Yet again: syllabicity

[CONNOLLY at LATTE.MEMPHIS.EDU]

[ moderator re-formatted ]

>>> Pat responded:

>>> One of the phenomena I believe I have identified in early language is that
>>> the plural morpheme was, at one point, simply stress-accentuation.

>Leo objected:

>> How odd then that the plural veb forms are not identical to singular ones
>> escept for the accent.  Ditto nouns.

>Pat responds:

>Sometimes I believe you are merely pulling my leg. The phonomenon I think I
>detected in early language has nothing to do with singular or plural verbal
>inflections, which are much later.

No, I'm not pulling your leg.  I do wonder, though: is "early language" a
distinct entity of some sort?  I had not taken it that way, thinking you meant
it rather as a collective for "early languages".  Unless, of course, you were
shifting the stress to the second syllble to indicate plurality: early
lan-GUAGE.

>The data I think that may support this hypothesis are the different forms of
>the (e.g.) IE verb for momentary and durative: durative *'CeC(V) and
>momentary *C0'CV (Lehmann 1974:186). Since early transitive verbal roots can
>be analyzed as *CV (object) + *CV (verbal idea), a pluralization of the
>verbal idea indicates repetition leading to perfectivity; the pluralization
>of the object indicates multiple objects leading to an interpretation of
>imperfectivity.

Unless you posit humongous numbers of consonant and vowel phonemes, don't you
then have a problem that the total number of phonetically different roots would
have been (C+1)(V), i.e. the number of consonants + 1 (for no consonant at all)
times the number of vowels?  Awfully limiting.

>Pat continued:

>>> Yes, I believe that vowel retention is generally a function of
>>> stress-accent but I find the explanation that *bherete "had three syllables
>>> with stress-accent" ununderstandable in terms of what I think of as
>>> stress-accent.
><snip>

>Then why write "had three syllables with stress-accent"? Spell out what you
>mean to say so that I will not have occasion to misinterpret it.

You have said that V (or e/o, or ^) is dropped unless under stress accent.
_bherete_ had three e's.  Hence: three accents.  That's what I meant.

>There can be many explanations for the retention of the principally
>stress-unaccented vowels as you well know. The likeliest is that the period
>during which lack of stress-accent caused zero-grade stem forms had passed.

But then, what happened to the expected *_bhr.te'_ from the period while the
rule still operated?  Such forms abound.  The Germanic strong preterite, which
reflects ablauting PIE perfect formations, still shows weak grades in the
plural.  So retention is no explanation, although a new analogical formation
would work -- that is, if one could find an old form with plural e-grade to
provide the analogy.

>Another: the necessity for differentiation of 3rd sing and 2nd pl. overrode
>normal patterns (if still operational). Etcetera.

With different grade of the root, there was surely differentiation.  It's the
e-grade of the root and the thematic vowel that requires explanation, not that
of the suffix.

> <snip>

>Leo, on another subject:

>> How so, if you are also claiming that PIE was ergative?  In ergative
>> languages, the morphological subject of an agent-patient sentence (barring
>> "antipassive" or the like) is precisely the patient, not the agent.

>Pat responds:

>Are you just trying to create confusion? There is no law that in an ergative
>language any agreement markers on the verb *must* refer to the patient.

[stuff omitted]

>If the athematic primary and secondary endings referred to the
>ergative (in an ergative context) or the nominative (in an accusative
>context), i.e. where agentially referential, there would obviously be
>no need for an inanimate *-d(i). The only traces I see in IE of a
>reference to the patient in IE inflections is in the element *-dh(V)-
>occurring in the middle (PL T[?]SA, 'body, self').

