En réponse à Mark J. Reed :

>Did <h> become silent in Late Latin before it splintered into the
>Romance languages, or did that development happen independently in
>them later?  I know that it's silent (apart from its use in digraphs)
>in modern French, Spanish, and Italian, and I think in Portuguese . . .

Actually, things are murkier than just "<h> became silent and that's it" 
:)) .  Indeed, <h> was already silent in Vulgar Latin even before the 
Empire. So the original Latin <h> was lost already before split. But sounds 
change, and /h/ reappeared in some Romance languages, to disappear again. 
In Spanish, it came from initial /f/ which turned into /h/ (except in front 
of /w/, which explains  Spanish <fuego> vs. <hablar> from Latin FOCUS and 
FABULARE, IIRC). This new /h/ disappeared again. In French it was 
reintroduced through Germanic loanwords, and disappeared again, but leaving 
a trace (the infamous "h aspiré" which isn't pronounced but prevents 
liaison to happen. We say <la hache> and not *<l'hache> because <hache> is 
a Germanic loanword in French. A very good way to learn whether a <h> is 
"aspirated" in French or not is to look at the etymology of the word :) ).

So at least in those two languages you have two kinds of written <h>s: an 
original Latin one which was already silent before Latin split, and a later 
one which was actually pronounced at least during some time in the language 
itself (and in French still has some influence, even if it's not pronounced 
as such. French is maybe the only language to have such a "negative 
phoneme", i.e. something which is unpronounced but does influence 
pronunciation :) ).

Christophe Grandsire.

You need a straight mind to invent a twisted conlang.