Fig. 11.2.Two ways of thinking about the CS-SpS connection
Letus returntotheissueofthehomogeneityof linguisticmeaning. Onemightfear thatthelinguisticintegrityof lexical
items is violated by allowing the meaning of a word to contain an SpS component: as sketched in Fig. 11.2a, SpS falls
outside the standard tripartite organization of language into phonology, syntax, and semantics. But consider the
evolutionary perspective. Suppose one deletes the phonological and syntactic structures. What is leftis a non-linguistic
association of cognitive structures in memory, much of which could be shared by a non-linguistic organism.
Phonological and syntactic structures can then be viewed as further structures tacked onto this concept to make it
linguistically expressible. Such a perspective is suggested in Fig. 11.2b.
With or without language, the mind has to have a way to unify multimodal representations and store them as units
(that is, to establish long-ter m me mory“binding”in the neuroscience sense). The structures that make this a“lexical
item”rather than just a “concept” simply represent an additional modality into which this concept extends: the
linguistic modality.
Admitting SpS as a component of lexical meaning provides immediate and welcome aid on the problem of
“completers.”For instance, Katz (1972) offers, as part of the semantic structure ofchair, features like [HAS A BACK]
and [HAS LEGS]. But whatkindoffeatures are these? Theyare notvery general: thekind ofback appropriateforchairs
will not do for people, camels, houses, or books; and something would not be a chair by virtue of having insect legs.
What's reallygoing onis thatwecallsome partofa chair thebackbecauseof(a)itsspatialconfigurationwithrespectto
other parts of the chair and (b) possibly its proximity to the back of the sitter. We call chair legslegsbecause they are
skinnyverticalsupports. WiththeprovisionofSpS, muchofthisdifficultyvanishes:SpScanencodewhata chair looks
like, including its part structure. The fact that certain parts of chairs are calledlegsandbackis encoded in the lexical
entries forthesewords, not necessarily in the entry forchair.
More generally, many perceptual properties can be encoded directly in SpS.Yellow, for instance, may be encoded in CS
as simplyKIND-OF COLOR, a reasonable