Linking, Optionality, and Ambiguity in Marathi

Optimality theory (OT; Prince and Smolensky 1993) has in a short time proven influential in theoretical phonology, and to a lesser degree in theoretical syntax. One of the main attractions of the theory is its strong commitment to typology. The basic hypothesis is that there is a set of universal constraints and languages differ only in how they rank the constraints. Since constraints are violable,1 this will result in the same input having a different optimal expression in different languages. The fact that constraint reranking is the only difference between languages is a very important assumption in OT and is necessary for its explanatory scope (Smolensky, 1996b; Bresnan, 2000), its learnability proofs (Tesar and Smolensky, 1998), and most importantly for typology. Once we assume “richness of the base” — as this assumption is often referred to — a natural typology arises. If there are n constraints in a strict ordering then there are n! different ways of ranking them. Since languages differ only in these rankings, it follows then that we predict n! languages. This is called “factorial typology”. Factorial typology is an important, automatic consequence of optimality-theoretic grammars. It is this property that truly sets OT grammars apart from ones in other generative frameworks. However, expressing something easily in a framework often comes at the cost of complicating the formalization of certain other facts. Frameworks in which constraints are inviolable can deal with ambiguity and optionality fairly well. If there are multiple ways of satisfying constraints, then there is optionality (in production) or ambiguity (in comprehension). Similarly, these frameworks have no significant problems with ineffability. This latter term refers to the fact that certain inputs to a grammar have no expression. In frameworks with inviolable constraints, these inputs simply violate one or more constraints and are therefore rejected by the grammar. However, it is precisely the properties of ineffability, ambiguity, and optionality that standard Optimality Theory (Prince and Smolensky, 1993) has trouble expressing. In this paper, I take steps to accommodate ambiguity and optionality in OT syntax. In particular I examine optionality in linking arguments to grammatical functions in Marathi and give an OT-LFG account of the relevant facts.2 In doing so, I will illustrate four points. The first is that linking can be achieved with a small set of cross-linguisically plausible, violable constraints. Second, optionality can be captured in OT by modifying the architecture of the theory only slightly. Third, in comprehension directed optimization, the same OT constraints that are used to capture the linking optionality in production can also capture the resulting ambiguity in the Marathi strings that correspond to the This work was supported in part by SSHRC Doctoral Fellowship 752-98-0424 and was completed as a Research Fellow at UCSC, where I was kindly sponsored by Jim McCloskey. I am also very grateful to the following people for discussion, comments, and suggestions: Farrell Ackerman, Judith Aissen, Arto Anttila, David Beaver, Joan Bresnan, Miriam Butt, Ashwini Deo, Alan Gous, Frank Keller, Paul Kiparsky, Chris Manning, Line Hove Mikkelsen, John Moore, Peter Sells, the participants in the Stanford OT Working Group, and the joint Stanford/UCSC Optimality Theory Syntax class. I would like to particularly thank Ida Toivonen, who has influenced my thoughts on OT and optionality through many long discussions. I accept sole responsibility for any remaining errors. 1It would be more precise to say that a form can be grammatical and still violate constraints. 2OT-LFG is a version of OT syntax which uses Lexical Functional Grammar (Bresnan, 1982; Dalrymple et al., 1995) for its representations and as its theory of GEN (see (1) below).