Re: [tied] Bader's article on *-os(y)o

From: Rob
Message: 32799
Date: 2004-05-20

It has been hard to keep up with all of the activity on here. This
is my first reply of many.

--- In cybalist@yahoogroups.com, enlil@... wrote:

> It's changed a lot in the past year. I now think that there were
> always thematic and athematic nouns in IE. IE *wodr "water" and
> *xewi- "bird" would derive from eMIE *wat:en and *hewai while
> *oxuyom "egg" was from MIE *hawayana (technically 'thematic').
> However both had the _same_ endings in their declensional endings
> right up until the last period of pre-IE when things changed.

I guess it depends on when you believe IE began.

> What you're asking about is when thematic and athematic declension
> became different. That I believe occured during early Late IE
> (c.5000 BCE). By this time most (but not all) thematic nouns became
> athematic because of loss of MIE unstressed *-a (aka Syncope). Some
> words in participle *-na- (> *-no-) survived via the Suffix
Resistance
> exception of Syncope or through phonotactic repair of otherwise
> 'odd' syllables. Other thematic nouns were created only in eLIE such
> as those caused by Nominative Misanalysis of genitival nouns in
> *-os and *-om as new nominative animates in *-o-s and
nominoaccusative
> inanimates in *-o-m. This was the start of adjectival declension
> which coincidentally adopted an animate *-s and inanimate *-m.
>
> When the thematic nominative was going to collide with a potentially
> identical plural and genitive singular, the plural kept its vowel *e
> unexpectedly to compensate and the genitive took on a new ending
> *-s-ya with postposed endingless locative *ya "at which" (> *yo-).
> The same retaining of vowel occurs also in the genitive plural
> *-o-_o_m to distinguish it from accusative *-o-m.

Yes, the genitive plural is a big oddity. I have not even begun
attempts to explain it.

> So eLIE was the start of nominal differentiation of thematic stems
> from athematic ones. But verbs are a different story.

I agree that the 'thematic vowels' of nouns and those of verbs have
very different origins.

> The verbs were also always athematic (MIE 3ps *es "is") or thematic
> (*kWera >"creates"). In late MIE it became *es-t& and *kWer&.-t&
> respectively by affixing demonstrative *ta to the 3ps and 3pp.
> Syncope regularly produced *est and *kWer&t (since MIE *a > lMIE *&
>
> eLIE NULL but MIE *e > lMIE *&. > eLIE *&). Final Voicing caused
> voicing of *-t to *-d in final position but not in *-ti where *t
> was medial.

If I may ask, where is the evidence for 3sg *-d?

> Schwa Diffusion in mid Late IE caused lengthening of *& before
voiced
> segments so *es-t and *kWer-&.-d (yet indicative *kWer-&-ti). During
> Schwa Merger of *& to *e and *&. to *a, analogy restored
voicelessness
> in *-d based on indicative *-ti and the thematic vowel of the
indicative
> was used in thematic stems. Thus *es-t and *kWer-e-t (based on the
> perfectly regular indicative *kWer-e-ti).

My observation has been that *o is very common before nasals. In the
thematic declensions, *o exists only before nasals: 1sg *-om(i), 1pl
*-ome(-), 3pl *-ont(i). Everywhere else, the vowel is *e. However,
*o also seems to be prevalent before nasals elsewhere.

> Complicated? Yes, I'm afraid so. It can't be any simpler. I tried.

On Piotr Gasiorowski's website, he lists aorists "derived from
duratives" as having an accented thematic vowel. I think that
perhaps all original aorists were this way. E.g. there was a basic
distinction between *bheugt "is/was fleeing" and *bhugét "escaped."
The durative forms had the accent shifts typical of the athematic
paradigm, while the aorist forms kept the stress on the "suffix"
(otherwise, the plural aorist and durative forms would have been the
same). In historically reconstructed PIE, things have become very
muddled. The ancient accent distinction had become a relic, with new
inflections being the temporal augments (*e- and *-i), the sigmatic
aorist, reduplication, etc. Furthermore, some earlier durative-
aorist pairs had sufficiently diverged in meaning to be treated as
completely separate concepts. In this light, I think it's at least
possible that the conservative-accented thematic paradigm (e.g.
*bhéromi) was a later innovation.

As a side note, I have a theory whereby the Greek "secondary aorist"
is actually primary. E.g. ephugon 'I fled' < *e-bhugóm, elipon 'I
left' < *e-likWóm.

> ? Do you know about the acrostatic pattern? The thematic stem accent
> _was_ regularized to the initial syllable in both verbs and nouns.
> So the accent in *bher-o-nti "they give" is on the first syllable
> of the _stem_, always. However, on the _suffix_ in athematic *?s-
onti
> "they are".

