Optimality Theoretic Syntax: a radical approach

Joan Bresnan

Department of Linguistics
Stanford University
Stanford, CA 94305

(bresnan@stanford.edu)
http://www-lfg.stanford.edu/bresnan/

Prerequisites: Participants should have some familiarity with syntactic theory, such as a prior course in syntax. Some acquaintance with unification-based formalisms would be useful. OT syntax will be introduced in a reasonably self-contained way.

Summary: The classical generative theory of knowledge of language is that a speaker's mental grammar is a pure combinatorial engine, blind to typology and resistant to grammar-external forces. It arises from domain-specific innate principles of UG, whose parametric variation becomes fixed upon exposure to a given linguistic experience. In this classical epistemology, markedness hierarchies such as the animacy hierarchy cannot play a role in the individual synchronic grammar of a present-day English speaker. These hierarchies are not universal; they are exception-ridden, both across languages and even within the individual languages where their effects sometimes appear.

The emergence of Optimality Theory, particularly stochastic OT, has introduced new ways of thinking about these fundamental issues, in phonology and syntax alike. The OT architecture provides a very natural way of capturing the softness and emergent quality of universal markedness hierarchies, modelling substantive functional/typological theories of linguistic structure, and integrating variation and change into the general theory.

(Excerpted and paraphrased from Bresnan and Aissen (2002) "Optimality and Functionality: Conjectures and Refutations", Natural Language and Linguistic Theory, 81--95.)

In this lecture series I will first introduce (classical, vanilla) OT by showing how it explains 'emergence of the unmarked' effects, typological asymmetries, and morphosyntactic inventories. I will then show how OT augmented with stochastic evaluation (Boersma 1998) can capture generalizations across hard and soft syntactic constraints that elude classical approaches.


Back to course listing.