Crossover asymmetries

Weak crossover and strong crossover are standardly attributed to at least partially different constraints, despite clear parallels in their distribution in English. Specifically, the standard analysis of strong crossover attributes it to Condition C, which plays no role in the analysis of weak crossover. This line of analysis predicts that the two types of crossover could in principle part ways, resulting in configurations that exhibit strong crossover but not weak crossover. In this article, we argue that scrambling in Hindi-Urdu bears out this prediction. We show that local scrambling displays secondary-strong-crossover effects but not secondary-weak-crossover effects, and we furthermore show that the distribution of strong crossover correlates with the distribution of Condition C connectivity. We furthermore argue that the distribution of strong crossover (and of Condition C connectivity) is crucially conditioned by case. Focusing primarily on the distribution of strong crossover and Condition C connectivity, we propose an analysis that extends to scrambling a DP-late-merge account of Condition C obviation, and we discuss the implications of this analysis.