Yes. The question is, why was it regularized? Paradigmatic
levelling? That is, otherwise the forms would have lost any common
phonetic basis?

> In nouns, thematic *wlkWo-s has initial accent, *wlkWosyo "of the
wolf"
> _still_ has initial accent. Yet athematic *kwon-s and genitive *kun-
os,
> the latter with _final_ accent. Get it? Thematic stems don't
preserve
> original accentuation.

Yes, I do get it. The form *wlkWos, with syllabic (zero-grade) *l,
surely means that it was earlier ending-accented: *wlkWós. Any
thoughts as to why the accent was retracted? It again seems like a
case of paradigmatic levelling to me, but beyond that I cannot yet
see anything.

> That's why I mention the rule of "Acrostatic Regularization" in mid
> Late IE that forced the accent firmly on the first syllable in
> thematic stems to avoid their accent flip-flops. Athematic nouns
> were relatively rarer than thematic nouns and their "flip-flops"
> continued on.

Right. By that time, the (remaining?) athematic nouns had become a
relic class.

> So, only athematic accent patterns can show my QAR pattern which
> technically now allows MIE to have a penultimate _and_ an
> antepenultimate accent (third-to-last syllable).

Right. I agree that the athematic nouns (and verbs) must have
preserved an earlier state of affairs.

> Early IE Phonotactic Constraint and Suffix Resistance.
>
> Early IE must have had allowable syllables versus syllables that
> weren't allowed in that language. Every language has rules on what
> a normal syllable looks like in it but every language is different.
> For example, zdr- is normal in Russian but completely unallowed in
> English. So I think that in eLIE, one thing that was unallowable
> was noun stems that were monosyllabic and ending in a vowel. Only
> pronouns like *ta- "that", *ka- "this" *kWa- "which"
(interrogative),
> *ya- "which" (relative), *e- "him, her, it" and *sa "that" could
have
> a simple form like CV. Another restriction was having three
consonants
> in a row. So nouns of an MIE form *CVCCa-sa in the nominative would
> have become postSyncope **CV:CC-s (with -CCC!!!) but since it wasn't
> allowable, it resulted in a preserved vowel between the stem and
> the unexpectedly 'clipped' nominative: *CVCC&-s. Hence some thematic
> nouns escaped the Syncope chopping block by Phonotactic Constraint.
> One of them might be *gHaido- "goat". Obviously these constaints
> changed in IE over time so that *-CCC was later allowable as in
> *bHe:r-s-t, the aorist 3ps of "carry".

This seems reasonable to me. Combinations of three (let alone more)
consonants seem to be very rare in historically reconstructed PIE.
To my knowledge, in such combinations (where they occur), at least
one member is a resonant or *s (as in the sigmatic aorist).

Let's take *wlkWos as an example. The root-form seems to be *w-l-
kW. Inserting unknown vowels gives us *walakWa. Penultimate accent
then gives *walákWa > *w@...@ > *wlakW > *wlekW. However, *wl never
occurs in PIE to my knowledge, so perhaps there was metathesis,
giving *welkW. Why, then, wasn't the form *welkWs instead of
*wlkWos? Aha, perhaps because of that rule you mentioned, Glen,
whereby (final) clusters of three or more consonants were not
allowed. Let's say this rule existed before *á > *é, so *walkW sa >
*walkW-sa > *walkWs was not allowed. Or is it more likely that a
schwa still existed after *walkW, which then became accented in the
compound *wálkW@-sa to eliminate the final tripartite cluster? This
would give *w@...@ > *wlkWá:s (compensatory lengthening when a post-
tonic syllable is lost) > *wlkWós?

> Suffix Resistance, on the other hand, preserves final vowels in
suffixes
> of the form -CV which aren't allowed to lose the vowel since it
would
> cause asyllabicity. Another big no-no. Morphemes must retain their
> syllabicity, except in rare exceptions such as nominative *-sa > *-
s,
> 3ps *-ta > *-t/*-t-i and inanimate *-ta > *-t > *-d. Those
exceptions
> I call "Clipping" and seem to _always_ involve suffixes derived from
> postfixed demonstratives.

It seems to me that thematic nouns had two sources.

1. "Phonotactic Constraint," as you mentioned above, to ensure that
clusters of three or more consonants did not occur (in word-final
position), e.g. *welkWs vs. *wlkWós.

2. Genitive adjectives, which I believe could encompass nominals of
the -tó, -nó, -ró types.

- Rob