CHAPTER 5 The Parallel Architecture

5.1 Introduction to Part II

The last three chapters have surveyed the overall program of generative grammar: situating language in the f-mind,
treating it as a formal combinatorial system, and confronting the problem of language acquisition. After more than
thirty years, as I have tried to show, this progra mis still co mpelling.

However, not all aspects of generative grammar that have survived all this time are so worthy. In particular, I wish to
take issue with a fundamental assumption embedded deep in the core of generative theory: that the free
combinatoriality of language is due to a single source, localized in syntactic structure. I have come to believe that this
“syntactocentric”architecture was an importantmistake—perhaps historicallyunavoidable, but a mistake nevertheless.
Part II, comprising Chapters 5–8, develops the alternative assumption that language has multiple parallel sources of
combinatoriality, each ofwhichcreatesitsown characteristictype ofstructure.Fig. 1.1 has already displayed thesort of
structure produced by such a parallel architecture: multiple structures partly linked by indices.

In practice, as we will see, theorists in phonology and semantics have ignored the assumption of syntactocentrism for
over twenty years; and some of the offshoots of mainline generative syntax also have come to incorporate multiple
generativecomponents. However, theconsequences of this break withtradition haveneverbeenseriously investigated
inthecontext of generativetheory as a whole. That is what I proposetodo here.Theoutcomeis a theory ofgrammar
whichistruetothefundamentalgoalsofgenerativelinguistics, and whichatthesametimeaffords a clearer integration
among the subfields of linguistics and between linguistics and related disciplines.

5.2 A short history of syntactocentris m

Let me make clearer what I mean by“syntactocentrism.”TheAspectsmodel