Abstract

Traditional grammars tend to treat inflectional systems as structured networks of elements, not as unstructured sets of forms with shared stems or roots. Although the exemplary patterns and leading forms of these descriptions bring out the structure of inflectional systems, traditional accounts are incomplete in some important respects. In particular, there is no method for measuring the implicational structure of a set of forms, no means of gauging the diagnostic value of specific forms within a set, and no generally accepted way of identifying the leading forms of a system. The problem of measuring this kind of inflectional structure can, however, be recast in a form that is susceptible to well-established techniques of analysis. The approach outlined in this talk proceeds from the observation that the relevant structure involves information that one form conveys about other forms. This information correlates with various of the entropy measures proposed initially in Shannon (1948). For example, the diagnostic value of a paradigm cell C correlates with the difference between the baseline entropy of a system and the conditional entropy of the system given knowledge of C. Entropy-based measures also diagnose the anomaly of a fully suppletive class system or the pathologically ‘uneconomical’ classes of Carstairs (1983), since in neither type of system does knowledge about any form reduce the uncertainty of other forms.