50 S T R U C T U R EASN D B E Y O N D
It may be useful to contrastthe sentencesin (33) with other non-DE determiners:
(34) a. Therewassomestudentwho had an incompleteor a failing grade
b. Somestudentwho missedclasswantedto taketheexamor contactthe advisor
Here there is no expectation that the studentsmentioned in (34a) may have both an
incomplete and a failing grade or that the student mentioned in (34b) wants to both
take the exam and contact the advisor. That is, within the restriction and scope of
some,the exclusivenessimplicature of or is (or typically can be) assumed.
Another interesting contrast concernsthe restriction versus the scope of the de-
terminerevery.As is well known, any is licensedin the restriction(which is DE) but
not in the scope(which is not DE) of every. And, in fact, SIs behaveaccordingly.
(35) Restrictionof every
got extra
a. .Everystudentwho wrotea squibor madea classroompresentation
credit.
b. #But not everystudentwho did both got extracredit.
c. Every studentwrote a squibor madea classroompresentation.
Sentence(35c) suggeststhat studentsdidn't do both, althoughin sentence(35a) the
implicature appearsto be suspended,as witnessedby the oddity of (35b). In assess-
ing their intuitions on casessuch as this, readersshould keep some caveatsin mind.
In particular, the defeasability of implicatures and the effect of shared contextual
knowledge have to be taken into due ionsideration. Consider,for example, sentence
(35c). If implicaturesare introducedlocally, sucha sentenceought to be understood
as 'oEverystudentwrote either a paper or made a presentationbut not both." I think
that this is what in fact happensin neutral contexts(i.e., contextsin which little is
known about the relevantfacts). Suppose,however,that we are consideringa class
where it is known beforehandthat besidesstudentsthat have done only one of the
two choresthere are also studentsthat did both of them. We would still describesuch
a situation by meansof (35c). The reasonfor this is fairly clear. We immediately see
that the implicature is incompatible with the context, so we throw it out. Yet using or
remains the best way to describethe relevant stateof affairs. The samegoes,mutatis
mutandis, for implicature removal. Sentenceslike (35a) appearto favor an inclusive
interpretation. But an exclusivenesscondition may be independently presentin the
conversational background. For example, we may be talking about a school where
you are actually penalized if you satisfy a course requirement in two ways. Or, to
give another example, consider a sentencelike:
' vii. It's enoughto know Italian or French(to be admittedto theprogram).
(36) It was a two coursemeal.But everyonewho skippedthe first or the secondcourse
enjoyedit more,for he wasn'ttoo full to appreciateit.
With a sentencelike this, we don't mean to include among the most satisfied cus-
tomers people who skipped both courses.We are told that the first and the second
coursecover,essentially,the whole meal. So underan inclusive construal,(36) winds
up saying that also those who skipped the whole meal enjoyed it. But this is a contra-
diction. fNotice, incidentally, that the existenceof caseslike (36) seemto constitute
further evidenceagainst globalism. We seemto be in the presencehere of an embed-
ded scalarimplicature.l
So what is being exactly claimed about SIs? The claim is that there are situa-
tions in which (standard)implicatures are by default presentand situationsin which
they are by default absent,and such situations are determined by structural factors.
By default interpretation, I simply mean the one that most people would give in cir-
cumstancesin which the context is unbiasedone way or the other. So the way in
which you want to testyour intuitions in assessingthe above(andthefollowing) claims
is by resortingto situationsthat are as much as possible "neutral." Sometimesthis
may be hard to do, but by and large I think it is possible.At any rate,this is certainly
an area of study where psycholinguistic experimentationcan usefully supplement
intuitions.
Let us move on to other relevant linguistic contexts and let us consider anteced-
entsversusconsequentsof conditionals. Any is licensed in the former but not in the
latter, and again SIs pattern accordingly.
(37) Antecedents of conditionals
a. If Paulor Bill come,Mary will be upset.
b. #But if PaulandBill bothcome,Mary won't be.
c. If Paulcomes,Mary or Suewill be upset.
In (37c)the unmarkedexpectationis that Mary and Suetypically won't both be upset.
There is a clear contrast with (37a), where the expectationis that if Paul and Bill
both come, Mary will be upset all the more.
The next class of casesconcernsclause-embeddingverbs that have a negative
coloring of some sort.
(38) Negativeembedding predicates
a. Dubitatives:doubt,deny
i. Johndoubtsthat Paulor Bill arein that room.
ii. # He doesn'tdoubtthatPaulandBill bothare.
b. Negativefactives:regret,be sorry
iii. Johnregretsthat Paulor Bill arein that roorh.
iv. #He doesn'tregretthat PaulandBill both are.
c. Negativepropositionalattitudes:fear,complain
v. Johnfearsthat Paulor Bill might not come.
vi. #Johndoesn'tfear that PaulandBill might both come.
d. Predicates of minirnumrequirement:be enough,suffice
viii. #It's not enoughto know both Italian and French.
This list is by no meansexhaustive.It is interestingto note that the verbs in (38) do
not pattern wholly uniformly with respectto any licensing. The verbs in (38a) do
licenseany. Those in (38b) typically do not. But they can be arguedto have some-
thing close enoughto the relevant property. For example, supposeI smoke and, in