Re: How does Armenian fit into the Indo-European family?

On Mar 18, 3:53 am, "John Atkinson" <johna...@xxxxxxxxxxx> wrote:
"Darkstar" <darkstar...@xxxxxxxx> wrote ...

John Atkinson wrote:
"Darkstar" <darkstar...@xxxxxxxx> wrote...
John Atkinson wrote:
"Darkstar" <darkstar...@xxxxxxxx> wrote...
John Atkinson wrote:
"Darkstar" <darkstar...@xxxxxxxx> wrote
[...]

. Do you have more examples [of *w > g]

gorc < *worgyo- (E work); gitem < weid- (E wit);

But */w/ is neatly reflected as /v/ in most IE languages.

It isn't in Germanic, Italic, Celtic, or Greek, for starters.

See corrections below.

Isn't it strange it's so different in Armenian?

Why? It's a pretty common shift. A recent example is the western
Romance languages which have borrowed words starting with /w/ from
Germanic a millenium or so ago, replacing the /w/ with /g/ -- Germ
*werra, warD- > Spanish guerra, guardia, etc.

I think going from a bilabial phoneme to a dorsal g/k is going a
little too far. Besides, originally it probably was <gUerra>/
<gUardia> as reflected in writing.

I agree. It was w > gw > g. (In Spanish, though not French, the "w"
is
still pronounced before a -- as in guardia. And quite likely the
same
thing happened in Armenian -- some people claim it did; it's
certainly
plausible, though, AFAIK, there is no actual evidence one way or
another, since gw, if it existed, became g well before Armenian was
first written down.

I would also assume the initial /
g-/ indicates it was borrowed from Celtiberian, not Germanic (?).

No. These words (which often have to do with military vocabulary)
are
widespread in Western Romance, and were probably diffused via Latin
before the main period of Germanic invasions, as a result of Roman
contact with German "confederates". -- but it has to have occurred
after
original Latin /w/ > /v/, of course!
I
can't recall any other evidence showing a direct transition w > g,
especially in the Anlaut.

The only possible explanation is to postulate an initial */H/

What do you mean by /H/? I've been using this symbol to denote a
PIE
laryngal, which, in Armenian as elsewhere, is lost initially,
after
changing the vowel. It seems you're using it for something else
(?)

Okay, you're right. I probably mean an aspirated */h/ or */x/,
although a laryngal-aspirated sound or some other back consonant is
also possible here.

The actual pronunciation of the three or four PIE "laryngals" is
unknown; in particular, it may not have involved the larynx. So
we're
not in conflict there -- only about whether /H/ actually occured in
the
particular words under consideration, and, if it did, whether it was
still there after proto-Armenian separated from proto-Greek, etc.

which was lost in all other branches
which had a different common ancestor. Consequently, */w/ was
lost
in
Proto-Armenian. And I don't think that Greek <organon> and
<oido>
fit
the theory of being related to their Armenian cognates or they
should
be <horganon>, <hoido>, etc.

Why? Initial /h-/ in Greek usually comes from /*s-/. Thus oida <
*wida
< PIE *weid- (cf Arm gitem) and orge:, power <*worg^e: (cf Arm
gorc).

I mean they should prbly be <kworganon>, <kwoido> or <korganon>,
<koido>, see below. Here, you're keeping to the point that g < w,
while I believe this is hardly possible in the Anlaut: g < *h, that
is.

Why is g < *h any more plausible than g < *gw < *w?

If PIE had a distinctly enunciated *gw or *gv, it would be found in
the other IE languages. The initial *h in *hw was light, so it easily
disappeared in most groups but got stronger in Celtic and Armenian.
The transition *gw < *w in the Anlaut is unlikely because it's hardly
attested in the world phonology,

I disagree that it's "hardly attested".



therefore it could have been *hw
right from the start.

[...]

I will summerize what I'm trying to say below.

The point of this thread is to see/show whether Armenian is
different
from the mainstream IE languages, or whether the latter share any
innovations.

