98 S T R U C T U R EASN D t s t Y ( J N U

n l. See, for example,Fauconnier(1975), Horn (1989:230ff.), Krifka (1995), Landman

d. {[smoke'(j)v smoke'(m)]n - S ({[smoke'()v smoke'(m)]'[smoke'() (1998),Israel (1998), among many others.
smoke'(m)l))) [tautologoustransformations]
2. The relevant entailment holds whenever the set of men is nonempty. Or, equivalently,
e.t[smoke,()vsmoke,(m)]1-[smoke,()nsmoke'(m)]}[bydef.ofS] we will be assuming a presuppositional view of determiners like every, according to which
everyN is taken to be uninterpretable as there are no Ns in the domain of quantification. See,
(8) Nobodythat smokesanddrinks getsto be 60' for example,de Jong and Verkuyl (1985).
LF: [ [No [bodythat t smokesanddrinks]l [t getsto be sixty]l 3. On scales,besidesthe works cited in the text, see,for example, Gazdu (1979) and
a. Let us write60(x)for "x getsto be (at least)sixty'" Hirschberg(1985). Heim (classlectures,Fall 1999) individuatesin monotonicitya necessary
-'
b. lltt getsto be sixtyllls: {60(x)n 61(x)} tby KRI condition for being part of a scale. For arguments that monotonicity is not also a sufficient
n n drink'(x)]l)
c. ll No bodythatsmokesanddrinksll= No(l,x[person'(x) [smoke'(x) condition. see Landman (1998).
d. |lNobody t h a ts m o k e s a n d d ri n k s | | ,= { N o (xx[person' (x)n[smoke' (x)n 4. Two caveats are in order. First, the notion of entailment is to be understood as gen-
drink'(x)]])n-S(No(Lx[person'(x)n[smoke'(x)ndrink'(x)]AlT))}tbySA] eralized in the usual way to all types (that "end i11["-see Parteeand Rooth 1983).Second,
[smoke'(x) n entailment must be understood as being relativized to contextually sharedknowledge in the
e. ll No bodythat smokesanddrinkslls= tNo(l,x[person'(x)n
n [smoke'(x) n No(},x[person'(x)
drink'(x)]), sense of Stalnaker (1978). See Heim (1984) and von Fintel (1999) for relevantelaborations
drink'(x)]])n- S({No(},x[person'(x)
ALT and Ap] of Stalnaker's view.
n [smoke'(x)v drink'(x)])])] [by def' of 5. U. Sauerland (personal communication) suggeststhat one way around this problem
n [smoke'(x)n
f. ll Nobodythat smokesanddrinksllE = {No(}"x[person'(x)
might be analyzing disjunction p v q by using such epistemic possibility operatorsas [0p I
dr ink ' ( x ) ll )n -N O (l ,x [p e rs o n ' (x )n [s mo ke' (x)vdri nk' (x)])])][bydetS ] Oql. The consequencesof this move need careful consideration. For example, it is not clear
n n
g. ll Nobodythat smokesanddrinkslls: {}"PtNO(}'x[person'(x)[smoke'(x) that a scalar treatment of the exclusivenessimplicature associatedwith disjunction can be
drink.(x)]])(P)nSoME((},x[person'(x)n[smoke'(x)vdrink'(x)])(P)1}[tautolo- maintainedunder such a view (without ad hoc assumptions).
goustransformations] 6. The localist may be accusedof predicting too strong truth implicatures with univer-
=
h. ll t tNo lbodythatt smokesanddrinks](ltgetsto be sixty]l ll" sal quantifiers. (a) is predicted to implicate (b):
{rPNo(},x[person'(x)n[smoke,(x)ndrink'(x)]]XP)nSoME((rx[person'(x)n (a) Everyone wrote a paper or made a presentation.
