Hello Hans, Considering the Turkish hyphenation patterns (which don't seem to be EC-encoded): I tried to decipher them, although some native speaker should probably take a look. I'm sending the original documentation and a kind of list of replacements. I would suggest to replace the following letters (see the attached file): i,i:,o:,u:,a=,i=,o=,c:,g=,s: and to ignore the first three patterns and all those containing any of the following sequences (they seem to be there just for the sake of Arabic transliteration): @,#,g:,z:,d!,h!,k!,s!,t!,z!,d=,n=,t=,z=,h=,s= (Or to regenerate the patterns with from scratch with a ruby script, which might be faster :) ftp://tug.ctan.org/pub/tex-archive/language/turkish/hyphen/turk_hyf.c Mojca