Jay David Atlas (16/01/02): (...) My views are the same as in my J. of Semantics paper of a few years ago. Larry Horn has not yet replied to them, in print, though he might well have given a lecture or two somewhere on it. (...) Larry Horn (17/01/02): Actually, I *have* replied to Jay's views in a paper I've given (in different incarnations) at the June 1999 PRAGMA conference in Tel Aviv, the Jan. 2000 LSA in Chicago, the March 2001 Formal Pragmatics Workshop in Berlin, the Sept. 2001 LAGB meeting in Reading, and in seminar talks at Tokyo and Amherst, under titles like "Assertion vs. Entailment" and "Assertoric Inertia". (Most of the arguments in the above presentations were based on the behavior of the scalar adverbs "almost" and "barely", but "only" was dealt with as well.) I'll be giving it once more in the negation/polarity parasession at the April 2002 CLS and then (finally) writing it up for the proceedings. Unfortunately I won't be able to attend the "One Day 'only'" proceedings in Amsterdam in May because of family obligations, I didn't want anyone to think Jay's 1996 JoS paper left me stumped or tongue-tied. P.S. If anyone's curious, my position is that "Only a (Fa)" does indeed ENTAIL Fa, as in Atlas (1991, 1993, 1996), but in uttering the former, the speaker does not ASSERT the prejacent (i.e. the positive proposition), which is therefore "assertorically inert". Thus "only" with NP focus is indeed asymmetric (as argued in Horn 1969 et seq.), but this asymmetry is to be explicated in terms of what is asserted rather than what is entailed or presupposed. It is assertion and not entailment that determines NPI licensing and various related diagnostics of "negativity"; in particular, NPI licensing is determined by "downward assertion" rather than downward entailment per se. Jay David Atlas (18/01/02): Larry's postscript to his email about his current view on the relationship between assertion and Negative Polarity Items is obviously important: (...) I have not had the chance to think carefully enough about his arguments in a version of "Assertoric Inertia" that he sent me some time ago. I should like to do that for the Amsterdam workshop. But whatever the validity of the arguments, I believe that he has moved in the right theoretical direction. He alluded to Atlas 1991, which was my first and long-delayed attempt to re-open discussion of 'only Proper name'. That's in Journal of Semantics Vol 8, Nos 1-2, "Topic/Comment, presupposition, logical form, and focus stress implicatures: the case of focal particles 'only' and 'also' ". Section 3 of that paper is entitled "The Problem of Assertion and Logical Form: Topic and Focus" . I try to deal with the role of assertion. I reprise that discussion in Atlas (1993) "The Importance of Being 'Only'" in J. of Semantics, Vol. 10, pp. 301-18, which reacts to Larry's "The Said and the Unsaid". I claimed in 1991, p. 139, and repeated in 1993, p. 306 the following: From the point of view of a theory of speech-acts, in asserting Only a is F, we do not thereby assert a is F, the way we would if the statement were to consist of a conjunction a is F and P. Rather, what we do assert [i.e. the content of the assertion] entails a is F, but it [the former] does not "say" it [the latter]. This feature of my analysis preserves Aquinas's and Geach's intuition that an "excluder" like only excludes everything other than what is named by the subject-term 'a' from "sharing in the predicate 'F' " and need not go on to say [that] something named by the subject-term does "share in the predicate". The distinction between what is asserted and what is entailed was also crucial to my later view. In my 1996 J. of Semantics paper, p. 299, I wrote: Since I (Atlas and Levinson 1981, Atlas 1988, 1989, 1991, 1993) have emphasized the crucial logical role that the Topic-Comment structure of assertions plays in the logical analysis of cleft statements, negative existence statements, and focal particle statements, my analysis of the Topic NP role of F in Only a is F suggests that it is a feature of asserting the sentence 'Only a is F' with the focal article 'only' that there should be a generalized conversational implicatum (which includes, on my view, accomodation) There is an F. These remarks are by way of supporting the kind of account of Larry is currently giving, whatever quibbles I may have about the details. I hope he will send me copies of some of the recent lectures he mentioned in his email, so I can see whether our views are as close as I suspect they have become. Larry Horn (18/01/02): "I have not had the chance to think carefully enough about his arguments in a version of "Assertoric Inertia" that he sent me some time ago. I should like to do that for the Amsterdam workshop.  But whatever the validity of the arguments, I believe that he has moved in the right theoretical direction. (...) From the point of view of a theory of speech-acts, in asserting Only a is F, we do not thereby assert a is F, the way we would if the statement were to consist of a conjunction a is F and P.  Rather, what we do assert [i.e. the content of the assertion] entails a is F, but it [the former] does not "say" it [the latter] ..." It may be worth pointing out that I cite this passage of Jay's with approval in my 1996 JoS paper on "only" ("Exclusive Company") and in "Negative polarity and vertical inference", the article overlapping with it I wrote for the Ottawa conference volume _Negation and Polarity:  Syntax and Semantics_ (Benjamins, 1997).  In my more recent papers (still unwritten, much less published), I've argued for extending the asserted-but-not-entailed line to the contributions of "barely", "almost", and possibly existential import in definite descriptions and factivity for emotives. I think I have indeed been moving closer to Jay on "only" (i.e. "moved in the right theoretical direction" as Jay puts it :)   But I favor the assertoric inertia approach to his pseudo-anti-additive line with respect to "only NP" as an NPI licenser (and with respect to the various Karttunen/Peters diagnostics for "conventional implicature" that I reanalyze as diagnostics for non-assertion) and we still part company on "almost" (at least as of his 1997 JoS paper). Jay David Atlas (24/04/02): I've been studying the handouts from your Tokyo and Reading lectures and thinking about assertoric inertia. It would be a neat solution to the entailment problem. But the problem of assertoric force embedded in the conjunctive paraphrases bothers me -- always has. And I never liked Grice's brackets. I criticized them in "On Presupposing", MIND 1978. (See also Atlas, PHILOSOPHY WITHOUT AMBIGUITY, 1989) I just can't figure out how the negative component of the 'only' sentence gets "asserted" and determines the co-occurrence of the NPIs without the affirmative part getting asserted, unless you make a strong claim about "deep structure" of the sentence having semantically separable or conjunctive components in the mind and introduce a notion of "mental assertion" (see my comments about mental assertion in my 'only' papers). And the mental processing then does a lot of work to determine what is grammatically acceptable. Then a necessary condition on grammatical co-occurrence of the NPIs would be assertoric force of the negative part of the meaning. That just raises my original question about why there is a negative part, and how negative it has to be, and on your new view why it has to be ASSERTED, to explain the NPI's occurrence. My original argument was to suggest that the behavior of 'only Proper Name' showed that a negative meaning was not required, but there is so much theoretical investment in NPIs being justified by something semantically negative, that was a deeply unwelcome suggestion. Yet even my logical form for 'Only Larry ever Fs', paraphrased by: One person, and if anyone does Larry does, ever Fs. or One person, and at most Larry, ever Fs has got a negative meaning component somewhere, even if only in the antecedent of the conditional, or buried in 'at most'. But if you're going to require that a negative semantic part of 'only Larry ever Fs' is ASSERTED, in order to justify the NPI, why not avoid the problem of an embedded assertoric force, and just claim that NPIs are justified by the presence of any semantic component as negative as 'if anyone Fs, Larry Fs' or 'at most Larry' ? Then the hard theoretical work is to say how such a component interacts with the NPIs in order to justify their co-occurrence.