Does anybody know a convincing etymology of German kein "no
..."?
The "official" etymology (cf. Kluge etc.) says that this form
comes from Old High German nihhein (nechein, etc.) by throwing out ni- (= IE *ne
"not") and "Verhärtung" ("hardening") of the -ch- in the initial
position. Middle High German knew also "kein" with the positive
meaning "irgendein" (any).
As for now, all is clear. The Proto-German (older than the 2nd
consonantal shift) was *ni-kein obviously. But what is its earlier
source?
Kluge links it with Goth. nih "and not" = Latin neque.
However, how the Germanic *h (= IE *k) could have yielded German (pre-second
shift) k (= IE *g)?
Another possible link, to Old Saxon nege^n and Dutch geen
"no... " is also suspected. Why -g- in OS and Dutch but -k- in German? Even if
we supposed that we have a rare example of g > k second consonantal shift in
anlaut, there is no way to explain the internal OS -g- (in nege^n) and OHG -ch-
(in nechein).
I have a personal hypothesis concerning these forms, and I
expect some comments on it. Perhaps no one has observed so far that an
intriguingly similar form is Polish "żaden" "no..." (= Germ. kein). There
are attestations of 14th-16th centuries in the form "niżadny" etc. (with
negative ni-), then this negative particle was thown away, and hence modern
"żaden". According to Boryś (Słownik etymologiczny języka polskiego, 2005), the
word under question is West Slavic only (with some East Slavic forms which could
be interdialectal borrowings). Its protoform could be *ni-z^e-edi^nu^ in
Proto-Slavic.
Now my own scenario. I assume that both forms, German and West
Slavic, are similarly built. The first element, Slavic *ni and Germanic *ne,
could be even the same word (the Slavic *ni < IE *ne ei, reduced to *ne in
German) or in German there is not the second IE part, the particle *ei, in this
word.
The second part is the IE particle *ge. It can be seen, among
others, in Germanic *mi-k (accusative "me"). It yielded *z^e in Slavic
regularly.
The third part is the numeral "1". It is *aina- in Germanic
(> ein in German) from IE *oino- but *ed(h)-oino- in
Slavic.
So, the common IE protoform could be *ne-(ei)-ge-oino- in both
German and West Slavic. It yielded *ne-k-aina- in Proto-Germanic, and
*ni-z^e-inu^, further *ni-z^e-ed-inu^ in Slavic.
What concerns inu^ / edinu^: It is noticeable that while the
original *oino- > in- is also preserved with the meaning "1" in SCS inorogu^
"unicorn", it has a new meaning "other" in other instances. When the new meaning
had developed, the old *ino-rogu^ was replaced by *edino-rogu^ (cf. modern
jednorożec "unicorn" in Polish). So, the assumed *ni-z^e-inu^ could
also have been replaced by *ni-z^e-ed-inu^ in the exacly the
same way as inorogu^ by *edinorogu^.
Germanic *nekaina- only survived in Old High German as *nekein
> nechein (2nd shift). In Slavic, the assumed *ni-z^e-ed-inu^ yielded
*niz^e^di^nu^ (with e+e > e^ "jat'"). The group z^ + e^ > z^a probably
regularly in Slavic (counterexamples may be all caused by analogy).
Originally there also existed a form without negation,
*ge-oino-. While there are no any traces of it in Slavic, it survived as kein
"irgendein" in MHG but disappeared soon. The initial negation was dropped in
both German and West Slavic, except some old Polish attestations. It is nothing
strange in it, but "ne" was similarly dropped in English and German development
(German "ich ne spreche" was replaced by "ich ne spreche nicht" and then by the
modern form "ich spreche nicht"). Omitting of negation in Slavic is rare but
just attested in this very word, so it is out of question.
And what with Dutch geen and Old Saxon nege^n? They do not fit
"kein" phonetically, so they are not directly related, even if constructed
similarly. Perhaps -g- comes from IE *ghe, another particle with similar
meaning. And perhaps they come from Pre-Verner Germanic *ne-h-aina-, with ne-h-
= Latin neque, like Kluge suggests.