Raymond Brown wrote:

> Besides one still has to account for analogy leveling out 'amica' ~
> 'amiche', while not leveling out 'amico' ~ 'amici'.  The hypothesis (*not*
> mine) I hold to explains these forms without analogy working in one case
> but not the other.

Well, analogy, unlike phonetic sound laws, is much less commonly
a completely regular phenomenon.   "Kine" may have become "cows", but
that doesn't mean "men" also had to become "mans".  (Although I'll grant
it does seem funny on an intuitive level that words so similar as 'amico'
and 'amica' did not undergo the same analogical leveling.)

> But natlangs unfortunately, as I'm sure you know, do not neatly evolved
> according to certainly fixed laws so that if, e.g. you fed Vulgar Latin
> into a computer it would then churn out modern French, Spanish or Italian.
> Human beings are so darned unpredictable.

True, but the underlying assumption behind that falsehood is that the universe
we live in is a deterministic one, that a "law" of Nature has to be a neat and
beautiful as Newtonian physics.  But the opposite of the falsehood is itself
not true:  language is not a haphazard, completely irregular phenomenon, but
is guided by probabilistic principles more like Einsteinian physics.  I'd say the
problem then with predicting language change is not in being able to know
whether there is any direction of (probable) drift, but, at least with the
development of linguistic theory so far, being able to take into account the
entirety of the system, much like economists trying to predict economic trends
with a few employment figures and the general price index.  That much won't cut it,
unfortunately.  ;)

Tom Wier <[log in to unmask]>
AIM: Deuterotom ICQ: 4315704
"Cogito ergo sum, sed credo ergo ero."