ABSTRACT
This chapter discusses the cooccurrence restrictions follow from several conditions in the grammar of the language under consideration. The reader will note that the languages with the strictest set of restrictions on the cooccurrence of similar segments—Souletin Basque, Sanskrit, and Cuzco Quechua—ali lack the identity effect. The cooccurrence of identical segments is prohibited. The chapter shows that it is similarity among the set of segments over which laryngeal cooccurrence restrictions operate that is at issue. The existence of the identity effect is determined by the relative rankings of BeIdentical, the constraints that penalize similarity, and constraints that enforce faithfulness to the input representations. The chapter provides an Optimality Theory analysis of the data. Tableau illustrates how the cooccurrence of non-identical ejective stops is prevented in Cuzco Quechua. This tableau provides first illustration of how Optimality Theory allows cooccurrence restrictions to be encoded in the grammar, rather than in a set of conditions levied on underlying forms.
