Apr 30:
Aristotelean contraposition consists of obverting an initial proposition, converting this obversion by switching the places of the first and second propositions (keeping any negation), and then obverting the conversion.
Obversion is often explained in this way:
A-type proposition: "All S are P" becomes "No S are non-P".
E-type proposition: "No S are P" becomes "All S are non-P".
I-type proposition: "Some S are P" becomes "Some S are not non-P".
O-type proposition: "Some S are not P" becomes "Some S are non-P".
I see two different unary operators "not" and "non-". I am informed that they are not equivalent, as "not" defines a negative proposition type (E or O) while "non-" is considered to define a positive proposition type.
My question is whether there is a meaningful distinction between these two operators. If there is a meaningful distinction, I see no explanation or interpretation of how "not non-P" differs from "P", and the list of examples (that seems to be repeated in multiple sources) seems woefully incomplete. If there is not a meaningful distinction coming from Aristotle, is there a meaningful distinction for modern logics (particularly paraconsistent logics)?
May 4:
Having read and considered the one answer to date and various comments and spent a good deal of time reading SEP (Aristotle's Logic, and The Traditional Square of Opposition), I am ready to start responding and following up.
Based on the answer and comments, it seems that my Apr 30 comparison of "not non-P" to "P" was malformed - it is effectively comparing one-and-a-half components of a proposition with one component - and I should instead compare "non-P" to "P" or "is-not non-P" to "is P". So part of my initial misunderstanding is resolved, I believe.
I like the replacement of "All" with "Every" that Mauro ALLEGANZA has been systematically using. The change prevents the need for a difference in number agreement (is/are) in the propositions, which reduces the amount of interpretation of what is changing from one proposition to another; and this change effectively eliminates the suggestion of a whole having properties that its parts may not share. ("The whole is greater than the sum of its parts" is something Aristotle coined.)
I also like Mauro ALLEGANZA's replacement of the complicated construction that operationally pairs a quantifier with "is" and infixes the subject. Using the "belongs to" relation makes proposition construction and proposition transformations both simpler.
What I do not see in any of the responses here or in the common presentations of obversion and contraposition is any treatment of existential import (i.e., which quantifiers should evaluate as true/false when one entity or the other or both is empty). SEP makes a good argument that "every" necessarily has existential import (i.e., "every" is not satisfied when the subject is empty) and that the O-proposition must not have existential import.
The arguments that SEP makes about existential import depend on the "contradictory" and "contrary" relationships between proposition types and on what is logically necessary to make Aristotle's syllogistic examples correct.
None of these necessary clarifications of the meaning of the proposition types is apparent in the examples or definitions of obversion that are currently presented on Wikipedia, for example. So, the common presentations of this aspect of so-called Aristotelean logic are indeed woefully incomplete, just as I worried on Apr 30. (I say so-called because I understand from SEP that Aristotle did not endorse contraposition and wrote about a few forms of obversion, but mainly these transformations were added between the 12th and 14th centuries CE and were then taught as part of "Aristotelean logic".)
Regarding paraconsistent logics, if the logical system evaluates correctly when empty entities are included, then it is compatible with a Kleene logic. Kleene logics and Priest logics are operationally identical but have different interpretations: Kleene nulls are unknown or undetermined, while Priest nulls are contradictory or overdetermined. A Priest logic need not be explosive and would therefore be paraconsistent. Consequently, a logic that is valid in the presence of empty entities could be paraconsistent. However, I was really wondering if there might be a more substantial difference between "not" and "non-" that might have led to something more interesting; but, obviously, that wonder is forlorn.