A recent EL&U question about the word inception led me to look into discussions of the (theoretical but not actual) verb inceive, which turned up a discussion in Harry Bochner, Simplicity in Generative Morphology (1992), of words that follow the pattern conception/conceive, deception/deceive, perception/perceive, and reception/receive. As Bochner observes, the gap at inception/inceive may hold some interest, but a more challenging issue for a person trying to identify predictable patterns of back formation involves the existence of interception/intercept where a person seeking regular and relatively uncomplicated patterns might have expected to find interception/interceive.
Bochner attempts to formulate relevant morphological rules for various word pattern families, but (as he says) "The less regular the rule is, the more expensive it is to learn words that exemplify it." (Unfortunately, his discussion is made harder to follow by the omission of several pages from the relevant section of the book that is available in Google Books.) Eventually, Bochner concludes that "while it remains true that we can predict the form of the -ion noun given the root, we cannot predict the form of the root given the -ion noun." He also argues that, absent a special-case grammatical rule to handle -ceive/cept words consistently, "we get the result that it is cheaper to learn the word intercept than interceive."
Analytically this argument may make sense, but I don't think that it satisfactorily explains why English adopted interception/intercept in the first place; perhaps it (the argument) doesn't attempt to do so. In any event, my question remains:
Why does English have interception/intercept instead of interception/interceive?