The Acquisition of the Weak Crossover Constraint : Evidence from Dutch Preschool Children

Abstract

The Weak Crossover Constraint restricts variable binding in sentences such as (1), in which an operator appears to move past a coindexed embedded pronoun (either overtly, or at LF). The pronouns in (1) can thus only be interpreted as deictic pronouns, referring to some previously specified individual in discourse.

(1) a. Hisi mother likes each boyi. b. Whoi does hisi mother like ti?

In the theoretical literature there has been much debate about the nature of the WCO-constraint. Some researchers have reduced it to restrictions on binding relations (e.g. Koopman and Sportiche 1982, Safir 1984, 2004, and Reinhart 1983). Others have reduced it to scope (e.g. Pica and Snyder 1995, Ruys 2000), prominence (Bresnan 1995, 1998), or have attempted to derive the constraint compositionally in different semantic frameworks (Büring 2004, Schlenker 2005). However, each approach has its own advantages and drawbacks. Acquisition research can help decide between competing accounts by testing specific predictions about L1-acquisition derived from these theories. However, previous L1-acquisition research has provided a conflicting picture (Thornton 1990, Lewis 2000), which may in part be due to methodological problems. This thesis reports two experiments which investigate the acquisition of the WCO constraint by Dutch preschoolers. The first experiment tests whether a methodological modification to improve the pragmatic felicity of the pronoun in the test statement improves children’s performance on WCO sentences (as compared to Lewis 2000). Furthermore, it tests a prediction of one of the scope based approaches to WCO (Ruys 2000), namely that children should reject the variable-bound interpretations of WCO sentences, while they should accept variable binding in sentences where a quantifier is seen to exceptionally scope over the sentence The results indicate that 61.9 percent of the children appear to have acquired the WCO constraint. This constitutes a significant improvement over Lewis (2000). In other words, improving the felicity of the test statement indeed improves children’s performance. Furthermore, the results provided evidence in favour of scope-based approaches to WCO. A second experiment tests the effects of two possibly confounding factors, namely the quantifier children are presented with, and the relative order between the pronoun and the quantifier. In this experiment the results of the first experiment were replicated under a different design. Furthermore, there is no effect of quantifier, but children are affected by the precedence relations obtaining between the pronoun and the quantifier. However, they do not differ from adults in this respect.