Re: dhuga:ter ('LARYNGEALS')

From: Miguel Carrasquer Vidal
Message: 55746
Date: 2008-03-23

On Sat, 22 Mar 2008 11:41:00 -0500, "Patrick Ryan"
<proto-language@...> wrote:

>By the way, is it not suspiciously coincidental that the 'laryngeal' theory
>has three coloring agents that just happen to coincide with the three
>attested vowels in PIE?

It's neither suspicious nor a coincidence. It's just what we
would expect. But note that the frequency distribution of
the laryngeals is quite unlike that of the vowels. PIE *e is
by far the most common vowel, followed by *o, and the least
common by far is *a. The most common laryngeals are *h2 and
*h1, the least common, by far, is *h3.

The PIE vowel distribution is atypical, of course, and
laryngeal colouring (as well as colouring by back velar *k,
*g and *gh) helped to set the situation right by
incrementing the frequency of /a/.

If pre-PIE had a more normal 3-vowel frequency distribution,
one would expect *aH (*h2) to be the most common "laryngeal"
(/a/ would normally be the most common vowel), followed by
an about equal frequency of *eH (*h1) and *oH (*h3) (about
equal frequency of /i/ (/e/) and /u/ (/o/)). This is not the
case.

>Yes, with unstressed position, I believe the long vowels react very
>similarly to what you anticipated with the exception that I have not yet
>found it necessary to distinguish schwas
>as *&1, *&2, and *&3 according to their origin in *e: (*eH), *a: (*aH), and
>*o: (*oH).

But Greek has.


>
>> Your explanation:
>>
>> >where the vowel quality itself distinguishes the root from
>> >other similar roots (*as- from *es-, for example), the
>> >lengthened root vowel could be safely shortened without
>> >loss of semantic integrity.
>>
>> This makes no sense to me. The vowel quality, you say, is
>> original, so it would *always* distinguish the root from
>> other similar roots. You provide no examples of what happens
>> when the vowel quality does *not* distinguish the root from
>> other similar roots [can there be any in the context of your
>> theory?]. It would be simpler to say that in initial
>> position, the vowel gets shortened somehow.
>
>***
>
>It has taken me a long time to try to make sense of it myself.
>
>Mainly, because it is very complicated though not really difficult.
>
>*Hes, *Has, and *Hos produce *e:s, *a:s, and *o:s. At first look, one might
>say that all could safely be shortened since the vowel quality without
>length distinguishes them semantically (*es, *as, *os) but short vowels in
>initial root position are the normal outcome of *A dependent on the
>stress-accent;

Are you saying that PIE had words beginning with a vowel?
The consensus rather seems to be that all words started with
at least one consonant.

>so, at least some formerly long vowels that were shortened
>seem to have been misinterpreted as outcomes of *A; as you can readily see,
>this would be specially easy for *é from *é:.
>
>Another variable is whether one or both of the complementary pairs (V: and
>V) have acquired root extensions that effective differentiate them
>semantically regardless of the length or even quality of the root-initial
>vowel.
>
>As for "shortening somehow", I believe distinctions are maintained as long
>as they serve a semantic purpose but when they do not, a kind of entropy
>simplifies anything which can be simplified (*aa => *a, e.g.).

I still haven't seen any examples of the "distinctions
maintained".

>***
>
>
>> So what about final position? In standard laryngeal theory,
>> it is easy to distinguish between the 1sg. perfect ending
>> *-h2a (> -a) and the feminine Nsg. *-ah2 (> -a:). The fact
>> that in the 1sg. perfect an initial consonant was present is
>> proven by the failure of Brugmann's law in Skt. 1sg. cakara
>> (*kWe-kWor-h2a) vs. 3sg. caka:ra (*kWe-kWor-e). The fact
>> that in the suffix *-ah2- there was a final consonant is
>> proven by the laryngeal hardening processes we have been
>> discussing (*-ah2-s > *-ak-s).
>
>***
>
>In the first place, I operate under the assumption that all formants were
>originally words; and as words, must follow PIE root-form constraints. This
>means that in two-element formants, only *CV is permissible.
>
>Therefore, the feminine cannot be -*aH2 but must be *-H2a (-*Ha(:) in my
>notation).
>
>Just for completeness, I believe this derives from pre-PIE *ha.
>
>The collective, which has the same PIE form, derives from pre-PIE *hha.
>
>I agree that *H(2) was the final consonant of the root when -*s was added.
>But I account for it differently.
>
>With a base root of *sAnA, the plain verbal form is *sÁnØ -> *sén-.
>
>With the addition of STATIVE *Ha (from pre-PIE ?a) to *sAna, we get
>*sAnAHa -> *sAnÁHa -> *s°néH so *H is in final position to be hardened
>by -*s. This nicely explains why the attested form is <senex> vs. <**senax>.

