primitive;thedistinctionbetweenitand alltheother colors thenis a matter of SpS, where colordistinctionsneedtobe
encoded anyway for perceptual purposes.
Moreinteresting aretheverbs. Forinstance,walk, jog, limp, strut, andshuffleare similarinbeingverbsofself-locomotion.
They differ only in manner of motion,one of those things that is extremely awkward to characterize through algebraic
features. However, we can distinguish these words by theirappearance (and how theyfeel in the body). Assuming SpS
cancodedynamicas wellas staticspatialconfigurations (Whoknowshow? Well,somelevelmust!),itwouldmake sense
to leave the distinctions of manner of motion among these verbs to the geometric/topological encoding of SpS, and
abandon the attempt tofind primitive algebraic features for this purpose.
As with all interacting components, interesting questions of balance of power arise between CS and SpS. How far can
content be bledout of CS intoSpS? For example,slitherandslideare somewhat more distant from theverbs above; are
they still differentiated fromwalkonly in SpS, or is there a CS difference as well? At present I don't know how to
decide. (For some other cases, see Jackendoff 1996d; van der Zee 2000; Bloom et al. 1996.)
Noticethatpushingsome semanticdistinctions intoSpSdoes notsimplifylexical semanticsas a whole.Alldistinctions
must stillbe encoded somehow (and I hope researchers on vision and action might be persuaded to collaborate in the
task). However, this move takes a great deal of the burden off the feature system of CS, where the problems with
“completers”arise. It also accounts for the fact that it is so hard to define words part of whose content is spatial: after
all, a picture is worth ten thousand words (and an image schema is worth a dozen pictures!).
This position further predicts that lexical items distinguished only by SpS structure will have identical grammatical
behavior, which for afirst pass seems to be correct. For instance, this explains the intuition behind Grimshaw's claim
thatwalk, jog, etc. are“synonymous”for linguistic purposes.
Wewillmake further use ofSpSinsubsequentsections. But it is time tomoveontothenext amplificationofsemantic
structure.
11.6 Other than necessar yand sufcient conditions
As may have already become apparent, the study of lexical meaning has given rise to considerable investigation into
categorization: how humans place individuals into categories, and how systems of categories are constructed mentally.
In traditional semanticterms, the problem can be framed in terms of statingthetruth-conditionsfor thesentenceThis
is an X, where X is the category in question.