“propositional,”though not strictly linguistic, and SpS is sort of“imagistic,”though not strictly visual.^181
My workinghypothesis at the moment (Jackendoff 1996d) is that the grammatical aspects of language make reference
only to CS, not to SpS. Nothing in grammar depends on detailed shapes of objects.^182 On the other hand, SpS is
language's indirect connection to visual, haptic, and proprioceptive perception, and to the control of action; it is
through the SpS connection that we can talk about what we see. This leads to the diagra mof co mponents and
interfaces in Fig. 11.1, in which the“meaning”component of Fig. 9.1 is opened out into CS and SpS.
Going back to lexical semantics, a stored piece of SpS can take the place of what many approaches (e.g. Rosch and
Mervis 1975; Putnam1975)informallyterman“imageofa prototypicalinstanceofthecategory.”For example,catwill
have a CS that identifies it as an animal and a member of the felinefamily and that specifies its typically being a pet. It
will also have an SpS that encodes how cats are shaped and colored and how they move. Not all lexical items, of
course, have an SpS component. Those expressing abstract concepts such asfairnessandvalueand those expressing
logical concepts such asand, if, andnotcome to mind as prominent examples. These will have only CS components. I
would imagine that words denoting sounds, tastes, smells, and feelings have components relating to those sense
modalities instead of to SpS.
Fig. 11.1.Incorporating the CS-SpS split in meaning
348 SEMANTIC AND CONCEPTUAL FOUNDATIONS
(^181) One can interpret Kosslyn's (1980) experiments on the distinction between“propositional”and“imagistic”processing as confirming this division of labor.
(^182) However,somelanguageshaveclassifiersystemsthatdependoncoarseshapes(e.g. Lakoff's 1987 exampleofJapanesehonthatgoeswithlongskinnyobjects);thesemight
or might not be encoded in CS, an interesting and as yet unaddressed empirical issue.