Dorit makes an excellent point about Katz on analyticity--especially about his needing to tie his style of semantic analysis to reference (and thus truth) determination.  Though I think I might instead have said, his needing to tie it to truth- (and thus reference-)determination.  This should open a rich vein.
     I agree with nearly everything Dorit said about my remarks.  Of course psychological reality per se doesn't entail cognitive-/representationalism, and I shouldn't have implied it did.  (But I wasn't so much presupposing c.-/r.`ism myself as just working with M&L's sense of "concept."  As Dorit says, we have to be eternally vigilant about these undercurrents.)  There is a perfectly good psychological-realist interpretation of Katz, as realist and concrete as you please, that still has little or nothing to do with M&L-concepts.
     To clarify:  When I originally said that there is "nearly constant unclarity" about (a) whether Katz is talking about "actual cognitive processing,"..., I didn't mean specifically the "more active and constitutive role in ordinary thought about ordinary things" that would be needed to make them into M&L-concepts.  I meant only, actual language processing within the speech center.
    However, my complaint about Katz' n. c. u. did not rely on the assumption that a psychological theory of "concepts" would have to be a theory of M&L-concepts.  It's obvious that Katz is not talking about M&L-concepts.  No one would think he was.  But the nearly constant unclarity is still there.