bhrghowidhon@... wrote:
>
> Although */h1/ and */h2/ trigger pre-aspiration according to Olsen &
> Rasmussen's convincing theories, */T/ followed by */h1/ doesn't yield */Th/
> (or, at least, this is commonly held; I'm open to alternative views, but I
> don't have any at present)
Note that *h2 doesn't seem to have triggered aspiration in a _preceding_
stop in PIE, although it did cause it in some of the branches (most
notably Indo-Iranian). If one accepts the Olsen-Rasmussen preaspiration
theory (and I, for one, agree with your characterisation of it as
"convincing"), we need to distinguish two chronological layers of
aspiration by an adjacent laryngeal:
(1) PIE: t > tH/{h1, h2}__ (this *tH probably counted as _one_ segment
rather than two, since the reflexes of *Th1 are different). The
additional condition is that the laryngeal must be consonantal (i.e.,
preceded by a vowel or a syllabic resonant). I'm not quite convinced
that any other stop was affected in this way, but the process may have
paved the way for the phonemicisation of at least */tH/ already in PIE.
(2) Dialectal: T > TH/__h2, apparently affecting all previously
unaspirated stops of whatever place of articulation in Indo-Iranian
(though perhaps not in all positions), and perhaps a subset thereof in
Slavic and Armenian (and ...?). Note that, at least in Indic, *h2
remained reflected as separate _consonantal_ segment for some time after
it had caused aspiration; cf. its metrical effects in Vedic.
Interestingly, a word-medial syllabic *&2 also produced aspiration as
well as an extra consonantal position, since its pre-Indic reflex was
realised as something like *[hï].
Piotr