Asymmetry in voiceless stop lenition

[Eduardo Ribeiro]

Dear colleagues,

A comparison between Karajá (Macro-Jê stock, Central Brazil) and
related languages (such as Proto-Jê) reveals that Karajá underwent a
pervasive diachronic process of voiceless-stop lenition; in CrV
clusters, voiceless stops were thoroughly eliminated; in other
positions, *p appears as /w/ and *t as /r/.

*k, however, still occurs as /k/, but with a twist: in male speech, it
may be eliminated altogether (kòhã 'armadillo' > òhã, etc.). Thus, it
seems likely that the genesis of male vs. female speech distinctions
in Karajá may be somehow related to such tendency towards stop
lenition.  Notice that if lenition treated all voiceless stops the
same way, *k would show up as a velar approximant in Karajá (a very
marked, unstable phoneme cross-linguistically, as far as I know).

I would appreciate any examples of similar asymmetries in the
diachronic and synchronic behavior of voiceless stops. Examples
illustrating differing reflexes of a sound change between male and
female speakers would also be greatly appreciated.

Thanks in advance,

Eduardo
http://wado.us