So far, we have:
1) E. <work>, Gr. <orga->, L.<opera>, probably also Russ.
<ver***'>,
etc, but Armenian <gorts>
2) Faroese <vita>, Sl. <vedat'>, Sanskrit <vedas>, Gr. <eidos,
Feidos>
''I saw', 'shape', L. <videre> 'to see', except Welsh <gwybod>
'know',
<gweled> 'see' where /gw/ is preserved,

You say "preserved"; this is your _assumption_; others reckon *w > gw

but Armenian <giter> 'I know'.
3) G.<wetter> 'weather', Sl.. <veter>, Lith. <vejas>, L.<ventus>,
'wind', Hindi <vayu>, Alb. <ere>, except Welsh <gwynt>, but
Armenian
<holm, kami> 'wind'

Are <holm> and <kami> cognate with all the rest? The others all come
from <*we:-(nt)->, blow.

According to M&A, <holm> comes from *Henh1mos, breath, which also
gives
Latin animus, Greek anemos. Seems to me /nm/ > /lm/ is very
plausible
(cf Latin <anima> > Spanish <alma>, etc. I admit that this doesn't
explain where the initial <h> in Armenian <holm> comes from!

But, it's not unique in this. In Armenian, h's are not infrequently
added to vowel-initial words, leading to doublets like <ogi>, breath, vs
<hogi>, spirit.

For no particular reason? Sometimes, the initial /h/ is acquired
historically as in Armenian <hanal> 'fall' from Old Armenian <enkel>
probably under influence of the velar /k/.

Do you know whether there's a word <olm>?

No, but I know there's a word <hosel> from <hosil> 'to fly' (< to soar
in the wind?). And it's not <kami>, sorry, but <k'ami>, the same
mistake you've been telling me about. Now with the <k'> there, the
Armenian origin of the word becomes more obvious. Might be West
Armenian, actually.

I don't know <kami>. If it's IE, it would presumably have come from
*gwem-, come, which seems implausible semantically -- also none of
the
references I have derive <kami> from *gwem- -- M&A claims *gwem- has
no
cognate in Armenian. My guess is that <kami> comes from a non-IE
language.

Japanese "kamikaze" :-) Seriously, with Armenian it's very hard to
tell, because its phonology is very different from the rest.

But the sound change rules, PIE > Armenian, are pretty well defined, and
exceptions are rare, no more common than in other branches of IE. The
fact that the final results are "very different" is immaterial.

Okay, more proof below.

*hw or *wh could > k` (cf *sweso:r-, sister > *hweur- > <k`oyr>), but
not k.

*w does > k in the cluster *k^w -- *k^wo:n-t-, dog, > *swund- >
skownd> -- as mentioned before. But I can't see how this could apply
to <kami>.

<Kami> indeed might be from another stock.

In any case, as you said yourself with respect to mi/mek, there's no
way
Armenian could have <m> in a word cognate with all those other words
you
listed in (3) above. This applies to both <holm> and <kami>.

I disagree here. [...] a relatively natural, progressive
assimilation in the Auslaut */hwent/ --> */holt/ --> */holn/--> Old
Armenian <holm> --> Modern <hov>. The transition to /m/ is influenced
by the bilabial /o/ which is from the bilabial */w/. Hence, the
existence of the original */w/ is supported by the presence of */o/.
As a result, the process seem quite smooth and natural.

Doesn't "seem" natural to me. /we/ > /o/ is not unusual, but I don't
think it happens in Armenian -- I know of no examples.

A possible explanation is below.

What's more,
"the transition to /m/ is influenced
by the bilabial /o/" is obviously wrong, since /o/ isn't bilabial!

Well, /o/ is formed by rounding lips, isn't it?

Here we observe Armenian behaving differently from the rest by
possibly preserving the initial */h/ from /*hw/, which could have
been
an aspirated (or partly laryngal?) consonant, reflected as
/k-/,/g-/,
/ h-/ in Armenian

If so, there'd have to be a conditioning factor to determine which of
the three it ended up as. What is that factor?

