En réponse à Raymond Brown <[log in to unmask]>:

> The development of the -oir ending of the infinitive took place _within
> French_; it was not inherited from VL.  The earliest French form, in
> fact,
> is _faillir_ << VL *fallíre.  The change from _fallir_ to the modern
> _falloir_ is almost certainly on analogy of _chaut_ ~ _chaloir_; _vaut_
> ~
> _valloir_.  So _faut_ ~ _faillir_ >> _faut_ ~ _falloir_.

The problem is that _faillir_ is a regular -ir verb, like finir, which gives
present participle _faillissant_, and the present "il faillit" (identical to the
simple past, like all those -ir verbs). So we have to think of _faut_ ~
_faillir_ >> _faut_ ~ _falloir_, and then analogy would make the remaining
_faut_ ~ _faillir_ >> _faillit_ ~ _faillir_. Is it possible?

> Thus from the one verb, the French have created _two_, since _faillir_
> (>>
> Eng. _fail_) survives (tho I believe _il faut_ = "he fails" is now very
> rare);

Not only rare, but incorrect, the correct form is "il faillit".

> The one Italian & two French verbs are derived from VL *fallíre for the
> Classical _fallere_ (all short vowels, with stress on the initial
> syllable)
> [perfect: fefelli, supine: falsum] "to deceive, trick, cheat".  There
> was a
> confusing shifting around of verbs between the Classical Latin 2nd, 3rd
> &
> 4th conjugations; nor were the shifts the same everywhere in the
> proto-Romance world.

Very true. Now I have to think what *fallíre would bring in Narbonósc (and what
meaning I will give it, Narbonósc is specialized in non-obvious semantic shifts
:) ).

> Fascinating, eh?

Indeed, and very educating for me. Narbonósc needs that :)