|Title:||Towards an Optimal Theory of Reflexivization||Add Dissertation|
|Author:||Silke Fischer||Update Dissertation|
|Email:||click here to access email|
|Institution:||Universität Tübingen, Department of Romance Languages|
Arnim von Stechow
|Abstract:||Based on the observation that the standard approaches to binding (Chomsky (1981)/Reinhart & Reuland (1993) and subsequent work) suffer from empirical inadequacies and have difficulties to account for crosslinguistic variation, this thesis sets out to develop an alternative binding theory which captures these shortcomings. The languages that are discussed include English, Dutch, German, Italian, Icelandic, and Faroese.
Furthermore, since it can be shown that a local derivational framework (cf. Chomsky (2000, 2001)) should be adopted for both conceptual and empirical reasons, the thesis investigates the question of whether binding can be successfully integrated into a derivational theory and explores the theoretical consequences of such an enterprise. Hence, after considering a global variant, the theory that is put forward in the end is a derivational approach to binding.
The underlying idea of the derivational analysis is that binding is encoded as feature checking with the antecedent acting as probe which selects the element it binds as goal. In the numeration it is only specified that there will be a binding relation between the designated antecedent and an element x; the concrete realization of the bound element x is determined in the course of the derivation. In the beginning, x is equipped with a realization matrix, which contains its possible realizations (anaphoric or pronominal specifications). After the completion of each phrase, optimization takes place and might restrict x's realization matrix by deleting the most anaphoric form it contains (depending on the respective language and the domain that has been reached). The result is the following: the longer x remains unbound, the less probable it is that x will be realized as an anaphor, because the anaphoric forms are gradually
deleted from the realization matrix in the course of the derivation and x must eventually select one of the remaining forms. When x is finally bound (i.e. when checking takes place), the most anaphoric specification that is left in the optimal realization matrix determines the vocabulary item that is chosen for Late Insertion (cf. Halle & Marantz (1993) and subsequent work on Distributed Morphology).
Technically, the optimal realization matrix is determined by two different kinds of violable constraints (cf. Prince & Smolensky (2004)): On the one hand, there is a version of Principle A constraints which are sensitive to domains of different size and which are violated by the anaphoric forms if x is still free in the respective domains. On the other hand, there are faithfulness constraints that punish any reduction of the realization matrix and might therefore delay the deletion of realization possibilities. The two groups of constraints are ordered in two universal subhierarchies; different interactions between these two hierarchies account for crosslinguistic variation, while universal properties of binding (like the fact that anaphors are used if the binding relation is relatively local and pronouns if it is less local) are captured due to the relative restrictiveness of the system.