This is a long-promised attempt to
make sense of "Scythian" ethnonyms, beginning with those whose Hellenised
version ends in <-tai>. They seem to contain *-ta:, the Northeast Iranian
collective suffix, eventually generalised as a plural marker in Sogdian,
Yaghnobi and Ossetic. It seems that *-ta:, whatever its etymology, was
originally a clitic particle attached to fully formed words -- the nom.sg. of
thematic nouns and the nom.pl. of at least some of the consonantal stems. Final
*-ah (< *-os) developed into a front vowel (apparently *-i), so we eventually
get the collectives *-ah + ta: > *-i-ta (usually rendered as Greek -etai) for
a-stems, and *-a: + ta: > *-a-ta (Greek -atai) for a:-stems. This regional
innovation can be dated to late Old Iranian times, but when the intervocalic
voiceless stops were voiced in some of the Northeastern dialects, the still
existent boundary between the word and the clitic blocked the voicing; as a
result, the *-t- of *-ta: has remained voiceless (as in Modern Ossetic). Thus,
for example, tribal names of the form <...-sagetai> reflect *-sagi#ta
< *-sakah#ta: '-Saka[pl.]'.
To sum up, names with (Greek)
<-a-tai> and <-e-tai> can be treated as extensions of originally
shorter ethnonyms. Their characteristic feature is the presence of *-ta:,
Hellenised like the plural of Greek stems in -te: < *-ta: (i.e. with
<-tai>), and always with a voiceless stop (rather than anything
rendered <-toi> or <-doi>). Attempts to etymologise such names in
Iranian terms should begin by stripping off the pluralising suffix. So much for
starters; I'll try to deal with individual items later on.
Piotr