We analyse the co-authorship networks of researchers affiliated at universities in Turkey by using two databases: the international SSCI database and the Turkish ULAKBIM database. We find that co-authorship networks are composed largely of isolated groups and there is little intersection between the two databases, permitting little knowledge diffusion. There seems to be two disparate populations of researchers. While some scholars publish mostly in the international journals, others target the national audience, and there is very little intersection between the two populations. The same observation is valid for universities, among which there is very little collaboration. Our results point out that while Turkish social sciences and humanities publications have been growing impressively in the last decade, domestic networks to ensure the dissemination of knowledge and of research output are very weak and should be supported by domestic policies.