Tosu mupthelasttwosections: Theoverallarchitectureofgra m mar consistsofa collectionofgenerativeco mponents
G 1 ...,Gn that create/license structures S 1 ,... Sn, plus a set of interfaces Ijk that constrain the relation between
structures of type Sjand structures of type Sk. (In principle, there may also be interfaces that address three or more
structures.) Typically, an interface Ijkdoes not“see”all of either Sj;orSk; it attends only to certain aspects of them. For
instance, the syntax–morphophonology interface does not“see”most of syntactic embedding, nor does it “see”
syntacticcategory distinctions such as NP vs. VP. Itsees only (primarily)thelinearorder ofwordsinsyntax, plus those
syntactic features such as case that map into affixes. So we can think of the distinctions in Sjand Skto which Ijkis
sensitiveas its“windows”ontoSjand Sk. Thisoverallarchitecturewillbereplicatedas wemovetofurthercomponents
of the grammar.
5.5 Semantics as a generative system
Let us turn our attention next to semantics. Assuming that the function of language is the expression and
communication of thoughts, I will identify semantics as the organization of those thoughts that language can express.
This is not the only construal of the term“semantics,”but Chapters 9 and 10 will argue that this is the appropriate
construal for an f-mentalistictheory of language. Anticipatingthatdiscussion, letus say thatthethoughts expressed by
language are structured in terms of a cognitive organization calledconceptual structure(CS). Conceptual structure is not
part of language per se—it is part of thought. It is the locus for the understanding of linguistic utterances in context,
incorporating pragmatic considerations and “world knowledge”; it is the cognitive structure in terms of which
reasoning and planning take place. That is, the hypothesized level of conceptual structure is intended as a theoretical
counterpart of what common sense calls“meaning.”
The hypothesis of conceptual structure goes beyond the goals of many linguistic semanticists, who wish tofind a kind
of specifically linguistic semantic structure that is more limited in scope, excluding all pragmatic or contextual
considerations. Sections 9.6 and 9.7 are devoted to showing why this limited goal ought to be abandoned. (However,
just in case I a mwrong about this, I will also show how such a level too can be worked into the architecture.)
It has become clear from the many approaches to semantics in the literature that semantics is a combinatorial system
independent of, and far richer than, syntactic structure. Formal semantics (Chierchia and McConnell-Ginet 1990;
Lappin1996 and CognitiveGrammar (Langacker 1987; Lakoff 1987)differonjustaboutevery issue butthisone:they
are both theories of meaning as a rich