The days in New England finally feel like they’re getting longer now, and with the increase in daylight, I’ve gone back to staring out of the window on my way home from school, enjoying the scenery, and, of course, thinking about my next linguistics endeavor. Today I wanted to discuss Optimality Theory, a general model of how grammars are structured that has various applications in phonology research.
OT was invented in 1991 by Alan Prince and Paul Smolensky, two well-known linguists. Its invention was revolutionary, stimulating research in syntax, semantics, sociolinguistics, historical linguistics, and most of all phonology.
Up until the creation of OT, the mainstream approach to modeling grammars was a blocking and triggering method created by Noam Chomsky and Howard Lasnik in 1977. Their method was based on the idea that phonological and syntactic processes are influenced by constraints on the output of the grammar; processes can be blocked or triggered by output constraints. In their approach, all of the transformations that could be applied to an input are optional. Any input can go through any, all, or none of the transformations. The result is submitted to surface-structure constraints, called filters. Any candidate surface structure that satisfies all of the filters is a well-formed sentence of the language. A transformation is blocked whenever a surface filter rules out the result of applying that transformation. Similarly, a transformation is triggered whenever a filter rules out the result of not applying to the transformation: the transformation must be applied according to the filter.
Chomsky and Lasnik’s approach was extremely influential in syntax, but not phonology, as it could not work on typical phonological data because its output constraints were inviolable. In other words, the theory, and those that followed it, were too specific, and there were far too many exceptions to the point where the theories simply did not apply. Parameterizing phonological typology would require a practically impossible number of constraints.
OT solves this problem by setting up a basic division between the operational component (GEN) of the grammar and the constraint component (EVAL). GEN constructs a set of candidate output forms that deviate from the input in various ways. It applies all linguistic operations freely, a property known as freedom of analysis. This freedom allows GEN to create every single possible candidate. EVAL selects a member of the set GEN creates and makes it the actual output of the grammar; it evaluates the members using some constraint hierarchy and selects its most harmonic or optimal member as the output of the grammar. It is important to note that OT constraints are violable, unlike Chomsky and Lasnik’s inviolable filters or the various pre-OT phonological constraints. The best output in OT is the output that violates the smallest number of important constraints; violating a grammar’s principal constraints gives the input a larger penalty.
The set of all of the constraints in OT is called CON. All languages use all of the constraints in CON, but the grammars are differentiated by the weighting of each constraint. There are two types of constraints. The first is markedness, a substantive theory of linguistic well-formedness that states that complex consonant clusters or that-trace sequences are bad. In simpler terms, markedness removes any inputs that violate universal linguistic rules. The second type of constraint is faithfulness, a conservative type of constraint that only requires that any output of the grammar resembles its corresponding input.
OT has been applied to many research problems, and it continues to be used to solve linguistics challenges today.
McCarthy, John J. What Is Optimality Theory? University of Massachusetts Amherst, Jan. 2007.