ABSTRACT
I argued in Chapter 2 that a certain hypothetical Latin-like language L diverged inflexionally from actual Latin in such a way as to render it impossible as a natural language. The difference between Latin and L is that in L each noun's inflexion for each of the twelve Case-Number combinations in the paradigm is determined independently. L therefore has what we called the Independent Distribution Characteristic; and I suggested that no natural language possesses this characteristic. But, if inflexions are not in fact distributed independently, our task now is to determine the nature of the restrictions which govern the organisation of inflexional resources into paradigms. In the course of developing more concrete proposals on this, I will present further material from three languages unrelated to each other or to Latin (or, at most, only extremely distantly related): Hungarian, Zulu and Dyirbal. The presentation will at first be informal, and from now on I will deliberately avoid using the term 'paradigm' until I come to define it for the purpose of formulating a claim about 'paradigm economy'.
