In our discussion of the categorial status of nominalizations in the previous chapter, we did not postulate a transformation to relate nominalized verbs to ordinary verbs, or nominalized clauses to ordinary clauses. In this chapter we will show that nominalized clauses, in which the head and its affixes determine the type of structure in which the head can occur, are best derived under the strong lexicalist hypothesis. The salient features of this hypothesis are:
Words are inserted into phrase structure positions. There is no formal difference between derivational and inflectional morphology. Word formation takes place in the lexicon. The lexicon is defined in terms of rule types that are different from those of the syntax.
The relation between elements or features of words and the structure that words appear in is not a transformational one. We will argue that features of words are related to syntactic structures through morphological control and percolation. These two mechanisms are formally distinct and are both needed.
In the ordinary case, the notion word as defined by the lexicon corresponds to the phonological word. We will argue that in Quechua, person, number, tense, and Case markers are true affixes, attached in the word formation component. Exceptions involve (phrase structure generated) clitics attached to the lexical word in the phonological component.