From: fournet.arnaud
Message: 50470
Date: 2007-11-01
----- Original Message -----From: Grzegorz JagodzinskiSent: Wednesday, October 24, 2007 2:45 AMSubject: Re: [tied] swallow vs. nighingaleAnd my notices about words mentioned in this thread.
0. General notice. Contrary to Neo-grammarian believers, I doubt in
existence absolute phonetic rules in general (I do never call them "laws"
for the reason explained elsewhere). Of course, there may exists rules
without exceptions - but this seems to be a very rare phenomenon. It is also
possible to formulate rules in such complex manner that the rule would
explain all known examples - but I doubt in practical use of such complex
rules, illustrated by single examples.===============
A.F
What distinguishes Alchemy from Chemistry is the absence of rules.
You need "rules" to make sure traceability exists. And the result of traceability is that you can prove something because you can also disprove it.
No rules means empty speculation.
==============
Instead, I notice that phonetic rules function really well on words which
fulfil two conditions: a) they have medium frequency, i.e. they are not very
rare and not very frequent, b) they are derivatives with clear
word-formation structure, and the phonological changes (to go) would not
destroy this structure. They may also act quite well when a new phoneme
arises or when a phoneme disappears from the language system.=============
A.F
Look at French from LAtin, and you will know that phonetic changes do not care what the structure of the words is.
=====================
Other words which do not accomplish these conditions may also subdue the
phonetic rule in question, but they may also develop irregularly, for
various reasons, starting from the Zipf's law (tendency to shortening long
words with high frequency), through tendency to morphological simplicity of
the language, popularly called "analogy", and finishing on folk etymology
(adideation incomprehensible terms to well known words of today's language).
The subject here is birds names. Because they are words with unclear
word-formation structure as a rule, this is why there is so much
irregularities in their development. As for me, there is nothing strange in
it. And, as a consequence, I see virtually no reason to suppose that "a bird
language", a postulated source for many bird names in western Indo-European
languages, ever existed. Of course it might have existed - but we could not
prove it then. The observed irregularities are too weak evidence, and they
all may be due to development of inherited vocabulary, or due to borrowing
wandering words from various sources.
1. "Goose", Latin anser. This word comes from rural dictionary, i.e. it is
dialectal, and then without the expected h- < IE *g^h-. Polish ga,sior
"gander" (= he-goose) may have the same suffix -er- (even if thematicized,
i.e. in the original form *-ero-, in Slavic). Some satem languages give
evidence for *g^h- here, so Slavic go,si^ (i^ = soft yer) "goose" is a part
of the kentum vocabulary (some join it with Italian substrate in the history
of the Slavic branch, some suppose that g^(h) ... s > g(h) ...s - the
original palatal hardened when an "s" was in the word). Anyway, g- is -
generally speaking - unexpected here (or: needs a special explanation) , as
well as the lack of h- in Latin.=============
A.F
PIE *ghans is a compound word : gh_H2 + H2_ns.
So there is no "rural" or "irregular" situation.
LAtin has the simple word H2_ns. Most other languages have the compound.
============================
2. "Swallow". An interesing word. Latin hirundo and Greek khelidwo:n are
believed not to be cognates by ultra-neogrammarian s (God be with them). I
see no reason for denying that they are cognates. I see also no reason for
naming one of the forms "corrupted" (in fact, both are "corrupted" in such
or another way). Of course they did not develop regularly, and there is
nothing strange in it. These are just non-motivated words (with no clear
word-formation structure and obscure meaning of the root). I need no more
explanation beyond this statement. Just such isolated words do develop
irregularly, by their nature. Searching for reasons why we have -r- in Latin
vs. -l- in Greek, why -u- in Latin but -i- in Greek, why -nd- in Latin
and -dw- in Greek means just wasting of time. It can clearly be seen in
Romance where this lexeme continued to develop irregularily as well, and all
the forms are "corrupted".================
A.F :I agree that these words are at least connected.Two interesting loan-words exist in Uralic Moksha Mordvin :Wolf : *vrk- hence Moksha vrg-az (r is voiced and syllabic)Swallow : *KrH-t hence Moksha krxt-ja(ks) or krxt-agarx is something like : rsh starts like r but become unvoiced and spirantr and rx are two different phonemes.rx in "swallow" suggests that "something" was there.Initial status of K is lost : could be g or k or k? or whatever.These loan-words are supposed to be from Iranian languages.Comparison with *gheli:d- yields a skeleton : *gh_l/r_H-.Long i: from i+H-Looks like a PIE word for "swallow".Turcic languages have something like *qar-lak for swallow-bird.===============
Italian róndine lost e- for unknown reason (and this is why it is not as
clear as one might think), Old French aronde had irregular a- instead of e-
(not even speaking that it developed from nominative, not accusative, which
itself is irregular), and the same irregular are Port., Galic., Spanish or
Catalonian forms (with no need for the factor of the Basque substrate).
