Davidson's article is very tough going, but I'm pretty sure the interpretation I offered in class is correct:  Just as some readers of Donnellan hold, against Kripke in “Speaker’s Reference and Semantic Reference,” that in addition to semantic referent (SR) and utterer's referent (UR), there is an actual referent (AR) that may vary independently (see again the handout "On the Question of the `Actual' Referent")--AR tracks with "what is said," or vice versa--Davidson is maintaining that  in addition to semantic meaning (a sentence's literal, compositional, "standard," conventional meaning) and utterer's or speaker-meaning, there is an actual meaning that may vary independently, and that actual meaning tracks with "what is said" or vice versa.
    The Kripkean position on malapropisms would be that, even in context, the sentence used has exactly its semantic meaning, while the speaker meant something different.  The hearer realizes that the speaker cannot have meant what s/he literally said, and works out what s/he meant instead; that is how and why communication succeeds.  End of story.  The idea of "actual meaning" is just a confusion as between sentence meaning and speaker-meaning, based on failure to see the difference between sentence- or propositional semantics and communication theory.
    When Davidson introduces his notion of "first meaning" on p. 474, it sounds as though he means semantic meaning: "what should be found by consulting a dictionary based on actual usage."  But then on p. 475, speakers' intentions come in: "A speaker necessarily intends first meaning to be grasped by his audience."  Major tension here, as is shown by precisely the case of malapropisms.  Mrs. Malaprop did not intend Captain Absolute to understand "a nice derangement of epitaphs" as meaning a nice derangement of epitaphs, which is what the dictionary would specify.  She intended him to understand it as meaning _a nice arrangement of epithets_.  Davidson continues, "...and it is grasped if communication succeeds"; that means that "first meaning" must go with the speaker's communicative intention rather than with semantic meaning.  So "first meaning" is not semantic meaning.  But it is not merely speaker-meaning either (p. 474); Davidson says that "the intended meaning seems to take over from the standard meaning," yet he wants to keep this from blurring the distinction between expression meaning and speaker-meaning; first meaning is a kind of expression meaning.
    The reasons I assimilate first meaning thus understood to "AR" as it figures in the Donnellan literature are (i) Davidson's own reference to Donnellan and his (Davidson's) verdict that "you can change the meaning [of words] provided you believe (and perhaps are justified in believing) that the interpreter has adequate clues for the new interpretation" (pp. 477-78); (ii) the similarity of the latter and all Davidson's other communicative-intention stuff to Adam's requirement on ARs (that the speaker’s intention would or could have been recognized by a reasonable and reasonably informed hearer); and (iii) that first meaning, like AR, goes with "what is said."
    Now, what about Davidson's paper considered as a critique of existing views, of his own Truth-Condition theory of meaning in particular?  On the present interpretation, Davidson does not engage any existing views I know of.  His Truth-Condition theory was a theory of semantic meaning, not of "first meaning" as I am now understanding the latter.  The principles (1), (2) and (3) are certainly widely held of semantic meaning.  Does anyone hold them of  "first meaning"?  A prior question is, had anyone besides Davidson ever discussed "first meaning" at all?  Not so far as I know.  Theories of meaning of the sort we've been surveying are theories of semantic meaning.  Grice and Schiffer were explicating speaker-meaning.  Perhaps someone had talked of "first" or "actual" meaning before Davidson wrote his article, but if so I missed it.
    That said:  Is what Davidson goes on to say about "first meaning," about prior theories and passing theories and how neither of them corresponds to "the language a person knows," true?  Assuming there is such a thing as first meaning, I'd say so; at least, I have no major quarrel with any of it.  I would add, in a somewhat contrary spirit, that first meaning is derived from semantic meaning; a speaker commits a malapropism only when she mistakes the word she uses for a different word whose semantic meaning is her word's first meaning; also, the hearer interprets the malapropism only by initially recognizing, in light of the word's own semantic meaning, that that meaning cannot have been either the word's first meaning or what the speaker meant by it.
    There remains the question of whether there is such a thing as "first meaning."  Despite my Adam-increased liking for AR, I am not so fond of "first meaning," and I still doubt that it exists.  I would like to hear a persuasive antiKripkean argument for the claim that the meaning communicated in a case of malapropism is a kind of expression meaning rather than merely speaker-meaning.  (In a recent unpublished paper on Davidson, Marga Reimer considers such an argument: based on Grice's technical notion of a "conversational implicature," but we have not studied that notion, and in any case Reimer provides a convincing objection to the argument.)