- S(NO(]'x[person(x) n [smoke(x)n
[smoke'(x)v drink'(x)]XP)l(]'xt60(x)l)n (b) Everyone either wrote a paper or made a classroom presentationand not both.
drink(x)l)((l,xt60(x)lAlr))) [bv SA]
The localist would predict the implicature to be
i. |l (tNo lbodythatt smokesanddrinks]][t getsto be sixty]] l|" = tNo(},x[person'(x)
n drink'(x)]]X(}'xt60(x)])n SoME((},x[person'(x) n [smoke'(x)v (c) It is not the case that everyone did both.
n [smoke,(x)
drink'(x)ll)((l"lxt60(x)l)n-S({No(},xtperson'(x)n[smoke'(x)ndrink'(x)]) I think that the (possible) presenceof the strong implicature (b) is plausible. Supposewe make
n [smoke'(x)n drink'(x)])((]'x[60(x)])'
((),xt59(x) l), NO(l'x[person'(x) a bet on (the truth o0 (a). I bet that everyonewrote a paperor made a classroompresentation.
(NO(Ixtperson(x) n [smoke'(x)n drink'(x)])((Ixt61(x) l)] )] tdef' of ALT and Then we find out that half of the people did both (while the other half, one of the two). What
GenAp would happen?I think there would be discussionon who won the bet. If the embeddedstrong
j. l| t tNo lbodythatt smokesanddrinks]][t getsto be sixty]] l|" : {No(},x[person'(x) implicature was simply not there, there should be no ground whatsoever for arguing in such
n[smoke,(x)ndrink'(x)]]X(xxt60(x)])nSoME((},x[person'(x)n[smoke'(x)v a case.
drink'(x)]X(}"xt60(x)])n-No(Ix[person'(x)n[smoke'(x)ndrink'(x)])((xx[59 7. I an assuming that (i) subjects originate within the VP, (ii) or can coordinate VPs,
(x) l)) [def.of S] and (iii) QR can apply at the VP level. Accordingly, the subject Mary in (27b) is extracted
= NO(]'x[person'(x) acrossthe board out of the coordinated VP. These assumptionsare madejust for explicitness
k. ll t [No [bodythatt smokesanddrinks]l[t getsto be sixty]l ll"
n v sake.None is crucial to the conceptual point at hand.
n [smoke,(x)n drink'(x)]]X(},xt60(x)])n SoME((},x[person'(x)[smoke'(x)
8. One of the few authors who takes an explicit position in favor of such a view is
drink'(x)lX(rxt60(x)l)lnSoME(l.x[person'(x)n[smoke'(x)ndrink'(x)])((]'xt59 Landman( 1998). His argumentsare developedin the context of a treatmentof numeralswithin
(x) l) [tautologous smallestsetcondition]
transformations;
an event-basedsemantics.The present chapter is an attempt at pursuing his proposal for all
implicatures (while remaining neutral with respect to event semantics).
Notes 9. An alternativemethodology would involve considering the implicature separatelyfrom
1999 atthe Center for Cognitive the meaning. Thus, for example (28a), repeatedhere, would implicate (b):
The frst version of this work was presentedin the fall of
pontignano workshop (which concluded chomsky's visit to ltaly)' (a) It's false that Sue harassedsome students.
Sciencein Lyon and at the
of california at Irvine, ucLA' uni-
Subsequentversions were preserrt"cut NELS, University (b) It's false that Sue harassedsome though not all the students.
in Leipzig and at MassachusettsInsti-
versity of Tuebingen (Fatl 2000); at the DGfs meeting
audiences contributed significantly to give shape But the results would not change. If (a) implicates (b), then (28c) ought to be a possible
tute of rechnology (Spring 2001). All those
to thank I. Caponigro, C' Cecchetto' S' Crain' J' Gajewski' continuation.
to the original ideas.i .tro *outd like
o. Percus, and u. Sauerland' 10. For some speakers(more tolerant in the use of the indicative future for optatives)
A. Gualmini, T. Guasti, E. Guerzoni, L. Meroni, F. Panzeri,