Você está na página 1de 1

N U M T N A / A N U t H E S Y N T A X / p R A c M A T t c sI N T E R F A C E

45

properties
2. Empirical of scalarimplicatures Uncontroversially, the "not all" implicature will normally
be presentin interpreting
(15a),which will be taken to convey "Someothough
not all sutdentsare waiting for
In this sectionI will first presentdatathat castsomedoubtson the traditional,strictly him'" But if implicaturesarecomputedglobally, ,urh implicature
is absentfrom (l5b).
modular view of how SIs come about. Such data will suggestthat implicatures are This seemsodd, for after all we were just repirting wtrat (l5a) gives
(or, at the very least, can be) introduced locally (i.e., in the scope domain of the us ground.sfor.
The sameholds for numerals:
scalarterm) and then projectedto (i. e., inherited by) larger embedding structures.
The main empirical generalizationsthat characterizeimplicature projection will (16) a. John:"My colleaguemakes$100an hour.,,
be discussed. b. Johnbelievesthathis colleague
makes$r00 an hour.

If the standardneo-Griceanview of numeralsis correct,


2.1. Are there embeddedimplicatures? the phrase,.$100an hour,,
in (16a),via the scalarimplicature,comesto have an "exactly''interpretation.
As mentioned in the introduction, the dominant view maintains that implicatures are But in
sentence(16b)' the same numerical phraseloses such
an interpretation.In fact, if
computed globally, that is, after the semanticsof the whole root sentencehas been implicaturesare global, there is no way for unmodified numerals
in embeddedclauses
computed.In this section I will present somepreliminary evidence that goes against to get an "exactly" interpretation. To put it in slightly different
terms,a sentencelike
this idea. If this proves to be true, we then need a way of thinking about implicature (l5b) or (16b) certainly can be understoodas if it
had an embeddedimplicature.For
computation differently from the standardneo-Gricean one. example,(16b) certainly can be understoodas imputing
to John the belief that his
Let us begin by pointing out that, according to the standardview, embedded colleaguemakesexactly $100 an hour. If this attribution
doesn't come about via a
implicaturesshould not exist. Consider,for example,a sentencelike (12): local implicature,then how does it come about?
A related set of problems comes from factive verbs, that
(I2) Johnbelievesthatsomestudents is, verbs that are taken
arewaitingfor him. to presupposethe truth of their complement. As we know, presuppositions
can nor-
mally be accommodated.Take a sentencesuchas "My bike
is outside.,,Ifyou didn,t
If implicatures are factored in at the embeddedlevel, this sentenceshould implicate already know that I have a bike, you can accommodate
such information without
this one: any problem. By the sametoken, supposesomeonetells
us,
(13) Johnbelievesthatnot everystudentis waitingfor him. (I7) Johnknowsthat somestudentsarewaitingfor
him.
If, on the other hand,implicaturesare computedat the root level, sentence(12)'s If we didn't know the relevantfacts(namely,that somestudents
arewaiting for John),
relevant alternativewould be (14a). And the implicature should be its negation, we would typically accommodatethem. I think that, in
(14b). fact, we will typically ac-
commodatealsothe implicaturegenerallyassociatedwith the
embeddedclause.That
is, we interpret (17) as
(14) a. Johnbelievesthateverystudentis waitingfor him.
b. It is not the casethatJohnbelievesthat everystudentis waiting for him. (18) Somethoughnot all studentsarewaitingfor John
andhe is awareof it.
Sentence(14b) is much weakerthan (13). The former merely saysthat itis compat- This interpretation (and the way in which we accommodate)
doesnot come for free
ible wrth John's beliefs that not every studentis waiting. But this doesn't mean he at all if implicatures are computed globally, whereas it
does if they are computed
excludessuch a possibility, as (13) does.So, if (12) implicates(l4a) [as opposedto locally. In the local approach to implicatures, (17) is interpreted
(13)1,the implicaturenormally associatedwith someis weakenedto the point of being as
virtually suspended. (19) Johnknowsthat somethoughnot all students
arewaiting for him.
What are our intuitions like about thesefacts? Perhapsintuitions in this domain
are not sharp enough to settle the issue. Let me add, then, some further relevant ob- So we have right there the fact to be accommodated.The global
mechanism,instead,
servations.SupposeJohn comes to us and utters sentence(15a). Generalconversa- would only authorizeone to conclude,on the basis of (17),
tional dynamics authorizesus to claim on the basis of John's utterancethat (15b)
holds: (20) It is not the casethat Johnknowsthat everystudent
is waitingfor him.
(15) a. John:"Somestudents arewaitingfor me." We obviouslyneedsomeextra assumptionto get from hereto (18),
and suchassump-
b. Johnbelievesthat somestudentsare waitingfor him. tions are not so straightforward to state.The usual assumptio;r
fi ;il speaker(the

Você também pode gostar