The acquisition of the English past tense inflection is the paradigm example of rule learning in the child language literature and has become something of a test case for theories of language development. This is unfortunate, as the idiosyncratic properties of the English system of marking tense make it a rather unrepresentative example of morphological development. In this paper, I contrast this familiar inflection with a much more complex morphological subsystem, the Polish genitive. The genitive case has three different markers, each restricted to a different subset of nouns, in both the singular and the plural. Analysis of the spontanous speech of three children between the ages of ; and ; showed that they generalized, and overgeneralized, all three singular endings. However, error rates were extremely low and there is no evidence that they treated any one ending as the ' default '. The genitive plural, on the other hand, showed a strikingly different pattern of acquisition, similar to that seen in English-speaking children learning the past tense. It is argued that in the latter two cases, the default-like character of one of the affixes is attributable to the properties of the relevant inflectional subsystems, not to the predispositions that children bring to the language-learning task. : For most linguists, it is axiomatic that lexical and grammatical knowledge are distinct aspects of a speaker 's competence. The lexicon is finite, whereas the number of complex units (inflected forms, phrases, and sentences) that a speaker can produce and understand is in principle infinite. Furthermore, the relationship between a word and its referent is arbitrary and hence un-[*] I would like to thank Magdalena