John Vertical wrote:
> If you count non-stops, Italic languages have examples of velar > labial in
> the form of *x > f, and thruout Uralic, developments of the type *G, *N > w
> (> v) are abundant.

Oh, right! That happened even in English. There's also the interesting interchange between /f/ ~ /h/ ~ /x/ ~ 0 in Iberian languages, at least some instances of which derive from Arabic /X\/.

It feels to me like this is most likely after a vowel and before another consonant or pause, as in English, but that isn't generally true of the Iberian examples.

On Feb 28, 2011, at 4:16 PM, Alex Fink wrote:

> On Mon, 28 Feb 2011 13:09:49 -0500, John Vertical <[log in to unmask]>
> wrote:
>> On Sat, 26 Feb 2011 15:01:35 -0600, Eric Christopherson wrote:
>>> It seems there are a lot of labial:velar (and sometimes labial:palatal)
>> correspondendences in North American language families. Does anyone know:
>>> 1. Does a change labial <-> velar tend to go in only one direction, or does
>> it go both ways? If only one, which?
>> I'm aware of at least labial > velar being attested (for example, in
>> Northern Samic *pS *ps *pt *pts > kS ks kt kts, and I've seen the same
>> change alleged as unconditional for some American languages).
> Mm, and on the flip side there's the change velar > labial before coronal,
> as in Romanian (e.g. _lemn_ < Lat _li:gnum_ where _gn_ is [Nn], _coapsa(_ <
> _coxa_, _fapt_ < _factum_).   
> In general I think I remember reading in Blevins gave me the impression that
> this interchange before coronals was significantly more common than, well,
> before vowels.

That certainly seems to be my impression.

There's also the phenomenon of shift of palatalized labials to palatal or coronal POA, sometimes with concomitant spirantization of affrication, as happened in e.g. Romance, Bantu, and Greek.