General comment

This is an extremely small sample on which to base a reconstruction, however the fricative correspondence seems to indicate that the Mandan term is a cognate, not a borrowing from Hidatsa. The aspiration status of *p here is unknown and Hidatsa hC normally implies an old obstruent cluster or geminate s. It appears that locatives o- ‘in, locative’ and a- ‘on, locative’ may both have been possible with this term.