basic vocabulary borrowing (was: IE "Urheimat" and evidence from Uralic linguistics)

[Ante Aikio]

[I wrote:]

Concerning the affix *-teksä(n) [in e.g. Finnish kahdeksan 'eight',
yhdeksän 'nine'], this explanation has been

>> recently revived: it has been argued that it is a loan from Proto-Iranian
>> *detsa. The phonetics are flawless; there are other examples of U *ks <
>> Iranian *ts (The cluster *ts was illegal in U, hence the substitution).
>> This explanation seems more plausible to me at least than the previous
>> rather fabricated theory that Finnish kahdeksan and yhdeksän developed
>> from *kakta e-k-sä-n 'two do not exist' (i.e., "two are missing from
>> ten") and *ükti e-k-sä-n.

[Adam Hyllested replied:]

> Well, what speaks  in favour of the latter theory is of course the fact
> that '10' in Finnish is not **teksa:n, but <kymmenta:/kymmenen>, a word
> that also exists in Mordvin, Yukaghir, and Omok. Furthermore, it
> resembles full grade of a root cognate to the IE *kmt- 'hand' (with the
> derivations *dekmt '10' and *(d)kmtom '100'); the zero grade shows up in U
> *ka:te 'hand', Finnish <ka:si>. Whether a loanword, a cognate, or a word
> of totally different origin, kymmenta: '10' must be older than '8' and
> '9', if these are borrowed from Iranian.

I don't think that Finnic-Mordvin *küm(m)ini '10', PU *käti 'hand and
arm' and PIE *kmt- have anything to do with each other. Loaning is
out of question, because PIE *mt would not give U *m(m) (in *küm(m)ini) or
*t (in *käti), since PU *-mt- was fully possible. Also, the words for '10'
and '100' have PIE palatal *k´, which regularly gives PU palatalized *s´
in loan words. Thus, in case of loaning one would rather expect U *s´VmtV
or something like that. Cognateship of course remains a possibility, but
this is pure speculation.

There is internal evidence suggesting suggesting that Fi.-Mordv. *kümmini
is secondary, whatever its origin might be. There is an apparently native
U word for '10' with a wider distribution: PU *luka '10', found in Saamic,
Mari and Manysi. This is obviously connected with PU *luki- 'count, say',
even though the morphology is unclear. The origin of *kümmini- is obscure,
but it resembles curiously Finnish kämmen 'palm of the hand'. However, the
sound correspondence (*-ü- ~ *-ä-) has no internal explanation.

And what are the Yukaghir and Omok cognates of *kümmini? This sounds
susceptible to me.

> By the way, on the basis of what material you are reconstructing a
> Proto-Iranian *detsa ?. PIE *dekmt became *das'a already in
> Indo-Iranian; compare Sanskrit <das'a> '10' and Avestan <dasa> '10'.

It's not my reconstruction really, but one of Asko Parpola and Jorma
Koivulehto. They maintain that PIE *e was retained in early
Proto-Iranian, which has something to with palatalization of velars in
Iranian, if I recall correctly - I can dig up the references for you, if
you are interested.

Regards,
Ante Aikio