From: "Marcus Smith"

> None of my references say, but the following looks promising.

> Many of the geminates are the result of deleting a high vowel (u and i),
> and then assimilating the first consonant.  For example, _kekka_ 'result'
> comes from _ketuka_ which was a Chinese-Japanese compound.  The same is
> true of _kokkoo_ 'diplomatic relations' from _kokukoo_ and _hatten_
> 'development' from _hatuten_.

This is the way I've always assumed it worked. Since the p, t, k, that ends
Chinese syllables are non-plosive and almost glottal stopped, this makes
sense to me.

> The word _kana_ 'symbol, character' (as
> found in katakana and hiragana) comes from _karina_ -- presumably it went
> to _kanna_ > _kana_.

I have trouble with this. Unless the characters ascribed to "ka-na" are
"ateji" (certain characters erroneously attributed to Japanese syllables)
(and I don't think that is the case here), then "ka" hails from Chinese
"jia3", and there is no final stop on that syllable. So ka-na should be
"ka-na" without all the intermediate steps.

(Later) Ah, I've just looked up "jia3" in the Japanese character dictionary
and it gives "kari". So the indigenous reading of the charater is "kari" and
the Sino reading is "ka". I could see that going either way: ka+na= kana, or
kari+na>kan+na>kana. Native Japanese elements tends to combine with one
another and Sino elements combine with their own kind, so your derivation
may well work. Still, the above examples like "kokkoo" are Sino-Sino combos,
so I think the process of what's going on in your first paragraph and that
of your later paragraph may be different.