--- In
cybalist@yahoogroups.com, "Patrick Ryan" <proto-language@...>
wrote:
> If you would give a couple of examples of what you specifically
> believe could only be effected by four 'laryngeals', I will attem to
> explain it with only two.
I said three, not four. But why should I read the examples that have
been filling the pages of all discussion on laryngeal theory since 1878
aloud to you? To deal a well-deserved blow of sublime embarrassment to
someone so unnervingly hammering at those he call "underread" and yet
himself in need of having the most fundamental and well-known examples of
everything explicitly served and explained to him before he can motivate
his outbursts? It is not important for me to expostulate with you over any
of this. When Saussure saw fit to expostulate with his contemporaries
about their primitive grasp of these matters, he presented just about the
whole case in one go, and now you want ammunition from me to enable you to
expostulate with Saussure and the following generations of researchers
that have confirmed and tightened his results? But yeah, why not? You may
do well to consult R.S.P.Beekes, The Development of the
Proto-Indo-European Laryngeals in Greek, The Hague - Paris 1969: Mouton.
One of the main objectives of that classical book was to prove that a
third laryngeal (viz., the o-colouring H3) is necessary, and that the
minimum number of laryngeals for PIE is three.
The classical set. roots show what you need to be shown for Greek beyond
the shadow of any doubt. In Greek, a full-grade sequence *erH1/*erH2/*erH3
yields /ere/, /era/, /ore/ (sic, by metathesis) before a following
consonant, depending on the identity of the laryngeal, and the
corresponding zero-grades are /re:/, /ra:/, /ro:/, again depending on the
nature of the laryngeal, and in that order. Some material will show this.
I beg the indulgence of the other readers of the list, but my system needs
to get this out.
The root *terH1- 'bore, drill' is seen in the instrument noun tére-tro-n
'drill, auger' and the verb teíro: 'I drill' (térH1-yo:) which forms PPP
tre:tós 'durchgebohrt' from *tr.H1-tó-s. This means that a zero-grade
sequence /rH1/ has here yielded Greek /re:/ between consonants. Another
example is Gk. hré:-to:r or hre:-té:r 'speaker' from the verb eíro: 'I
say', Hitt. weriezzi 'speaks', which is IE *werH1-ye/o-, agent noun
*wr.H1-té:r, PPP *wr.H1-tó-s in Gk. hre:tós; here too *rH1 is reflected by
Gk. /re:/ between consonants.
The root of keránnu:mi 'I mix', aor. ekérasa from IE root aorist *k^ér&2-
(secondarily transferred to s-aorist form as /kera-s-/ as very common in
Greek), forms the PPP á-kra:tos 'unmixed' from IE *k^r.H2-tó-s; there is
an exact match with the Sanskrit compound á:-s'i:rta- 'hinzugemischt'.
-The genitive kra:atós of the word for 'head' is identical with the
Sanskrit ablative adverb s'i:rs.a-tás 'from the head', obviously
continuing *k^r.H2sn.-tós, i.e. with a zero-grade sequence /rH2/ reflected
as /ra:/ between original consonants.
The verb bibró:sko: 'I devour' belongs to a root with e-vocalism, Lith.
gérti 'drink', Arm. eker 'ate'. The Lith. tone demands a schwa which the
obvious analysis *gWérH3- readily delivers; the same goes for the strange
fact that the Armenian verb has (otherwise) joined the middle voice, 1sg
keray, 3sg keraw, obviously on the basis of the very stem *gWer&3- which
became /kera-/ and so looked like a middle-voice stem. Since sk-presents
regularly have zero-grade, the Gk. verb reflects *gWi-gWr.H3-sk^o:, with a
clear reflection of a PIE phonemic zero-grade sequence /rH3/ as Gk. /ro:/
between consonants. - The verb which is Sanskrit str.n.á:ti, aor.
ástari:t, ptc. sti:rn.á- 'spread out', forms Gk. aor. estóresa, again from
a root-aor. stem *store from *stéro (or a prestage much like it), IE 3sg
*stér&3-t. The e-vocalism of the aorist is indirectly reflected in Latin
sterno: and more directly in Russian pro-sterét' (from PSl. *ste"r-ti by
*anybody's* rules). Here too a zero-grade phonemic sequence *rH3 has
yielded Greek /ro:/ between consonants.
If there is no laryngeal, the situation is different: dérkomai, aor.
édrakon; kradíe:; most often however the product /ra/ is metathesized to
/ar/ so as to match the full-grade order of vowel + sonant: kardía: (cf.
kêr), phtheíro: phthartós, etc. This shows that a zero-grade segment
consisting of plain /r/ unaccompanied by laryngeals yields /ra/ (or /ar/)
in Greek between consonants.
Now in these sets we have seen IE anteconsonantal érH turn up as Gk.
/ere/, /era/, /ore/ (by metathesis) depending on the identity of the
laryngeal following the /r/. We have also seen the corresponding
zero-grades surface as Greek /re:/, /ra:/, /ro:/, again depending on the
nature of the laryngeal involved, and in this order. Now, the non-Greek
reflexes of these segments are mostly without any discrimination of the
laryngeal, thus Latin ra: for all three, Germanic /ur/ for all three,
Lithuanian ìr for all three, and Sanskrit i:r/u:r (Iranian ar) for all
three. This has led some to postulate a pre-PIE merger of the laryngeal in
these sequences, positing then only *r.H (or even *r.:), but that is
incorrect. Two of the three "long sonants" are kept apart in Tocharian as
seen by Klaus T. Schmidt, and they are all three kept apart in Greek as
summarized in the preceding. This has been known with great certainty
since the comprehensive study of the Greek facts by Beekes, and of the
main problem of Proto-Indo-European Schwebeablaut dealt with by Raimo
Anttila, both from 1969. Both scholars also showed, immediately assisted
by Rix, that some of the analogical explanations for schwa reflected as
Greek /e/ and /o/ are morphologically impossible, and that there
consequently is no possibility of accounting for the Greek laryngeal
reflexes by a PIE inventory containing less than three laryngeals.
Jens