The attested form is senex from *senaks. There is no +-eks,
only *-aks (Greek -ax, Slavic -akU, Armenian -ac, -ak`).

>Apparently, *° in the root syllable, manifests itself as *e.

Well, it shouldn't, +snah2 would have been a perfectly
acceptable root (cf. *sm-ih2 "1 (f.)").
The root is in full grade here, as expected in a thematic
formation (*sen=a-h2-).

>With <sena:tus>, we need only simply assume that the root to *-Ha was added
>was *sÁnA -> *sénØ + Ha -> *s°nHá -> *sena:.

The formation is in no way different (*sen=a-h2=t-o-).

>Of course, I do not doubt that the perfect ending was consonant-initial:
>*Ha.
>
> As for the Law that Brugmann disavowed, we had, as you may recall, a long
>discussion of its merits on this or the Nostratic list. It is riddled with
>so many qualifications that only 'linguistic correctness' could inspire
>someone to call it a law, in my humble opinion.

Brugmann's law is solid. The few apparent counterexamples
can easily be explained.

>The only reason I mention this at all is because I believe the 3rd p. sing.
>perfect formant is *-He.
>
>***
>
>
>> Another important difference between laryngeal theory and
>> your "vocalic theory", is what happens when we have a
>> laryngeal at both sides of the vowel. Again, in laryngeal
>> theory there is no problem. For instance, the PIE
>> instrumental suffix is *-eh1 > -e: (in your "vocalic
>> theory", I suppose that would be *-e(:)H with "original"
>> *e).
>
>***
>
>No, as you now know, I would have to reconstruct it as -*He(:).
>
>***
>
> Now when added to the feminine marker *-eh2-/*-o-ih2-,
>> the instrumental becomes *-ah2-ah1 > -a: (e.g. Lith -à) and
>> *-o-ih2-ah2 > *-oj(j)a: (Slavic -ojoN, with secondary *-m
>> [Skt. -aya:]). We see that the colouring effect of *h2 is
>> stronger than the neutral effect of *h1. I don't think
>> there is a way to explain this using vowel qualities in lieu
>> of laryngeal qualities.
>
>
>***
>
>There may not be a way to explain it but I will make an attempt.
>
>H2a+H1e (your notation, my form), in my notation: *Ha(:)+He(:).
>
>If the stress-accent was on the feminine ending: *Há(:)&.
>
>If not *Ha(:)&.
>
>If one allow a loss of *& after a long vowel, we arrive at the desired -*a:.
>
>For Skt. -aya:, we need only suppose a final syllable stress-accent:
>*&He(:) -> -ia:.
>
>I will leave it to you to comment on whether Skt. -ia: could become -*aya:.

No it couldn't.

>As for Balto-Slavic, I am completely ignorant of the turn and twists of it.
>
>If you feel the Balto-Slavic data is incommensurable with my proposed
>reconstrcutions, you will have to lead me through it like a child.

The Slavic a:-stem instrumental -ojoN, like the Skt.
instrumental -aya: and the Armenian oblique -oj^, derive
from PIE *-o-jh2-eh1 (that is: thematic vowel *-o- (this is
the expected shape before a voiced segment) + feminine
suffix *-jh2- in zero grade + stressed instrumental suffix
-éh1) > *-ojh2áh1. In Skt. *-ojHa: regularly gives -aya:
(without Brugmann), in Armenian we have -oj^ with regular
loss of the final vowel and jH > jj > j^. In Slavic we would
have expected *-aja: > +-oja, but an *-N was added (as often
happens in vowel-final case-endings), giving *-aja:N >
*-ajo:N > *-ajaN > -ojoN. The Slavic form cannot be from
*-oje:(N), which would have given +-ojeN.

This shows that the sequence *h2eh1 gives (H)a: (as in Latin
*h2eh1sah2 > a:ra). Apparently, *h3 in turn tops *h2 in a
word like *h2eh3m(e)n [or possibly *h3eh2-mn-] > Lat. o:men.

If the laryngeal merely preserved the quality of the
original vowel, we would always have laryngeals of the same
colouring [the same subscript in terms of the laryngeal
theory] on both ends of the vowel. The fact that we do have
*h2eh1, *h3eh2/*h2eh3 etc. disproves that. The fact that we
_don't_ seem to have *h1eh1, *h2eh2 ot *h3eh3 tends to
confirm that *h1, *h2 and *h3 were separate phonemes, given
that consonants do not normally repeat themselves in PIE
roots (reduplications excepted).

=======================
Miguel Carrasquer Vidal
miguelc@...