On the distribution and complexity of gender and numeral classifiers

This paper surveys the occurrence of gender and numeral classifiers in the languages of the world and evaluates statistically whether there is a complexity trade-off between these two linguistic patterns. Complexity is measured as overt coding of the pattern in a language, an approach that has been shown earlier to provide a reliable first estimate for possible trade-offs between typological variables. The data come from a genealogically and areally stratified sample of 360 languages. The relationship between gender and numeral classifiers in this data was researched by constructing Generalized Linear Mixed Models. According to the results a significant inverse relationship occurs between the variables independently of genealogical affiliation and geographical areas. The distributions are explained functionally by economy, that is, the tendency to avoid using multiple patterns in the same functional domain.