Van Wijk's study of Dutch plurals provides a valuable contribution to the discussion of default rules in language. The outcome of this dissertation asks for a serious reconsideration of the claim, as made in Pinker's (Reference Pinker1999) words and rules theory, that default rules are unique and, as such, insensitive to phonological and semantic regularities of the stem. Dutch allows, for instance, the pluralization of novel words, names, acronyms, and loanwords with two different suffixes, –en and –s, which is not expected if a default rule is unique. Data collected in a series of experiments indicate, moreover, that children learning Dutch as their first language as well as native adults attach both suffixes to nonwords.
On the basis of the observed variability, van Wijk argues for a words and constraints approach, according to which regular plurals in Dutch are formed on the basis of an ordered set of violable, conflicting constraints. These constraints can be rather diverse in nature. Some are universal; others are language-specific. Some constraints are sensitive to phonology, whereas others are not. The ranking of the constraints, and hence the choice of the suffix, is a matter of development. Additionally, suffix selection is determined by input distributions. Van Wijk maintains the distinction between grammar and lexicon and assumes that irregular plurals are stored in the mental lexicon.
The strength of the words and constraints approach lies in the analysis of interindividual and intra-individual differences. Because of this, the framework could provide a fruitful starting point for explaining patterns in the data of second language learners, in particular because these data generally show a higher degree of variability than those of native learners. It is conceivable that learner varieties are caused by the transfer of language-specific constraints. Commonalities between native learners and (different groups of) nonnative learners could involve those aspects of the Dutch plural related to input frequencies.