Elmo's Sister's Ball: The Problem of Acquiring Nominal Recursion

Abstract

Recursion is the ability to iterate syntactic constituents inside constituents. Experimental and corpus literature indicates that comprehending recursive structures is difficult for children. Since recursion is a fundamental property of language, the result of the application of core minimal tools (merge, label, and select), why does it present an acquisition challenge? We propose that while Merge itself is universal, its application is determined by selection, which is extensively parameterized. We investigate whether children's difficulties can be defined in terms of the complexity of the selectional demands involved. This investigation leads to three empirical questions: i. How does NP recursion compare to NP coordination in acquisition? ii. Do children acquire each level of recursion independently or is recursion triggered automatically once the initial selectional requirements are acquired? iii. Are there differences in the acquisition of different types of nominal recursion? We compare the selectional requirements of English genitives, PPs, and NP coordination. Our syntactic analysis suggests that genitive recursion has more complex selectional demands than PP recursion. We designed a study eliciting production of two level embedded prepositional and genitive NPs. Preschool-aged English-speaking children (n = 46) and adults (n = 11) participated in an elicitation study. Children were comparable to adults in their ability to produce three coordinated NPs, but had substantial difficulty linking three NPs using recursion. First- and second-level embedding were clearly distinct steps in development. These findings have implications for the independence of phrasal and selectional development in children and for our understanding of structural complexity in child language and in theory.