Level-ordering in lexical development.

This paper examines the claim that lexical word formation rules are ordered at three levels of application (Kiparsky, 1982). Lexical rules of affixation and compounding differ with respect to phonological effects, semantic regularity and productivity. With respect to these properties, rules can be assigned to one of three “levels” and are applied sequentially. Level-ordering predicts that irregular plurals may be formed at level 1 prior to compounding at level 2. Thus, forms such as mice-infested are acceptable. However, regular plurals formed at level 3 may not precede compounding, therefore predicting that *rats-infested is not acceptable. A learnability problem arises since the child almost never hears compounds containing irregular plurals. Given that the input appears to underdetermine the relevant constraints, it is suggested that level-ordering is an innate structural property of the lexicon. It is predicted that children should show no evidence of having to learn the constraints of levelordering with respect to pluralization and compounding. An experiment with 33 threeto jive-year-olds elicited singular, plural and compound forms of regular, irregular and pluralia tantum nouns (also at level 1). Results showed that: (1) Children almost never produced regular plurals inside compounds (e.g., *rats-eater); (2) As soon as children used irregular plurals, they used them inside compounds (e.g., mice-eater); (3) Plurulia tantum nouns were also used inside compounds (e.g., clothes-eater) although for variousphonological and semantic reasons there appeared to be a difference for the individual nouns within this class. The results strongly support the notion that level-ordering constrains the child’s word-formation rules, independent of the input received. Some possible mechanisms for assigning rules to their appropriate levels are discussed. *The research reported here was carried out while the author was a Sloan Post-Doctoral Fellow at Stanford University. Many thanks to Susan Lierle for running subjects. I am also grateful to Melissa Bowerman, Susan Carey, Kathie Carpenter, Eve Clark, Susan Gelman, Steven Pinker and Meg Withgot for helpful comments. A version of this paper was originally presented at the Ninth Annual Boston University Conference on Child Language Development, October 1984. Requests for reprints should be sent to: Peter Gordon, Department of Psychology, University of Pittsburgh, Pittsburgh, PA 15260, U.S.A. OOlO-0277/85/$6.80 OElsevier Sequoia/Printed in The Netherlands