There can be several different factors at work. Right from the top of
my head,
*/hwe/, /hwo/ -> */gi/, */go/ is regular, but /ho/ may be partly
onomatopoetic. Cf. the many languages with the initial /f/, /h/, /b/,
/
v/, /u/ in <blow>, <wind>, etc., like Japanese <fubuki> 'storm',
Korean <param> 'wind', Hindi <heva>, Persian <bad> 'wind', Lithuanian
<pusti> 'to blow', Greek <fuso>, Czech <fukati> etc. In fact, words
denoting 'wind' with an initial /g/ are not so easy to find. But again
the Japanese "kaze".

and as /g-/ in Celtic. Here I assume that /w/ cannot
transform into /g/ directly, at least not in the beginning of the
word.

Therefore,
*<hw-> or *<Hw-> -- > <w> in most IE languages
-- > <gw-> in early Celtic
-- > <kh->, <g-> or <k-> in Armenian

Although one might immediatly object that this is not a pure
innovation, just a loss or rise in strength of an aspirated /h/.
But
judging the amount of the mainstream IE languages, it would be
unlikely the loss was so regular in so many dialects, therefore we
might assume there was a common ancestor which did not include
Armenian.

The reason why it's important to set Armenian apart is the
following.

IMO, the initial /d/ in *<duwo> or /tr/ in *<treyes> have no direct
connection to <erku>-<erek>

Armenian for "three" is <erek`>, not <erek>. Here <k`> denotes the
aspirated velar stop, which replaced <s> as a plural suffix, as I
pointed out before. This apparently isn't a sound change, it's a
substitution. Reason: /s/ > /k`> doesn't seem to occur anywhere
except
in the plural suffix -- elsewhere, /s/ > /0/ (mostly), or remains
unchanged (especialy after n).

Okay. It's probably should be denoted as /x/ to avoid confusion (?).

What, then, would you do with <p`> and <t`>? Anyway, "x" in IPA (and
Russian!) denotes the voiceless velar fricative, not the voiceless
aspirated velar stop. So to transcribe <k`> as "x" would be much more
confusing!

The outcome of this paragraph is that /kh/ doesn't come from the /s/
of */treyes/. More reason to believe these words are not directly
cognate.

I've just pointed out that "k`" regularly substitutes for "*s" in
plurals. <otk`>, feet, cf Greek <podes>; (-mk`>, 1st person plural
ending, cf Latin <-mus>, etc etc etc.

Whether or not the ending -k` is cognate with *-s (and, as I said, it
quite likely isn't), has nothing to do with whether <ere-> is cognate
with */treye->

Armenian clearly distinguishes aspirated and unaspirated voiceless
stops, and has done so as far back as the written evidence goes.
Please
don't use the same symbol for different phonemes, it confuses
everyone.

Point taken.

, because the direct reconstruction is far
too complex phonologically. A simple one would be:

In Armenian:
*<THeTHhaus> -->*<eTHhau> --> *<erhu> --> <erku>
*<THerese> --> *<THereh> --> <erekh>

This time at least you show the aspiration!

Maybe you think this is pedantic, but please don't use <..> for
postulated _sounds_. <...> denoted orthography, a transliteration of
the _written_ language.

Right again. Was in a hurry.

For all those starred bits, you should use
either /.../ (if they're meant to be phonemes), or [...] if they're
supposed to be actual (postulated) sounds. (Or ".." if you want to
be
intentionally vague :-)

This is what's always confused me. The IPA has [...] for phonemes.

No it doesn't. IPA is a _phonetic_ alphabet, not a _phonemic_ alphabet.
It has nothing to do with phonemes. Though people wanting to discuss
the phonemes of particular languages do find it convenient to "borrow"
IPA symbols to denote them.

Why /.../ for phonemes, then? What do you mean by 'actual' sounds?

Phones. What comes out of your mouth, as measured instrumentally.

Ok, I see.

Aren't phonemes actual?

That's a controversial question. For those that claim they are (or
those, like me, who don't care much, but find the concept useful),
phonemes are what the hearer, a native speaker of the same language,
thinks he hears. One phoneme can corresponds to several different
"actual sounds" (allophones).

Is there a formally accepted standard?

