The frequency of individual segments and of each particular syllable has been counted in lists of about 2000 to 5000 lexical entries in a sample of languages drawn from the major areal/genetic groupings. An analysis of a subset of these languages with small segment inventories and simple phonotactics (Maddieson and Precoda, 1992) has shown that the frequency of a particular syllable type can be better predicted from the overall segment frequencies than from the assumption that adjoining articulatory similar segments are preferred, as suggested by Janson (1986). Additional languages are being added to the database to test the generality of this conclusion over a more diverse set of languages. The extended set includes Chinese, Polish, Korean, Comanche, Kwakw’ala, Yupik, !Xóõ, Ngizim, Maninka, Gbaya, Fasu, Turkish, as well as Hawaiian, Pirahã, Rotokas, Shipibo, and Kadazan which were reported on earlier.