Peter LeFanu Lumsdaine <p.l.lumsdaine@gmail.com> added:
Both constructions have plenty of examples; the OED online’s wild-card search is useful here, e.g.
http://www.oed.com/search?searchType=dictionary&q=*orical&_searchBtn=Search.
This gives 81 words with -orical, against 279 with -orial. The -orial examples are mostly from verb roots — dictatorial, professorial, etc.
Of course, these two don't stem from any "dictatory" or "professory", or?
— but with some exceptions: armorial, (im)memorial, and so on. I’m not enough of a linguist to see any full explanation for which words get which suffix.
The question is not which sort of ending occurs more frequently, -orical or -orial, but whether there's a "rule" by which a final "-y" on a noun gets converted to a final, adjectival, "-ial". I suggested earlier, by a long list of such nouns, that, if there is such a rule, it's more honored in the breach than in the observance. Again, if there is such a rule, why does it not apply to the nouns bigotry, burglary, comedy, empathy, felicity, poetry, progeny, prosody, registry, sodomy, sophistry, story, symphony, therapy, ..., zealotry? We have neither storial nor storical, for instance (despite "historical" (but not "historial") from history), nor ... (left to the reader) ... . Even for words like memory, remedy, or testimony, the results of the *+y => *+ial "rule" have meanings rather far from the adjectival "relating to the notion of *+y " meanings that most *+ical constructs derived from corresponding *+y nouns have. Memorial, for example, is a noun, signifying an object serving to recall a given memory, not an adjective signifying "relating to memory"; remedial, though an adjective, signifies, "serving to remediate", not "relating to the notion of a remedy"; and a testimonial is, again, a noun, not an adjective meaning "of, or related to, testimony". Perhaps only arterial, from artery, comes close, but even it means "flowing as through an artery", rather than "related to, or having to do with, arteries". And alluvial, jovial, sartorial, are words in the *+ial camp that do not, however, arise from any application of Peter Pease's proposed "rule", any more than do radial or medial -- or their ^c^ counterparts radical and medical: there is no noun alluvy, jovy, or sartory -- or rady or medy -- to apply such a rule to, any more than there is a cony to engender conical, or comy for comical.
But in the case of categories, the OED backs up what others have written: categorists are/were simply following standard usage. “Categorical” is older and more widely used, going back to 1598, and with plenty of both colloquial and technical usage. “Categorial” appears in 1912 in philosophy, and from the 50’s in linguistics, but remains mostly restricted to these fields. Google N-grams gives a quick view of the comparative frequency:
Comparative frequency of these endings is -- must I say again? -- irrelevant. (Ah: and Pease's "rule" doesn't work on "frequency", either? How comial :-) .) Cheers, and peace, -- Fred [For admin and other information see: http://www.mta.ca/~cat-dist/ ]