For particular languages, yes, usually. But a phonemic analysis is
unique to a particular language (or even a particular dialect of a
language), so a "formally accepted standard" that would apply to all
languages is an absurd concept.

I don't want to get involved in yet another discussion of the phoneme
concept -- it's been done to death here before, and others here are much
better qualified than me to deal with it. I advise you to read any
basic linguistics text (or Wikipedia, whose article on the subject is
quite good, IMO).

Wiki can be excellent if used carefully.

Okay, back to the main point.
First off, more examples on the *hw-correspondences.

4) Old Arm <havgit>, but Welsh <wy>, pl.<wyau>, Breton <vi>, Irish
<ubh>, Latin <ovum>, Greek <augo, avgho>, Frisian <eike>, Alb <veja,
veze>, Slovak <vajce>, Latv <ola>, Pers <xaya>. Appearantly, from
either */hwavg-/ or */wavg-/.
Again, we have the same correlation (Arm. */h-/ vs. others */w-/, /
v-/.) But this time the Celts have no initial /gw-/.The /gw/ in */
gwavg/ did not survive, due to phonetic complexity, hence possibly */
wevg/ or */wuvg/ > /ubh, wy/
Also note the preservation of the ending /-it/ in Arm, which is not
found in the rest of IELs.

5) Old Armenian /hot/ 'flower', but Lusitan /kwet/, Polnish /kwiat/,
Lithuanian <kvietka>, Welsh <gwellt> 'grass', Frisian <gjers> 'grass',
Hindi <ghas>, Marathi <gevet>, Greek <khortari>

6) Old Arm /kanac'/ 'green', [/delin/ 'yellow'], but Welsh /gwyrd/,
Lat. /verd-/ 'green', /helvus/ 'yellow', Slav. /zelen-/ 'green', /
zholt-/ 'yellow', Lith /zha~lias/ 'green', Alban /gjelber/ 'green', /
verdh/ 'yellow', Snkt. /haris/ 'green', Avest /zairi/ 'yellow'.
"Green" and "yellow" have the same base in IELs, but possibly not in
Armenian.

7) Old Arm. /kin/, but Welsh /gwraig/, Lith /zmona/, Slovenian /
zhena/, Goth /qino/, E 'wife' Alb /gruja/, Persian /zan/, Icelandic /
kona/, Snsk /janis/

8) Arm /erb/ 'when,' /ur/ 'where', /inc'/ 'what' but the IELs with
their classical interrogative *kwo-, *kwe- base.
9) Arm /ord/ 'worm', but Welsh /gallt/, G /wurm/, L /vermis/, Hindi /
kira/, Greek /s-kuliki/, Lith /kirmis/, Sl. /cherv'/, Baluchi /kirm/,
Alb /krimb/
In these two examples the initial */h/ in Armenian must have been lost
early on.

The outcome is that IE *hw/gw behaves predictably in Arm., but changes
into a lot of other things in the mainstream IELs (except in Celtic,
where it's mostly *gw-). This indicates that the transitions in Arm.
ocurred very early, while the mainstream IELs had time to
progressively develop distinct vowels (*hwa, *hwe, *hwi, *hwo) and
strengthen the /hw/-part in different ways (*kv, *gw, *hv, *w, etc),
which finally gave rise to the differences in assimilation of *hw.
This might mean that the emergence of strong /w/ in *hw in the
mainstream IELs was a recent innovation that took place after Armenian
had already branched off. To disprove this conjecture, one could try
to find some examples of /b-/, /p/-, /v-/ that would result from the
transformation of *hw- in Armenian, and thus indicate the presence of
a bilabial phoneme at an early state. But there seem to be none in
Swadesh's 200-word list, which implies that /w/ of *hw never existed
in Arm.! Therefore, this /w/ sound could actually be an *innovation*
in the mainstream IELs...

Therefore, we have
*/hu-, ho-/ --> /k'V-/, /hV-/, /V-/ in Arm
*/hu-, ho-/ --> */hw-/ --> mostly as /gw/ in Celtic, but variously as /
v/, /u/, /o/, /g/, /h/, /k/, /g/, /kw/, /gw/, /hw/ in most IELs.

Or what do you think?


.