Ayesha Kidwai
Thesis Supervisor: Anvita Abbi
Jawaharlal Nehru University, New Delhi, 1995
This dissertation explores the syntactic and semantic/pragmatic properties
of Hindi and Urdu scrambling within the framework of Chomsky (1992). It
argues for a uniform analysis of Hindi and Urdu scrambling as adjunction to
XP, and formalizes the link between clause-internal leftward (CIL) scrambling
and preverbal 
WH-focusing in Hindi and Urdu. It therefore departs
from standard assumptions of scrambling as optional and/or semantically
vacuous.
Chapter 1 outlines the theoretical framework within which the phenomenon of CIL scrambling is investigated, and explores the problems that it poses for the minimalist program.
Chapter 2 investigates the syntactic properties of CIL from the perspective of the theory of movement and finds that the movement involved in scrambling cannot be characterized as either movement for Case/agreement, WH-movement, Topicalization or (S-Structure) QR. Following Muller and Sternefeld (1993), the discussion here suggests that, crosslinguistically, scrambling can only be characterized as adjunction to XP. The chapter concludes with a discussion of the status of XP-adjunction in the minimalist program, and proposes that the theoretically optimal characterization of adjunction would be one which provided the operation with a morphological motivation -- i.e. one in which XP-adjunction would, like substitution, be feature-driven movement.
Chapter 3 is designed to explain the facts in scrambled configurations, taken by Mahajan (1990) to be evidence for an A-movement analysis of the operation. Chapter 2 shows that the binding-theoretic evidence Mahajan presents is not only inconclusive with regards to an A-movement analysis, but also that if the binding facts in Hindi and Urdu default order are studied carefully, even Mahajan's data actually points in the direction of an adjunction analysis of CIL scrambling. Chapter 3 develops a Binding Theory that can explain the binding judgements obtained in scrambled configurations and still maintain CIL scrambling to be adjunction to XP. In this context, the structure of double object constructions in UG, the LF-raising approach to pronouns and reflexives, the theories of binding and coreference of Reinhart (1991) and Reinhart and Grodzinsky (1993), and the theory of reconstruction are examined and developed. The Binding Theory that emerges from these explorations is shown to have coverage beyond scrambling data, as it can provide an explanation for binding in double object constructions in a number of languages, as well as a principled distinction between Weak and Strong Crossover.
With this last argument against an adjunction analysis of CIL scrambling dispensed with, Chapter 4 turns to a consideration of the related issues of morphological motivations for XP-adjunction in general, and CIL scrambling in particular. The chapter claims that scrambling is employed for focusing elements in the preverbal focus position. The thesis describes this position as a Focus Phrase projection immediately dominating VP, which is activated by the scrambled XP. The mechanisms of head-activation are hypothesized to involve a version of dynamic agreement (Rizzi 1991), by which the scrambled XP transmits an N-feature to the head of the Focus Phrase. By this account, scrambling is no longer characterizable as optional (since it must take place whenever the [+FOCUS] feature is involved), or semantically vacuous (since scrambling has an expressed LF-effect -- focusing).
Chapter 5 considers the specificity effects noted in scrambled configurations. It suggests that with a few modifications to Diesing's (1992) Mapping Hypothesis, these facts can be explained by an adjunction analysis of scrambling. The chapter contradicts assumptions of a causal link between object agreement and specificity and makes some informal speculations regarding the sites of (in)definiteness in Hindi and Urdu.
Chapter 6 concludes the dissertation with a brief consideration of the issues of rightward and long-distance scrambling. It is claimed that rightward scrambling is actually base-generated, and that long-distance scrambling across tensed clause boundaries is ungrammatical.
Besides the issues surrounding scrambling, the dissertation offers some new
theoretical and empirical proposals. In Chapter 2, an analysis of the Hindi
and Urdu -to particle as a topic particle is offered; in Chapter 3
the impossibility of raising X
-reflexives across tensed boundaries is explained
in terms of Watanabe's (1993) layered Case theory. In the same chapter, a universal
proposal for the structure of ditransitives in UG is developed, that maintains
a strict version of UTAH and involves Case checking in a VP internal
AGR-oP. Dative shift is argued to require IO Case checking in a broadly L-related
position to this AGR-o. This is used to explain the binding and extraction facts in
double object constructions in a number of languages, including Albanian, English and
Italian.