incommonis nottheir semanticsbut their abilitytooccur in noun positionsinrelationtoverbs and prepositions, their
abilityto govern number/gender agreement and take case endings (in languages that have such functions), their ability
to occur with certain kinds of quantificational expressions such asmany, much, andall, and so forth—all syntactic
properties.
To besure, althoughnotallnouns denoteconcreteobjects, allwordsforconcreteobjectsare nouns: thereis a one-way
but not a two-way implication. And the nouns of early child languagedoindeed all denote concrete objects. (What
woulda two-year-oldhaveto say about perfectionandfinesse?) The consensus (e.g. Pinker 1989) is thatchildren f-use
this one-way implication in their language acquisition. The words for concrete objects are learnable in part because
they are visible and people can point the mout. Having learned the words for concrete objects, children f-know that
these words are going to be nouns. In turn, this enables the mto f-look for the gra m matical properties of nouns, fro m
which they can start f-figuring outstorm, cost, and the like. Notice that this learning strategy requires a non-zero
Universal Grammar: UG must stipulate that Concrete Objects in semantics are encoded syntactically as Nouns.
This isonlybywayofmotivatinga distinctionbetweensyntacticand semanticcategories attheverysimplestlevel.The
disparity will accumulate as we proceed.
5.6 The tripartite theory and some variants
Beforefurtherexploringthelexiconand thesyntax–semanticsinterface, we can sum up themainpointofthischapter:
the grammar overall has the tripartite organization sketched in Fig. 5.4.
Fig. 5.4 The tripartite parallel architecture