Looking at the tension between these theories fro mour present very concrete standpoint of storage in the brain, we
arrive at the conclusion that both approaches are correct, but for different classes of words. This not an optimally
elegant solution—but of course it is not clear that the brain is always optimally elegant.
Thesametensionappearsamongsyntactictheories. Itwasmentionedearlierthatmanytheoriessuchas HPSGachieve
a homogeneous treatment of morphology by doing it all“in the lexicon.”Such theories level out the distinction
between words, a grammatical notion, and lexical items, a notion having to do with storage; they treat all words as
lexical items. Thus in this respect they parallel lexeme-based morphological theories. The price is having to claim
(usually tacitly) that the entire paradig mof inflection for every word is present in the lexicon. Late Government-
Binding Theory (Baker 1988); Pollock 1989) and the Minimalist Program (Chomsky 1995) level out the distinction in
theoppositedirection:theytreatinflectionalaffixes as X°syntacticcategories, and hencelikewordsintheir potentialto
project complement and specifier structure. Because they stress the free combination of affixes, they parallel the
morpheme-based morphological theories. The price here is complex derivations that reduce elaborate syntactic
structure to the observed morphological forms, and an inadequate treatment of semiproductivity. In bothapproaches,
part of the proble mco mes fro mexpecting a ho mogeneous theory.^80
162 ARCHITECTURAL FOUNDATIONS
(^80) See Jackendoff (1997a : chs. 5 and 6 and and) for more discussion of morphology in the present framework, including some discussion of allomorphy and morphological
classes. The reader is advised, however, to take the treatment of morphological blocking there with a grain of salt. It can't possibly be that complicated.