the uncertainty is limited to a specified set of individuals. This type of focus is evoked by question types such as (55).
(55) Which one of them—David, Pat, or Joe—went to the party?In English, the cleft construction(56) seems specificallyto pick out this sort of focus. For instance, (56) seems a more
felicitous answer to (55) than to (52.).^220
(56) It was PAT who went to the party.Here the focus is an individual presumed to be already known to the hearer; it is only this individual's role in the
situationthatis new. I'llcall thisrestrictive focus. A thirdtypeis illustrated in (54a): it is used tocontrast twoentitiesalong
some dimension or to correct thehearer's presumptionabout something in the common ground. I'll call thiscontrastive
focus. A special type of contrastive focus ismetalinguistic focus, where the speaker is commenting on the choice of words
or pronunciation:
(57) a.You didn't hear me right: I said BILL, not JILL.
b. You say to-MAY-to, and I say to-MAH-to.These types of focus might be encoded by diacritic features on [Focus] in the information structure tier; various
interface rules (including lexical items such aswhichand constructions such as cleft) would be sensitive to these
features.
The communicative function of focus is to specify what information in the sentence is new to the hearer. Given that
the overall function of communication is to inform the receiver of something new, it stands to reason that every
sentence should have a focus.^221 The communicativefunction of the rest of the sentence, the presupposition, is to link
the new information to what the hearer already knows.
There exist sentences in which everything is new, that is, the entire sentence is focus.
PHRASAL SEMANTICS 411
(^220) As an answer to (52), (56) reads more into the question than the question literally asks—which commonly happens, of course.
(^221) A number of people have suggested to me that this statement is overblown. For example, the use of language in religious ritual is not to inform anyone (even God) of
anything. The same is true in ritualized greetings (Hey, pal, howya doin'?). So maybe we're only talking about the kinds of examples typically discussed when people study
information structure. It bears looking into. Even in such cases, exactly what the speaker intends as new may have to be worked out pragmatically by thehearer (i.e. by
enriched composition). A good example is Lambrecht'sYou LIED to me! Presumably the hearer knows this already; what is new is the informationthat this is now part of
the common ground.