For those four languages the references given are:
http://wals.info/refdb/record/Franchetto-2002,
http://wals.info/refdb/record/Franchetto-2004,
http://wals.info/refdb/record/Merlan-1982,
http://wals.info/refdb/record/Andersen-1988,
http://wals.info/refdb/record/Simeoni-1978,
http://wals.info/refdb/record/Rumsey-1982,
and http://wals.info/refdb/record/Coate-and-Oates-1970.
You will also see 13 languages with SVO for feature 81A but VS for feature 82A. Namely; Breton, Chontal Maya, Huastec, Huave (San Mateo del Mar), Iaai, Mapudungun, Mocoví, Muna, Nahuatl (Michoacán), Paumarí, Pilagá, Salinan, and Tepehua (Tlachichilco).
Can anyone find out why these apparent inconsistencies exist? Or why they're "only apparent" (if in fact they are only apparent)?
If you look at http://wals.info/feature/combined/81A/83A or http://wals.info/feature/combined/81A/8 ... format=map you will see one language (Yukulta) with SVO for feature 81A but OV for feature 83A. The reference given for both features is http://wals.info/refdb/record/Keen-1983.
Can anyone explain that? Might the reference already contain the explanation? Or, if some CBBer lives or lived or studies or studied near the part of Australia where Yukulta (aka Yugulda aka Ganggalida) is spoken, or where Sandra Keen or RMW Dixon or Barry J Blake work, might you already know the answer?
Edit: On 3/20/14 I posted comments on WALS.info about Kuikúro 82A, Breton 82A, and Yukulta 83A.