lateral flaps, l, and r

[Spike L Gildea]

In the Cariban family (northern South America), there is a segment which
I reconstruct to a lateral flap in Proto-Cariban.  In a couple of the nine
Cariban languages I have worked on, it is realized almost entirely as a
non-lateral flap (like the [r] in Am.English _butter_); in the others,
I found free variation of the lateral flap with both a non-lateral flap
[r] and a non-flapped lateral [l], but with a strong skewing in the
frequency of each allophone depending on the preceding vowel: mostly
non-lateral flap [r] (and never a straight lateral [l]) following high
vowels, almost exclusively the lateral flap following /e/, and mostly a
standard lateral [l] (and almost never the non-lateral flap) preceding
/a/ and /o/ (cf. the pattern reported by Lichtenstein in his posting).

As far as the potential for change, what is interesting is that (1) in the
allophones, either the lateral feature or the flapping can be lost,
leaving the other as the sole distinctive feature for the segment, and
(2) these allophones are each attested as having become the sole phoneme
in some dialects of modern languages -- /r/ in some dialects of Panare (and
maybe Yukpa -- both from my own field notes), and /l/ in some dialects of
Carib Proper (a.k.a. Kari'na, Galibi) spoken in French Guiana
(Renault-Lescure 1986 and Meira pc).

Spike