Btw., Portuguese andorinha may be affected by Basque or may be not affected.
But even if it was affected, the word became similar to a certain known
word, independently of its meaning.
Compare a similar instance in Polish where wielbl/a,d "camel" became similar
to "wielki" = "great" and "bl/a,d" = "error". This word has come from Greek
name of elephant (through Gothic), and **wielba,d would be expected in
Polish, not wielbl/a,d. It may be so that in Portuguese a kind of folk
etymology affected the word in question - but it is not a necessary
explanation. I meen we need not present it to explain the difference between
the expected and the observed forms. Which is more, nobody can prove that it
really was folk etymology here.
3. The proposed IE partial reconstruction *g^hel- (correctly based also on
Albanian), or even *g^helu-, suggests that the first part of the name of the
bird had connection with the adjective mean "green" (cf. Polish zielony <
*zelen- < *g^hel-en-), "gold-coloured" (cf. Polish zl/oty < *g^hel-to-) or
"yellow" (cf. Polish z.o'l/ty < IE kentumized *ghel-to-). The Germanic word
for swallow, *swalu-, might have been similarly "colourful", cf. Common
Slavic *solvu^ (u^ = hard yer) "yellow-grey" , attested in Russian solovoj,
and Germanic words without -w-, like English sallow, or OHG salo 'dark,
muddy'.
In Slavic *solviji^ was the name for nightingale, not swallow (cf. Polish
sl/owik, Czech slavík, Russian solovej); so, -w- in Germainic may be an
argument for Neo-grammarians that swallow has nothing to do with sallow -
but it is not the case in Slavic. It should be an argument for a wise
etymologist not to uncritically believe in saint phonetic "laws".
The Slavic form for "swallow" cannot be reconstructed even within one
language: Polish has "jasko'l/ka" while in Old Polish also "jastkul/ka" is
attested; Russian has "lastocka" and "lastka". So, even the hypothesis that
we have *g^hl-a:st- here, cognate to Latin and Greek words, cannot be
rejected definitely.
Btw., the most known example of -s-/-sw- is the IE etymon for "6". I also
call to mind that the popular and very widespread reconstruction of the type
of *s(w)eks (with -w- mobile???) cannot be taken seriously (it is enough to
compare Latin sex and Welsh chwech, or even Greek weks with digamma).
Etymologists who place it in their dictionaries and other works does not
seem to know all IE material well enough, or reject some evidence for
unknown reason. But the initial s^- in Slavic (szes'c' in Polish) cannot be
explained as either *s- or *sw-, and needs an additional k-. Avestan xs^vas^
attests this k- almost directly. The same about Greek kséstriks "six-rowed
barley" (for example, quoted in
http://www.sydgram. nsw.edu.au/ CollegeSt/ extension/ may99/arithmetic .pdf)
together with heks < weks 'six'. And finally, the same about Sanskrit s.as.
which "should not" have cerebralization in the anlaut sound - but it has
(clearly as a result of irregular development of the initial group *ksw-).
And finally, we can find indirect arguments for *ksw-, not *sw- and not *s-,
in Georgian ekvsi and in Finno-Ugric where *kuuti ~ *kutti "six" may be
cognate to the Indo-European form (of course, this is not an argument for
those who believe that the IE-ans are from Mars while the Uralic people from
Venus, or inversely; if somebody believes in common origin of the humans,
he/she should not just protest against Nostratic comparisons) .
These examples show that irregular development does exist, and that regular
and irregular forms may exist in one language side-by-side (see Greek). Of
course there is no strange in the fact that *-w- has preserved in some
languages while it (irregularily) disappeared in the other, since *k- has
behaved similarly. Phonetic rules just not always function, this is all. And
so, the presence of -w- in "swallow" and the absence of it in "sallow"
cannot be taken as a proof of lack of relation between the two words,
especially that a similar pair exists in Slavic.
I also notice that the Slavic word and the Germanic word which have the same
origin (in my opinion) are names of two different birds. I also see nothing
special in it.
And on some older stuff, mentioned also in this thread:
4. "Blackbird": if Germanic *a- in *amVsla- was from a laryngeal, it would
be rather strange since Germanic, as a rule, has not preserved initial
laryngeals as vowels. But if the word is Semitic, and *a- was the article,
which are its cognates in known Semitic languages?============
A.F
*a being the article is not the only possibility.
You have the Form IV in verbs that allow C_C_C to become ?_C_C_C.
This can be something like that in PIE waiting to be described properly
==================
5. "Lark". OE la:verce points at Gmc. *laiwVrk- which is also proved by
Finnish leivonen. The Latin < Celtic form alauda may, or may not, belong
here. If it is related, why we have a- not in Germanic here, like in the
previous example? And two questions remain: a) is the Finnish form isolated
within the Uralic family, and then we should suppose it was borrowed from
Germanic, or contrary, it is not isolated, and may have been the source for
Gmc., and b) if the word is Semitic, which are its cognetes?
