This presumes that linguistic semantics is made up from different kinds of units than contextualized meaning—they
are different levels of structure in the same sense that phonology and syntax are. In effect, this breaks the interface
between linguistic for mand meaning of Fig. 9.1 into two stages:“semanticsfirst and pragmatics afterwards.”Such is
the position, as I understand it, in Katz (1972), Chomsky (1975), and Sperber and Wilson (1986), for instance.^139
Alternatively, linguistic semantics could be a“stripped down”version of contextualized meaning, comprising only a
subset of its units and/or distinctions(perhaps parallelto the relation between thesound structures encoded by lexical
phonology and phonetics respectively). We might sketch this organization as Fig. 9.3.
This still keeps semantic interpretation separate from its integration with nonlinguistic elements of meaning
(“pragmatics”). The only difference is that we can see the latter process as an enrichment of linguistic meaning rather
than a mapping into an separate format of structure.
What might lie behind thedesire to separate semantics from conceptualization? One motivation, I suspect, is a lurking
fear that general-purpose knowledge and belief are a bottomless pit, and that in order to make the enterprise of
semantics manageable it must somehow be restricted. And thereforesomedistinction must be made so we can stop
before drowning in endless detail.
Myownattitudeisthattheseams ofthemindmustbedeterminedempirically, notonthebasisofour fears. A priori, it
wouldmake more sense for therenottobeanyspecial levelof linguisticsemantics: thiswouldjust requireonemore bit
of work for evolution to do beyond evolving phonology and syntax. In addition, there is a methodological issue: we
can't tell whether there is a sea mwithout establishing what lies on both sides of it. Thus investigating linguistic
semantics
Fig. 9.3.“Linguistic semantics”as a subset of contextualized meaning
SEMANTICS AS A MENTALISTIC ENTERPRISE 283
(^139) Chomsky (1975 :105) has a diagram almost identical to Fig. 9.2. He calls linguistic semantics“logical form,”and considers the left-hand interface (hisSR-1) a part of
sentencegrammar. He characterizes the right-hand interface(hisSR-2)as“other semantic rules.”Of course his arrows go only fro mleft to right, because his gra m mar is
syntactocentric.