Measuring affix rivalry as a gradient relationship

Abstract

In many languages, affixes can be used to derive words with the same semantic types and can therefore compete in word formation. This paper discusses how to quantitatively assess the competition between derivational affixes based on their semantic similarity. Two possible measures of affix rivalry drawn from studies in ecology are examined: the Sørensen index, which considers the proportion of shared functions between rival affixes; and the Percentage similarity coefficient, which is based on the realization frequency of functions. Two complementary measures (Balanced richness and Balanced abundance) are also proposed to further analyze the semantic dissimilarity between rival affixes. Using the semantic competition between six French deverbal suffixes as a case study, we show how these four measures suit the quantification of affix rivalry and help capture different aspects of the phenomenon.