If you have the following two statements:
$\forall x \: P(x)$
and
$\exists x \: P(x)$
then the first is the superaltern of the second, and the second is the subaltern of the first.
So it is really a relationship between two statements where the one is an eistential, and the other a universal, but that are otherwise the same. Thus, for example, even though something like $P$ implies $P\lor Q$ we wouldn't call $P$ the superaltern of $P\lor Q$ .... Indeed, there would be no 'its (one) superaltern' in suc a case. So agai this is really about two quantificational statements with the only difference the type of quantifier.
In the square, the superaltern will be at the top, and the subaltern at the bottom, which is where I think the terminology comes from.