This dissertation applies a particular theory of language
 acquisition and representation, Optimality Theory (Prince
 and Smolensky 1993, Grimshaw 1997), and a particular
 learning algorithm within this theory, the Constraint
 Demotion Algorithm (Tesar and Smolensky 2000), to the
 problem of how second language acquisition of pro-drop 
 takes place for learners whose first language does not 
 instantiate the grammatical properties traditionally 
 associated with pro-drop. 



The overarching goal of this study is to provide an 
 account of the developmental stages in the second language
 learning of three grammatical properties: null subjects, 
 inversion, and that-trace.  Although there is no lack of 
 such accounts from earlier generative perspectives, the 
 need remains for a comprehensive developmental account 
 from an Optimality-theoretic perspective. This 

dissertation begins to address that need.



The study is based on several empirical tests (a 
 translation task, a pilot study, and a grammaticality 
 judgment task) that were administered to 370 adult native 
 English speakers studying Spanish at the University of 
 South Carolina or the Pennsylvania State University.  Each 
 task was designed to investigate learner competencies 
 regarding null subjects, inversion, and that-trace.  A key 
 conclusion from these studies is that the acquisition of 
 Spanish by native speakers of English involves a reranking 
 of universal syntactic and discoursal constraints in these 
 languages.  Specifically, this dissertation argues that 
 acquisition of Spanish occurs through the demotion of 
 certain syntactic constraints in the English native 
 grammar so that these constraints are dominated by 
 discoursal constraints in the Spanish second language 
 grammar.