I've thought of an interesting defense for Carston's "cancellable explicature" (CE) view against the two simple arguments that were offered against it in class.
    My simple anti-Carston argument was this:  (1) Explicatures are said.  [Ignore the conflicting issue about "higher-order explicatures," though Carston might flatly deny (1).]  (2) What is (actually) said cannot be cancelled.  Therefore, (3) explicatures cannot be cancelled.  Therefore, CE is false.
    Trevor's simple argument was:  (1) Explicatures are said.  [Ignore the conflicting issue....]  (2) What is said in an utterance does not depend upon events that do or don't happen later.  (3) According to CE, explicature depends on (later) noncancellation of the original utterance.  Therefore, CE is false.
    The defense makes use of the idea, floated by (I think) both Jon and Brent, that cancellation is not so easy to distinguish from disambiguation.
    As I reconstructed Carston's reasoning, the disputed explicatures are supposed to have been generated from the original utterances by "development," in particular by "enrichment"; the idea is that actual slots or blanks or template-holes in the uttered sentence's logical form are being filled in by context.  In (75a), "ready" has to mean _ready to [DO X]_, and even when that blank is filled with "leave," "leave" itself has to mean _leave [LOCATION Y] for [LOCATION Z]_.  One gets no full truth condition or complete propositional content until those holes are plugged.  Likewise, in (77a), "jumped" has to mean _jumped [PREPOSITIONAL PHRASE]_; one can't jump without jumping somewhere somehow.  In (6a), when the verb "play" is transitive," "plays" (obviously) has to mean _plays [DIRECT OBJECT]_, where the direct object is a musical instrument or a game or some other type of thing that can be played.
    Now, this is not so different from contextual reference-determination, especially of pronouns.  "That" in "That's a shame" has to be assigned a referent ("Do you mean it's a shame that he demonstrated against the war in the firat place, or it's a shame that he went to jail for it?").  The sentence's logical form has a slot in it: _X is a shame_.  "Diane hopes that Robyn will like her new boss" needs to be disambiguated as between Robyn's liking Diane's new boss and Robyn's liking her own new boss, so "her" is a slot in logical form.
    Now, notice that there is a quasi-cancellation phenomenon with pronouns:  "That's a shame--no, not that he went to jail, but that he demonstrated in the first place."  "Robyn hopes that Diane will buy her book--careful, I don't mean Robyn's book, I mean Diane's own book."  And the same for proper names:  "I got an e-mail from John Roberts this morning--not John Roberts our philosopher of science, I mean John Roberts our former graduate student."
    Of course these are not cancellations of implicatures.  In the case of proper names, they are disambiguations of the names themselves.  In the case of pronouns, it is harder to say what to call them, because pronouns are not considered ambiguous; but, without there being a single word for it, the quasi-cancellations are wardings off of the most obvious but unintended assignments of referents.
    Now, Carston may assimilate the "cancellations" of her explicatures to these quasi-cancellations.  Pragmatic "enrichment" is slot-filling, which is very like the contextual assigning of referents to names or to pronouns.  When I cancel the explicature of (77a) by saying "Not over the edge, mind you, just up and down in place," I'm telling the hearer not to fill the [PREPOSITIONAL PHRASE] blank with the obvious candidate, but to use "up and down in place" instead.  This is not precisely disambiguation, but neither are the quasi-cancellations we do with pronouns, and it is much more like disambiguation than like the cancellation of a conversational implicature.
    So we can say that although explicatures are quasi-cancellable, they are not cancellable in the same sense implicatures are, so the label "CE" is a bit inappropriate; but now it is no good objection to Carston's specific claims of explicature that the alleged explicature is cancellable, because she can reply that it is only quasi-cancellable (until you or I prove otherwise).
    How does this position blunt the two simple anti-CE arguments?   Against my argument:  Carston can grant (3) but deny that the falsity of CE follows.  Though explicatures cannot strictly be cancelled, they can be quasi-cancelled, and that is enough for her purposes.
    Against Trevor's argument:  Carston can say that (2) is, perhaps surprisingly, false.  What is said can after all depend on what does or doesn't happen later.  If someone asserts "Robyn hopes that Diane will buy her book," and does not cancel as above, the obvious candidate is filled in, and the speaker has said that Robyn hopes that Diane will buy her, Robyn's, book.
    (Someone may protest that the speaker's intentions play a bigger role than this last argument allows:  If the speaker sincerely meant Diane's book, then even if her/his utterance was misleading, it would be incorrect to report her as having said that Robyn hopes that Diane will buy Robyn's book.  Maybe.  But consider this Humpty-Dumpty-ish case:  You say something about George Bush.  I say, "He is an idiot."  The next day you protest that I shouldn't have been so unpatriotic as to call the President an idiot when there's a war on.  I respond, "I didn't.  When I said `he' I wasn't referring to Bush.  Mentally I was referring to a second cousin of mine named Bluto."  Do I get away with that, even if it's true that in my mind I was referring to Bluto?)

    OK, am I convinced?  I am not.  There's a remaining difference between Carston's explicatures and the pronoun examples.  It's that since pronouns themselves are singular terms, referents have to be assigned to them in order to "develop" a full truth condition.  That is not true of "jump" or "play."  So far as Carston has shown, their slots could be filled by bound variables, and the truth condition completed by existential quantification: thus, the sentence "He plays well" could mean just _John Murray plays some thing [or some instrument] well_, which is a full truth condition.  (Of course I would speaker-mean, and implicate, that it's the violin Murray plays well.)  Likewise, the sentence "He ran to the edge of the cliff and jumped" could mean only _Lionel ran to the edge of the cliff and did some jumping (in some way or other)_.  Neither of these counter-suggestions is at all embarrassed by the facts respectively that "play" requires a direct object and that one can't jump without jumping somewhere somehow.

    As always, you be the judge.