Piotr asked:
>What is the real evidence for a laryngeal in the dual ending?
I wish I had it all at my fingertips, but I don't.
(a) length is suggestive (e.g. Greek -o: in the thematic declension; fem -a:
as opposed to short -a in the 3rd declension, and surviving as -a: [not e:]
in the first)
(b) The behaviour of the dual in Vedic. e.g. the long i: of duals is not
shortened in pronuncation before a vowel (see McDonnel's grammar p437,f n3).
Likewise the duals in -i:, -u: and -e do not cause the elision of a
following initial a-.
(c) The dual forms in -u are suggestive if compared with other situations
where we posit a laryngeal and find forms with -u, e.g. Sanskrit dadau (I
gave <*deh3) but cakara (no -u) for the words that do not end in a long
vowel < a laryngeal.
It would take too long to hunt out the other evidence, but I suspect it will
be easier to explain the forms by using a laryngeal than without one.
Peter