While accepting that distribution is a coherent notion, I argue that it is nevertheless irrelevant to the working of the syllogism. Instead, I propose: (i) that a term's being distributed or undistributed in a proposition is its capacity to be replaced in a truth-preserving substitution with a narrower or a wider term; (ii) that which capacity the term has is determined by whether it occurs as the predicate of a negative or of an affirmative statement of the proposition; and (iii) that it is only the term's occurrence as the predicate of a negative or an affirmative statement–rather than its distribution value–that is relevant to syllogistic entailment.
"The Irrelevance of Distribution for the Syllogism." Notre Dame J. Formal Logic 35 (3) 433 - 449, /Summer 1994. https://doi.org/10.1305/ndjfl/1040511349