| Under a non-lexicalist view of word formation, such as Distributed Morphology (Halle and Marantz 1993), morphemes combine to form complex words during or after---but not before---narrow syntactic derivation. Such a model inevitably requires the availability of downward transformations, e.g. affix-hopping. This thesis provides a detailed investigation into such downward movements. Whereas previous analyses have relegated downward movements to a position outside of core derivational processes (e.g. Chomsky 1981 and, to a lesser extent, Embick and Noyer 2001), I argue that certain downward movements, namely head-to-head Lowering, form part of the central architecture of syntactic derivation and are motivated by fundamental properties of that architecture, such as phase impenetrability (Chomsky 2001).;Chapter 3 addresses tense-hopping, a canonical case of downward movement. I argue that certain asymmetries between English and Swedish provide evidence that these two languages derive their respective tense-hopping patterns via different means. Namely, Swedish tense-hopping is a case of Lowering, whereas English tense-hopping results from Local Dislocation (following Ochi 1999). Additionally, I propose a detailed theory of the Lowering vs. Raising distinction. Based in the observation that Lowering only ever takes place across a phase boundary, I posit a Phase Head Impenetrability Condition (PHIC), under which features embedded in a complex phase head become inaccessible as a result of Spell-out. Lowering occurs as a last resort feature-checking operation when the next highest head targets one of these embedded features; Raising occurs otherwise. I address several repercussions of this analysis, and in Chapter 4 I show that the PHIC allows for a straightforward account of the aux-raising vs. tense-hopping asymmetry in English. More precisely, I claim that auxiliary verbs are merged in the same phase as finite tense, and so the PHIC does not apply between these two elements, unlike with main verbs.;The analyses presented in this thesis all share a common goal: to show that downward movements can and should be incorporated into core linguistic theory.;Though this thesis addresses certain properties of other types of apparent downward movement (e.g. morpho-phonological merger; i.e. Local Dislocation), it focuses primarily on the defining characteristics of head-to-head Lowering. Central to this investigation is the observation that Lowering is a highly syntactic operation. In Chapter 2, I argue that a Lowering head may freely target any intermediate syntactic position of the complex head of its complement, thus deriving several cases of morphological optionality; e.g. reduplicative variability in Tagalog and Ndebele and the variable positions of agreement markers in Turkish. |
