It took place after the monophthongisation of *oi > *e^/*i in (late Proto-)Slavic. Several Germanic words were borrowed early enough to undergo the second palatalisation. The Krivichian dialect, on the other hand, was not affected, which means that the change took place when Slavic was already beginning to disintegrate. It was no longer an active phonetic process in the historically attested Slavic languages; e.g. it did not affect new Greek loans in Old Church Slavic. I'd date it at the 5th century, approximately. <köbäk> would have had to be borrowed about the time the palatalisation was under way and palatalised *k' (> *c) already existed as a phoneme in Slavic; an earlier loan would have been handled differently, and we'd be at a loss to explain why the word appears in east Slavic only. I doubt, however, if we can get the right outcome anyway. At the time of the second palatalisation the phonetic value of Slavic *o was [a], with a fronted allophone (later merging with *e) after palatal *j. <köbäk> (or the like) would have been Slavicised as *[kjäbak-] > *cebok-, or perhaps *[kjübak-] > *cubok-. I can't see how we can get *cobak- as required for Sergei's explanation of Trubachev's etymology to work.
My idea is that <sobaka> was borrowed from a source that had something like *s&baka -- not into Proto-Slavic but into relatively late East Slavic after the rise of new vowel qualities ([a] > [o], [a:] > [a]). The weakness of this hypothesis is that I can't identify the intermediary between Iranian and Slavic.
Piotr
----- Original Message -----
From: "george knysh" <gknysh@...>
To: <cybalist@yahoogroups.com>
Sent: Monday, October 07, 2002 8:25 PM
Subject: Re: [tied] Let dogs have their day too
> *****GK: What would be the temporal parameters for
> this process [second palatalisation] (earliest possible--latest possible)?