The head of an NP is taken to be that component which determines the catego
ries of the NP as a whole. First impression of an NP in Jarawara (Arawa fam
ily, Brazil) involving inalienable possession (e.g. o-mano 'my arm') is tha
t there is conflict of criteria concerning what is head. The gender of the
NP, for verbal suffix agreement, is determined by the possessor (here Isg p
refix o-), suggesting that this should be taken to be head. But the whole N
P counts as 3rd person, for verbal prefix agreement, suggesting that the po
ssessed noun (mano 'arm') should be taken as head. Furthermore, the NP coun
ts as inanimate. Detailed analysis shows that there is in fact no conflict.
All NPs in this language are 3rd person (1st and 2nd persons being confine
d to head marking within the predicate and functioning as possessors within
an NP, not as full NPs). And all NPs involving inalienable possession coun
t as inanimate. The only variable is gender, which is determined by the pos
sessor; plainly, this is the unequivocal head of the NP.