No, I'm no trying to cause confusion; I just think you don't realize the
implications of what you're claiming.  Some languages don't mark any NPs on the
verb, a great many mark only one, and some mark more (Basque can have up to
three).  When only one is marked, it is almost invariably the morphological
subject.  Accusative languages normally make the highest-ranking noun phrase
(by a hierarchy of "deep cases" of the sort proposed by Fillmore 1968) the
morphological subject; for an agent-patient sentence, this will be the agent.
Ergative languages normally make the patient the morphological subject.  (I say
"normally" because many languages can upset the process by passive or
"antipassive" formations, and some verbs may make unusual choices.)  When only
one NP is marked on the verb, it is this morphological subject -- i.e., in
ergative languages, the patient in an agent-patient sentence.

I should add that there are a few languages with ergative casemarking but
accusative verb agreement.  So yes, there is no *law* that verb agreement
markers in an ergative language will *always* refer to the patient; but
overwhelmingly they do.  So I'm trying to say that in suggesting that those of
an ergative PIE did not, you are proposing a typologically unusual system,
probably without realizing it.  BTW, what do you mean by "ergative" or
"accusative context"?

Pat continued:

>>>Regarding -*t and -*te, I do not believe that any grammatical morpheme
>>>in IE can originally have had the form -*C since I believe that all
>>>grammatical morphemes are originally grammaticalized -*Ce (at a
>>>minimum) non-grammatical morphemes. On this basis, both -*t and -*te
>>>must derive from earlier -*tV. In the absence of evidence to
>>>differentiate them, I assume a unitary origin.

Leo objected:

>>They have different meanings.  The null hypothesis should therefore be
>>that they are distinct, even if you are unable to find a phonetic
>>distinction. Beliefs about the shape of free morphemes have nothing to
>>do with the case.

Pat rejoins:

>Oh, yes but they do. And please, let us not get into another sterile
>discussion of null hypotheses, for which five linguists evince six
>opinions.

I'm not fond of "null hypotheses", and I apologize for using the jargony term,
which perhaps hid the point I wanted to make.  If morphemes are the smallest
units capable of bearing meanings, then phonetically identical chunks with very
different meanings should be assumed to be different, not the same.  In other
words, does a "let ball" in Tennis have anything to do with our verb _let_?
Phonetically, they're identical; the meaning is different, indeed, opposite.
They should therefore be assumed to be different morphemes.  (Historically,
this happens to be true, but that's just icing on the cake.)  Only the
*strongest* evidence of earlier shared meaning could then overturn this
assumption.

Pat continued:

>>For *te-w-to-, although we would both acknowledge -*to, I am not going
>>to be able to persuade you that a morpheme *te- could be the basis to
>>which a collective morpheme -*w was added --- in a paragraph or two
>>because you are unwilling to look beyond IE where *CeC roots are the
>>general rule. It is my belief that every IE *CeC root can potentially
>>be analyzed into *CV + *CV, and that these monosyllabic morphemes are
>>recognizable is some early languages: e.g. Egyptian <t>, 'loaf', is
>>cognate with IE -*dV, neuter formant.

Leo answered:

>>You're right: you can't convince me.  But not because I have any
>>preconceived ideas about root shapes in PIE (not that I know of any
>>*roots* that are that short).  The problem is that your semantics are
>>simply beyond the pale of anything that could be called linguistic
>>*science*.  Looking beyond PIE won't change that.

Pat responds:

>Semantics is in the eye of the beholder. For a look at what I believe are
>reasonable semantic relationships, see
>http://www.geocities.com/Athens/Forum/2803/ProtoLanguage-Monosyllables.htm#T
?O

I will look, I promise.  But I will not accept this identification.  Why not?
Because it depends on one dental stop + one unidentified vowel.  We do not even
know that either the stop or the vowel is the same in both languages.  The
words belong to different morphologial-syntactic categories: noun in Egyptian,
gender+case marker in PIE.  The only semantic correlation is that after all,
loaves are things.  This identification is, as I just said, beyond the pale of
anything that could be called linguistic *science*.  Given the limited number
of phonetically distinct roots available under your proposals, the most one
could say was that the two were *homonyms*.  But homonyms are not identities,
any more than the let ball has been let through.

Leo

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