a“pseudo-cleft”construction, as in (52e) and (53e). The remaining replies show that this correspondence is not
random: it reflects a regular relation between the question and the answer.
There is a clear semantic/pragmatic intuition behind these results: the stressed part of the replies in (52) and (53) is
information^218 that K intends for J to add to their joint belief (in the sense of section 10.11), or to the“common
ground”in the sense of Stalnaker (1978). Thewh-phrases in J's questions indicate what information J wishes; the
inappropriate answers do not supply it. The standard ter mfor this newly supplied infor mation is thefocusof the
sentence (alternativelyrheme). The remaining part of the sentence can be called thepresupposition;we will refine this
shortly.
Turning for a moment to thequestionsin (52) and (53): we would like to say that the question-answer pairs share a
common presupposition, and that the stressed phrase in the answer is a focus. By analogy, then, it makes sense to say
that thewh-word in a question is also a focus, but with a twist: instead of supplying new information, it supplies a
variable to befilled with new information.
What sort of a grammatical entity is the focus? Many writers (including Jackendoff 1972) have identified it in terms of
a feature [+F] or the like marked in syntax. But this misses the point. First of all, one still must account for the
semantics/pragmatics of focus. A syntactic feature can at best correspond to or express this semantic/pragmatic
property of the sentence, which properly belongs in conceptual structure. Second, in (52/53a), the focus is expressed
through prosody alone, with ordinary garden-variety syntax. Here a syntactic feature [+F] serves only to“pass
information”fro mse mantics to phonology. Third, the syntactic expression of focus is enor mously varied, as seen
already for English in (52) and (53). Many of the writers citedabove (see especiallyVan Valin and LaPolla 1997) stress
that the varietyof expression is far greater when one looks across the languages of the world, and many languages are
far moregrammaticallyobsessivethanEnglishinexpressingfocus. Hencea simplefeature(or node)reallydoesnotdo
much work in syntax—one needs in addition complex ways to work out the actual realization of this feature in each
language.
Fro mthe present perspective, we can see that a syntactic feature [+F] is si mply an artifact of syntactocentris m, the
assumptionthateverything inmeaning has tobederivedfrom something generated insyntax. The parallelarchitecture
permits us to separate the semantic/pragmatic property of being focussed from
PHRASAL SEMANTICS 409
(^218) Note, by the way, that we can relinquish the ban on the term“information”here, since we really are talking about a speaker informing a hearer, not just about structural
distinctions rattling around inside a brain.