A Method of Language Sampling

In recent years more attention is being paid to the quality of language samples in typological work. Without an adequate sampling strategy, samples may suffer from various kinds of bias. In this article we propose a sampling method in which the genetic criterion is taken as the most important: samples created with this method will reflect optimally the diversity of the languages of the world. On the basis of the internal structure of each genetic language tree a measure is computed that reflects the linguistic diversity in the language families represented by these trees. This measure is used to determine how many languages from each phylum should be selected, given any required sample size.