Object relatives made easy: A cross-linguistic comparison of the constraints influencing young children's processing of relative clauses

Abstract

We present the results from four studies, two corpora and two experimental, which suggest that English- and German-speaking children (3;1–4;9 years) use multiple constraints to process and produce object relative clauses. Our two corpora studies show that children produce object relatives that reflect the distributional and discourse regularities of the input. Specifically, the results show that when children produce object relatives they most often do so with (a) an inanimate head noun, and (b) a pronominal relative clause subject. Our experimental findings show that children use these constraints to process and produce this construction type. Moreover, when children were required to repeat the object relatives they most often use in naturalistic speech, the subject-object asymmetry in processing of relative clauses disappeared. We also report cross-linguistic differences in children's rate of acquisition which reflect properties of the input language. Overall, our results suggest that children are sensitive to the same constraints on relative clause processing as adults.

Acknowledgements

This research was in part supported by a postdoctoral fellowship to the first author from the Max Planck Institute for Evolutionary Anthropology. We would like to thank Anna Roby and Susanne Grassman for help in testing and coding, and Franklin Chang, Paul Bloom and two anonymous reviewers for helpful comments. Portions of this research were presented at the 19^th Annual CUNY Conference on Human Sentence Processing.

Notes

¹Gordon et al. (2001) argued that the difficulty ascribed to object relatives can be attributed to the fact that the NPs participating in the RC have mainly been of the same semantic type – animate human NPs. They predict that this causes an interference effect that affects thematic role assignment, and that an object NP that contains two semantically different NPs in any position should alleviate object NP difficulty. Mak et al. (2006) and Traxler et al. (2005) have both shown that the similarity-based approach does not predict patterns of processing difficulty in adults, since, crucially, object relatives that have an animate head and an inanimate RC subject (e.g., the dog that the ball hit) are not easier than object RCs that have semantically identical NPs.

²This is actually only true for masculine NPs, as there is greater ambiguity in case marking for feminine and neuter NPs.