| description |
This thesis concerns itself with the core syntactic phenomenon traditionally thought of within Principles and Parameters approaches in terms of movement. The point of departure is the observation that in two important respects, the characterization of this phenomenon in the recent Minimalist model of grammar (Chomsky 1993, 1995) (in terms of the operation Move) seems to fall short of that in the earlier GB model (Chomsky 1981, 1986) (in terms of the rule Move-a): first, the notion that movement operations apply freely seems impossible to maintain - a theoretical inadequacy; second, there is no obvious way of dealing with "optionality" phenomena - an empirical inadequacy. This thesis argues, however, that these apparent serious inadequacies of the Minimalist framework are in fact principledly soluble, and crucially without reverting to a GB-type model.The thesis falls into two parts, corresponding to the theoretical and empirical problems noted above. The central proposal of Part I is the Copy Hypothesis (Chapter 2): "all copies in a chain are active in the computational system". The relevance of this proposal is that, due ultimately to very fundamental properties of the standard Minimalist model, it actually appears impossible to maintain the notion that movement operations apply to any element - contrary to the Copy Hypothesis. However, I show how general conditions on movement are in fact sufficient to properly regulate the activity of traces, and give detailed arguments against Chomsky's (1995) proposal that "trace is immobile". Further to this, I show that the Copy Hypothesis has empirical applications involving the behaviour of wh-objects and associates of there in English, as well as computational complexity implications (Chapters 3 and 4). The Copy Hypothesis of Chapter 2 goes on to play an important role in Part II, in which I takes up the topic of optionality. The economy principles assumed to constrain derivations (in particular Last Resort) seem to exclude the possibility of optionality within the computational system. Since there is a certain amount of data which do appear to involve such optionality, the Minimalist framework evidently faces a major empirical problem, again seeming to lose out earlier models in which optionality data could be characterized simply in terms of optional application of Move-a. In Chapters 5 and 6, I show that there is in fact scope for some syntactic optionality within the derivational economy system. A system is developed whereby economy conditions in conjunction with feature properties of lexical items can derive variation in the timing of movement relative to Spell-Out. In this way, I account for optionality data (plus associated non-optionality effects) from French (optionality of participle agreement), English and Swedish (optional partial associate-movement with non-Case/agreement-checking expletives there and det 'it'), Icelandic, German and Dutch (optional overt Object Shift).
|
