> There is another proposal that seems pretty obvious to me, yet I've
> seen no discussion of it (though I can't imagine I'm the first to see
> it). I would reinterpret */t d dh/ as */th t d/...
My first response was rather brief - apologies. Breakfast was calling.
There have been several attempts to reinterpret the traditional three way
stop to make it more typologically acceptable. The precise system you
suggest was put forward by Toby Giffen in 1988.
Szemerenyi has a discussion of this sort of thing, partly at section 4.7
(page 54) and partly at section 6.7.1 (pages 142-145). Bomhard discusses
it in "IE and the Nostratic Hypothesis" chapter 3 (see especially pages 44 -
47).
Both accept that the traditional three-way contrast is typologically
unacceptable, but both also cite (different) evidence that there are
attested languages which have precisely this contrast. Bomhard points out
that languages analysed with this contrast are usually phonetically
misunderstood, and the voiced aspirate series is really something else.
Both of them reject the idea that the *dh series could have been voiceless.
Bomhard rejects the idea of fricatives of any kind. Both point out that
Pedersen had suggested a system where traditional */t d dh/ came from an
earlier */d t th/.
The problem is the amount of musical chairs needed if we remove voicing from
traditional *d and *dh, and aspiration from *dh. A system */ph b bh/ has
even been suggested to solve the typological improbability.
The trouble is that any three-way system contains an irrelevance. If the
third series in the traditional */t d dh/ is marked by aspiration, voicing
is irrelevant - yet voicing has to be reconstructed almost everywhere. If
the first series in your *th t d/ is marked by aspiration, lack of voicing
is likewise irrelevant, yet we have to reconstruct unvoiced consonants
almost everywhere.
There is at the moment no satisfying solution. Yours might well be right,
or someone else's - but there are problems with them all, and so the
traditional one, which at least accounts for the instability of the system,
tends to hold the field.
Peter