Marc Richards' 2004 dissertation "Object Shift and Scrambling in North and West Germanic: A Case Study in Symmetrical Syntax" (University of Cambridge) can be obtained from the author (it will also be posted on LingBuzz soon).

Abstract:

This thesis examines the well-known phenomena of Object Shift and Scrambling in Germanic from the perspective of a strictly minimalist, purely symmetrical syntax, arguing that their properties derive straightforwardly from the fundamental ‘symmetry-breaking’ strategies that ensure the linearization of such a system.

Chapter Two offers a unified analysis of Object Shift and Scrambling as parametrically determined variants of a head-complement ordering parameter – a version of Kayne’s LCA operative at the syntax-PF interface. The verb-object order-preservation effect known as Holmberg’s Generalization is immediately implied. The obligatory nature of Object Shift with weak pronouns is then shown to provide direct evidence that the phase boundaries defined by Chomsky’s Phase Impenetrability Condition (PIC) delimit a phonological as well as syntactic unit.

Chapters Three and Four explore the further technical and theoretical ramifications of the proposed linearization parameter for the derivational system of Chomsky’s Minimalist Inquiries and Derivation by Phase. In Chapter Three, I argue that the coexistence of order-preserving and order-permuting movement types in a single grammar lends further support to Chomsky’s phases (as linearization domains), and indicates the presence of a defective v phase-head selecting passive and unaccusative VPs. I offer some modifications that remove the weak/strong-phase distinction and yield a unified, lexical-array-based reformulation of the PIC. These lead to the main proposal of the chapter: the specifier of this light-v head provides the only possible merge-site for (there-type) expletives in the Probe-Goal-Agree framework. This low merge-site for expletives solves a number of technical, conceptual and empirical problems faced by the standard (Merge-TP) approaches and, in particular, allows a superior analysis of Transitive Expletive Constructions to emerge.

Chapter Four investigates the role of Case in the proposed account of linear-preservation (‘shape’) effects. I argue that Case-features assume a central importance at the syntax-PF interface in regulating the timing of Transfer/Spell-Out, so that an active element is locally identified as nonfinal for PF/linearization purposes. The predicted interplay between movement, shape, and phasal Spell-Out accounts for all the empirical facts observed across the Germanic paradigm. Finally, to bolster the case for Case, a defence is mounted for the indispensability of Case-features in the computation of LF. On the basis of the vP-analysis of expletives proposed in Chapter Three and a strong form of the activeness hypothesis, I propose a novel, unified analysis of Person-Case and definiteness restrictions that derives and explains the commonalities in behaviour between expletives and (Icelandic) quirky case. Defective intervention (and the Match/Agree distinction) is eliminated, dissolving into a heterogeneous range of phenomena that reduce, variously, to PIC effects, Agree-(in)activeness, the timing of optional-EPP-driven movement, and (in the case of Match-driven Move and multiple Agree) a parametrized approach to φ-completeness.

A much simpler, neater system emerges, one in which nonlexical macroparameters (such as the proposed head-directionality parameter) find a natural home as interface desymmetrization strategies that dispose of superfluous (symmetric) syntactic information.