Btw. The Polish word for "lark" is "skowronek". The Russian word is
"z^avoronok" . The first parts of the words are "corrupted", cf. also various
Ukrainian forms: z^ajvoronok, z^ovranok, dz^evoronok. All we can reconstruct
for Proto-Slavic is the second part *-vorn-. Taking into consideration the
variety of Slavic forms, I would not be very surprised if it appeared that
Germanic and Slavic forms are cognates (note -vorn- in Slavic and -wVrk- in
Germanic, absent in Finnish), to resentment of Neo-grammarians. ..
And on two non-bird-related terms from Schrijver:
6. "Ore". Germanic had evidently two independent words for "ore", from two
sources (one of them may have been inherited). One of them was *aruta-z,
*arutja-n, *arutja-taugo (OE o:ra, Swedish örtug, German Erz); note the
irregular development in Old English (lack of **-t). The second was OIcel.
rauði, and possibly the source of Finnish rauta - of course if the latter is
a borrowing. Note different stop in both forms, t : ð < IE *d : *dh. Because
of the fact, the first form does not seem to be inherited.
To the list of attested forms one should add Hebrew ?a:ra:d_ "bronze" (where
?a:- is NOT an article) and Sumerian urudu "copper". So, it may have been an
old Wanderwort, or a wandering word, borrowed into many languages in line,
from an unknown original source. It does not seem to have an article, and
the difference between a-full and a-less forms may be due different sources
rather than articles or laryngeals.
The presence of the prothetic e- in Greek erythros "red" is not a good piece
of evidence. Notice that Greek has no words with initial r- (rh- is from
consonantal groups, like *wr-, *sr-). A phonetic rule to add a prothetic
vowel to IE words with initial r- may have existed once in this language,
and the IE reconstruction *reudh- is also possible, instead of *h1reudh-.
Which is more, such a reconstruction is more probable because 3 consanants
(r, w, dh) are already in the root.
7. The reconstruction **teroP- for sulphur seems to me to be Martian, i.e.
taken from Mars (not good for the author himself...). Even taking apart
Greek astrape: which may not belong here at all (because of its meaning), I
see no reason for detaching this lexeme from Gmc. *swibla?-z, *swabla-z (OE
swefl, German Schwefel, Gothic swibl) < IE *sweplo-, Latin sulpur, sulphur <
IE *sulpo-r-, and also Slavic se^ra which lacks any "official" etymology
(formally, from IE *soira: or *se:ra:). OIr. sraib without -t- would be
original then, < *s(e)rapi? (maybe Celtologists have better ideas).
It is, however, worth noticing that Proto-Bulgar sarpur meant 'yellow chalk'
(modern Chuvash sara^ 'yellow', pura^ 'chalk'); this word may be so an old
Altaic borrowing into Indo-European; note also -ph- in Latin which is
characteristic for borrowed words rather than inherited ones. The Slavic
form may have lost the second part, in other languages a > u in the first
syllable (possible as a result of assimilation - but in real we cannot say
why it was so because we do not know the source of the borrowing, and its
phonology), in Germanic the development was *surpur > *surpul > *supul >
*swapul (or similarly), in Latin *surpur > *sulp(h)ur.
On various irregular changes in names of other substances, see IE names for
metals. Gmc. silubra- "silver", Slavic *serbro (note r.. r without
dissimilation) , Prussian sirablan (with r ... l, inversely than in Gmc.) are
certainly cognates (except not certain dissimilations and vowel changes in
Slavic which are "irregular") . But Lithuanian sidabras is not regular any
more because of "corrupted" -d- (from the dissimilated -r-). Greek side:ros
means "iron" but obviously belongs here. It has also -d-, like Lithuanian
(but no traces of *-bh- in addition, except the "suspected" length of -e:-).
Finally Basque zilhar "silver" also seems to belong to this family.
But note that the Greek form has another meaning, "iron". In Latin, there is
a word ferrum for iron, evidently from *fersom < *bhersom. Note that this is
not much than a "corrupted" form of the previous word with methatesis:
*bhers- ~ *serbh-. Or rather the previous form is "corrupted" and adideated
to the name of sulphur. The Latin form has cognates in Semitic: Hebrew
barzel (a "quadrilitteralis" (four-consonantal) stem, so rather not
originally Semitic), Akkad. parzillu (with p- ~ b-; in Latin -il ~ -el was
lost). Gmc. *brasa- "brass" may be (to Neogrammarians' dispair) another
mutation of the same word.
Grzegorz Jagodzin'ski
____________ _________ _________ _________ _________ _________ _
All new Yahoo! Mail "The new Interface is stunning in its simplicity and ease of use." - PC Magazine
http://uk.docs. yahoo.com/ nowyoucan. html