In modal counterpoint for more than two parts, should every pair of voices be in "good" two-part counterpoint?

Question

When checking my own three- and four-part modal exercises, I regularly run into questions like the following: if the bass has the CF, the alto is in 2nd species and the soprano in 3rd species, how exactly should I evaluate the counterpoint between the alto and the soprano? Is this just a version of two-part 2nd species in halved note values, or should I allow myself more leeway? If the latter, how much leeway?

Of course there are going to be obvious exceptions. Such as that the interval of the perfect fourth is allowable between the upper voices in a chord of the sixth. But other than such modifications, which rules should I abide by when evaluating the counterpoint between the upper voices?

Answer 1

Yes. Otherwise one loses the feel of 3 independent parts.

Answer 2

Yes.
Elaborating on ttw's to-the-point answer: one of counterpoint's general aims is how to write voices that keep the listener's attention engaged over a sustained passage of time, by avoiding redundancy or "cheating," particularly the redundancy encoded in two-species rules. Those rules apply at all time scales, to all pairs of voices. It's tedious to check all those pairs, but after a while you just learn not to break those rules in the first place.

Eventually, organists can even improvise four-part textures on the fly, without breaking those rules. Remember that J.S. Bach, whom we regard as the master contrapuntalist, in his day was esteemed instead for his improvisational prowess. (Which might explain his compositional volume: from modern examination of sketchbooks we see that he didn't hit the backspace key nearly as much as, say, Beethoven.)