonant) were metathesized to VCHV, whence VCV:, is, to say the least, novel.

Third, the *historical origin* of the [e:] has nothing to do with its
*synchronic* phonemic status at any given stage of PIE.  Whether it results
from laryngeal metathesis or through the lengthening of a stresses final
syllable of a root with no desinence (not my idea, but a good one) has nothing
to do with whether it is an allophone of /e/.

>If the long vowel of the nominative were original rather
>than a result of easily understood phonlogical developments, it
>*would* show up as more than -0- in, e.g. the genitive.

No one is claiming that _e:_ in _pate:r_ is original.  You, I, and Lehmann all
agree that it was somehow secondary.

[stuff omitted]

Pat responded:

>An /e:/ which is the result of phonological processes or morphology
>still cannot be considered a phoneme IMHO. For me to accept the
>phonemic status of [e:], I would need to see two roots: Ce/oC and
>Ce:/o:C, with different meanings. And yes, I meant to write e:/o:. If
>e: is phonemic, we should expect to see it participating in Ablaut.

Pat, I regret to have to say so again, but you simply do not understand what a
phoneme is.  Phonemes are minimal distinctive units of *sound*, not of
*meaning*.  Do study up on this.

As for ablaut, e:o ablaut is attested for traditional lengthened grade e: and
for traditional "original" e: i.e. eH.  Beside Gk. _pate:r_ 'father' we find
both _phra:to:r_ and _phra:te:r_ 'member of a clan' (orig. 'brother').  And for
Gothic _saian_ 'sow' < *seH- we find reduplicated preterite (originally
perfect) _sai-so_ < _*se-soH-_.

<snip>

[Leo asks for evidence that 3.sg.perfect -e resulted from -He, as Pat claimed.]

Pat responds:

>1) On general principles, since inflections are grammaticalized
>morphemes, and IE has no morphemes beginning in a vowel, any
>inflection that manifests itself apparently as a V, should be, ab
>origine, be presumed to be HV.

Since when is "general principle" a response to a request for evidence and
examples?  Isn't your claim that no IE morphemes began with a vowel an enormous
thesis that needs in-depth analysis?  It also seems to be a thoroughly novel
thesis as well: while others have claimed that IE *roots* could not begin with
a vowel, I have not until now seen that claim made for *other* morphemes.

Let me give you one case to consider.  What do you think of the thematic vowel
morpheme {-e/o-} (curly brackets being the proper notation for morphemes),
which is found in both nominal and verbal forms?  Is there *any* evidence for a
laryngeal there, since the morpheme was never initial?

I would also suggest that you review *carefully* the concept of the morpheme
before answering.  Many authors insinuate, or even claim, that the morpheme is
the minimal unit of *meaning*.  Though common, this is simply wrong.  The
morpheme is the minimal significant unit of word formation, and hence the
smallest unit capable of *bearing* meaning.  That's rather different.  Some
morphemes (the thematic vowel is one such) have no obvious meaning.  (This is
not to say that thematic and athematic verbs formed from the same root must
therefore have the same meaning.  Rather, we must say that root + {-e/o-} may
have a different meaning than bare root.  This is not surprising, since in
living languages, compound verbs need not have a meaning equivalent to the sum
of the parts: _understand_ is semanticly not "under" plus "stand".)

>2) For whatever interest in may be, I published in Mother Tongue an
>essay describing the differences between the person as vocalic
>differences, each proceeded by H{1}, i.e. /?/:
><http:/www/geocities.com/Athens/Forum/2803/PERSPRO1.htm>

>Please do not attribute all my views of 1990 to me now, however.

How are we to know what to believe and what not to?

[stuff omitted]

>I do not believe that the earliest Nostratic had what we would
>properly call pronouns. I believe all pronouns are only nouns in a
>specialized use.

Typologically, this is acceptable; certainly it seems right for Japanese.  But
the Japanese "pronoun" words look and act like nouns in every way, which cannot
be said of the IE set.

[stuff omitted]

Pat said:

>Yes, I believe that there was a *noun*, which would have the reflex of
>*to in IE, which meant 'tribal member', and was used in various
>positions that we would characterize as pronominal or inflectional. We
>even have an extended form of this *to in *teuta:-, 'people (probably
>better 'tribe')'.

But that still doesn't answer my question about the extraordinary brevity of
*to- (better: *te/o-, with ablaut).  Recall that if you actually accept
Lehmann's version of /^/ as a prosodic feature that comes and goes (and hence
not part of the root), the result is an absurdly short *nominal* root /t-/.
Thus *_teu-_ i.e. /tew-/ or */tw-/ looks like the bare minimum for anything
nominal, and that's pushing it a bit.

Leo

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